On April 30, 2026, the California Supreme Court decided In re Kowalczyk. I already wrote about this in Kowalczyk and the End of Pretend Bail Seven justices, zero dissents, one foreclosed workaround. In my previous article, I mentioned the “next front” or battlefield that I expect to come from this.

The more I’ve thought about what happened, the more irritated it makes me. Because there was a concurrence that — for those who want the tl;dr version — is 100% dicta and 100% political advertisement.

In In re Kowalczyk, Chief Justice Guerrero, writing for a unanimous court, resolved a constitutional ambiguity that In re Humphrey had left on the table in 2021: the relationship between Article I, section 12 guaranteeing release on bail subject to specifically enumerated exceptions and Article I, section 28(f)(3), the “Public Safety Bail” provision added by Proposition 9.

The Court held that section 12 controls and that section 28(f)(3) plays within its monkeybars rather than building a new playground. Applied to Mr. Kowalczyk, a homeless man held in pretrial custody on a $75,000 bail he could not pay after attempting to use found credit cards to buy a $7 cheeseburger, the holding meant the trial court had erred. Liberty is the norm. Detention is the carefully limited exception. That is a holding. It is binding.

Justice Wiley joined that holding without reservation. He then wrote separately to do something else.

What a Concurrence is Supposed to Do

A concurrence has a small repertoire. It can supply an alternative legal ground for the result. It can sharpen or qualify the majority’s reasoning. It can flag a doctrinal question reserved for a future case. It can register a narrow disagreement that doesn’t disturb the disposition. These are the moves. They share a common feature: they are still doing legal work, still tied to the case in front of the court.

What a concurrence does not do — what it has no business doing — is launch into freestanding policy commentary on a contested social question while joining the majority’s reasoning in full. The grammar of the form rules it out. If a justice has joined the majority opinion, by definition nothing in the separate writing is necessary to the decision. It supplies no rule. It binds no court. It is dicta, and, here, not the productive kind that clarifies law for the next litigant. It is dicta unmoored from the act of judging.

The Unnecessary Invitation

Justice Wiley’s concurrence, brief as it is, reads as a sustained policy meditation. He asks what an ideal bail system would look like. He wonders whether there are models California should emulate from other jurisdictions or from federal practice. He concedes — explicitly — that the judiciary, unlike the political branches, cannot freely investigate, consult, compromise, or face the voters on policy platforms. He says the political branches have “tremendous advantages over the judiciary” in designing pretrial release regimes. And he closes with the line that gives this piece its title: “I hope the invitation is accepted.”

That is not a holding. It is not even reasoning. It is a wish — addressed to the Legislature and the Governor — published in a volume of the California Reports.

Read the text carefully and notice what it does not contain. Justice Wiley does not articulate a different legal framework for the section 12/section 28(f)(3) question. He does not narrow or qualify the majority’s reasoning. He does not stake out a separate ground for the result. He does not even register a concern about the holding he joined. He has used the institutional vehicle of a California Supreme Court concurrence to encourage two coequal branches to take up a policy direction he seemingly prefers — that’s the best reason I can figure for why he bothered to write this.

Whatever the merits of pretrial-release reform — and to quote someone else who is always dicking around where he needn’t, “many people” believe there are serious ones — that is not the assignment.

Why This is Unbecoming

Judges hold their authority on a narrow franchise. They decide cases. The reason judicial opinions carry weight is the same reason they’re constrained: judges speak when a live controversy comes before them, and only so far as their words are necessary to resolve it. Strip that constraint and a judicial opinion becomes a newspaper column or a Substack post with a gavel attached.

There is a sharper version of the point in this case. Justice Wiley sits on the Court of Appeal, Second District. He occupies a seat on the Supreme Court only by temporary assignment, filling the vacancy left by Justice Jenkins’s retirement. He is a guest at the high court’s bench. The borrowed pulpit makes the political tone even worse. A justice on a temporary commission has, if anything, more reason to keep the work close to the case and the case alone.

The dicta-as-political-advertisement form has a particular pathology. It allows the writer to disclaim any operational effect — it’s just a concurrence, it binds nothing — while collecting the rhetorical premium of the institutional letterhead. It is arbitrage. Lobbyists use op-eds; this judge used a concurrence.

The political-advertisement reading I’m arguing here isn’t me promoting some controversial view. CalMatters, covering the decision for a general audience, described the concurrence in plain English as having “essentially begged for the matter to be resolved by the governor or the Legislature.” That is a journalist, not a defense lawyer, calling a concurrence what it is: supplication.

A judge who wants to address the political branches on policy has every legitimate venue available to him. A law review article. A speech. A commission. What he may not do — what he should not do — is fold the policy preference into the California Reports and pretend it is judicial work.

Legal Force vs. the Operational Danger

The dicta has no legal force. But legal force and operational force are not the same thing. Lawmakers and prosecutors who have spent the better part of a decade agitating for an aggressive reading of section 28(f)(3) — for a return to the unaffordable-bail status quo under a public-safety theory that serves only to add further pressure for detained people to plead out so they can go home — now have a sitting California Supreme Court “justice” telling them, on the record, “The door is open”. They will cite Justice Wiley not as authority (he supplied none) but as cover. Even the justice who joined the holding hopes you’ll act.

That cover is the damage. Kowalczyk will be quoted in legislative testimony for years. So will “Wiley’s Wish”. The first will do legal work. The second will do political work. Both will travel under the same imprimatur.

Coda

The Chief Justice did the work. She resolved a constitutional ambiguity that had been pending since Humphrey sidestepped it five years ago. The holding is careful, the reasoning patient, and the result faithful to a state constitutional right that has been part of California law since 1849. Whatever one thinks of pretrial-detention policy, that is what a judicial opinion looks like.

The concurrence is something else. It is a flat-out political advertisement. There is no law, nor subtlety, at all.

The California Reports are not the right venue for such chicanery.

I have been a criminal defense attorney in Fresno County for nearly twenty years. As a student, I opened my office with “Rick Horowitz, Independent Law Clerk” on the door and began researching and writing for criminal defense lawyers. Before I became an attorney, I sat second chair in multiple special circumstance murder trials and argued before the Fifth Appellate District Court. And I have been litigating the Humphrey line of cases and related issues since before the original Humphrey decision came down.

On January 6, 2026, I wrote a Daily Journal article with my co-author, colleague, and friend, Eric Schweitzer, of Schweitzer & Davidian, in which we stated that California’s bail problem was not legal uncertainty. The relevant law had been clear since 1849. That’s one year before California officially became a state, part of the United States.

In that article, we said that the California Supreme Court’s bail jurisprudence over the previous decade gave little reason to believe it would endorse a regime permitting detention by pricing freedom out of existence.

We specifically noted:

The Court of Appeal in Kowalczyk attempted to reconcile Article I, sections 12 and 28 in a way that risks normalizing unaffordable bail. The Supreme Court’s grant of review suggests discomfort with that result. The court has not yet issued a final opinion, and it may yet refine its reasoning. But in light of the court’s bail jurisprudence over the last decade, there is little reason to believe it will endorse a regime that permits detention by pricing freedom out of existence.

— Eric H. Schweitzer & Rick Horowitz, Bail Realities and Policy Fictions: The Truth About Affordable Bail, Daily Journal (Jan. 6, 2026)

On February 18, 2026, I created a video and asked this question, “was what was happening in California courtrooms confusion, or resistance?”

On February 21, 2026, I posted another video walking through the original Daily Journal argument in long form.

Then. On April 30. The California Supreme Court answered the question. Unanimously.

And it opened its opinion with the exact same 1849 framing that I had used in the Daily Journal piece.

The entire thesis of my Daily Journal argument was that trial court judges were the busiest of all bees — and the sting was all the more painful to those of us who had dug into the case law — in finding “workarounds” to the entire line of cases descending from Humphrey.

But, as I had predicted, Kowalczyk has foreclosed the workarounds. Unaffordable bail used as detention by any other name is no longer a gray area. And the question I posed in February has been answered.

I’ve been making this argument not just in print, but in trial courts. In a motion filed in Fresno County in February — citing Kowalczyk’s appellate decision as wrongly decided and Brown as the better authority — I argued that unaffordable bail set “knowing full well that it was the equivalent of a pretrial detention order” was constitutionally impermissible. The Supreme Court has now adopted that position.

What Kowalczyk Actually Held

The Supreme Court, in taking up Kowalczyk, limited the parties to two issues and that is the focus of this section.

Can Section 12 and Section 28 Be Reconciled?

First, the Court addressed the question of whether Article I, section 12, and Article I, section 28(f)(3) of the California Constitution were in conflict, or could be reconciled. The petitioner argued they were reconcilable — and the Court of Appeal had similarly found this to be the case. The respondent — and it’s interesting to note here that the Attorney General apparently played no part in this case, not even as amicus — argued that the two sections were in conflict and that section 28(f)(3) should override, or repeal, section 12.

Had the court accepted the respondent’s position, bail would be essentially discretionary for all crimes. Judges would have carte blanche to deny bail whenever they thought not doing so could have blowback consequences on them (e.g., for elections). And the default would have flipped from “liberty is the norm” to “pretrial detention is the norm”, which, incidentally, it has been for some time in certain counties.

That’s pretty much what the whole fight over bail has been about. And we will see below that this fight is probably far from over.

Fortunately, for now, the Supreme Court sided with the petitioner on this one. In re Kowalczyk, No. S277910, 2026 Cal. LEXIS 2206, at *20 (Cal. Apr. 30, 2026) (“We agree with petitioner.”)

Suffice it to say that the Court found sections 12 and 28(f)(3) could be harmonized. They rejected the idea that there was an implied repeal of section 12 when 28(f)(3) was passed by voters.

They went on to state that section 12 reiterated what had been true since 1849: “All persons shall be bailable”. Kowalczyk, 2026 Cal. LEXIS 2206, at *12, quoting Article I, section 7, of the California Constitution of 1849. In other words, “liberty is the norm”. Id. at *5, *48 (quoting In re Humphrey, 11 Cal. 5th 135, 155 (2021)). Section 12 then — and only then — mentions “very limited exceptions” in section 12 subsections to the rule that “pretrial release was guaranteed for all noncapital defendants”. Id. at *12 (citing In re Underwood, 9 Cal. 3d 345, 350 (1973)). The opinion also notes this on page 20:

Section 12 first prescribes a broad right to release on bail and then identifies, in list format, specific exceptions from that right.

— Kowalczyk, 2026 Cal. LEXIS 2206, at *25

The respondent had suggested that the use of “may” in section 28(f)(3) abrogated the restrictions placed upon judges by section 12.

The Court disagreed for numerous reasons. But salient here was a reminder that since 1849, the idea was always to limit the ability of a judge to deny bail, because the default is pretrial release.

[T]he right to release on bail, as currently codified in section 12, has been a constitutional right since our state’s founding. […] Such a right was included in our state’s earliest Constitution “to abrogate the common law rule that bail was a matter of judicial discretion by conferring an absolute right to bail except in a narrow class of cases.”

— Kowalczyk, 2026 Cal. LEXIS 2206, at *23–24 (quoting In re Law, 10 Cal. 3d 21, 25 (1973))

The Court noted that the type of discretion the respondent (prosecutors) wanted “has never been part of our state’s law.” Id. at *24. (By the way, it’s interesting to note that the Court mentions “1849” seven times in the majority opinion. Similarly, the Court also repeatedly notes the primacy of pretrial liberty — with limited exceptions spelled out in section 12 — as opposed to pretrial detention.)

In the end, the “may” of section 28(f)(3) reads as “it could happen that” someone is released on pretrial bail as required by section 12; i.e., pursuant to the conditions of section 12, someone may have the right to bail. In setting bail, section 28 adds some considerations (See Kowalczyk, 2026 Cal. LEXIS 2206, at *18 n.9) but does not broaden the scope of those who may be denied bail beyond those cases delineated in section 12.

Bail Must Not Be “Unreasonable”

You’ll forgive me, I hope, for pointing this out: in the Daily Journal article we suggested that the Supreme Court likely took up the Kowalczyk case because it conflicted with what we considered to be the better analysis of In re Brown, 76 Cal.App.5th 296 (2022).

Well, okay, it wasn’t because of what we thought. Perhaps it would be more clear to say that we thought they took it up for a particularly reason. And, then, as the next paragraph points out, they said that’s why they took it up. Although I see hints of our article in the opinion including certain word choices, I have absolutely no idea if they read it and they most certainly did not cite to it!

It could just be that “great minds think alike.”

The Supreme Court noted the conflicting appellate opinions as a reason for having taken up Kowalczyk and, as we will see — though it did not expressly say this, essentially sided with Brown, just as we had suggested. Kowalczyk, 2026 Cal. LEXIS 2206, at *11, *18–19.

In fact, in the most recent “Humphrey Motion” I filed, I had a section titled “Brown and Kowalczyk Seemingly Conflict; Between Brown and Kowalczyk, the Court Should Follow Brown”. That case is currently before the Fifth District Court of Appeal on my Petition for a Writ of Mandate/Prohibition after bail was (I believe) improperly denied. In a footnote of my Motion, I noted:

While Kowalczyk has not yet been decided by the Supreme Court, it is worth noting Nunez-Dosangos v. Superior Court, 107 Cal. App. 5th 283 (2024), cites Kowalczyk, addresses due process, and took a unfavorable view to pretrial detention.

— from my most recent “Humphrey Motion” in a trial court

I argued that the Appellate Court’s version of Kowalczyk was internally incoherent because setting bail “if there is a valid basis for detention” was either superfluous because detention is already authorized or contradictory because no number could secure release. The Supreme Court reached the same destination by a different route through its footnote 21. Kowalczyk, 2026 Cal. LEXIS 2206, at *37-40 & n.21. If I may embrace a tiny bit of hubris here, I think my reasoning was sharper; the Court’s resolution more cautious.

Anywayser, at page 32 of Kowalczyk, the Court essentially summarizes all that it is going to argue relating to the ongoing controversy initially addressed by Humphrey regarding affordable bail.

Based on our constitutional framework, which prioritizes liberty by mandating that nearly all noncapital criminal defendants have a general right to release on bail, combined with the equal protection and due process principles announced in Humphrey, we conclude that bail determinations (including the amount of bail that is set) must be reasonable under the totality of the circumstances. As a general rule, the right to reasonable bail means that a court may not set bail in an amount that is objectively unattainable for the defendant.

— Kowalczyk, 2026 Cal. LEXIS 2206, at *37 (bold emphasis added; plain italics in original).

Things do get tightened up a little bit.

When it comes to considering monetary bail, the defendant must do more than present conclusory assertions of indigency or an inability to pay. […] But courts cannot use artificially high or objectively unattainable bail as an end run to effectuate pretrial detention where such detention is not authorized by section 12.

— Id. at *37–38

This is a point where I think we might see a new battleground evolving. Because, as the Court noted earlier in its opinion:

This does not mean that bail may only be set in an amount that is easily affordable or convenient to the defendant, or that the court must accept unsupported, conclusory assertions of indigency or an inability to pay. Rather, consistent with our constitutional framework regarding bail determinations and our prior precedent, bail must generally be set in an amount that is reasonably attainable, in order to effectuate the defendant’s constitutional right to release on bail.

— Id. at *7–8

Ultimately, I think all but the most recalcitrant of courts are going to have to recognize that the Supreme Court has essentially reiterated that it is not kidding when it comes to the essential points:

1. Pretrial release, rather than pretrial detention, is the ruling paradigm; “liberty is the norm”.

2. Only those “exceptions” to bail supported by section 12 are allowed.

3. Bail must be “reasonably attainable” (whatever that ends up meaning will probably be subject to some litigation).

Kowalczyk, 2026 Cal. LEXIS 2206, at *48–49.

Kowalczyk is just the latest from the Supreme Court reiterating these points. And it’s a sad fact that we have to keep returning to this. In my mind, the “constitutional framework” is not unclear and never has been — not since 1849 — it’s just, as I said in the Daily Journal being ignored by judges for reasons we’ll get to below.

Seven Justices, Zero Dissents

As I said, Kowalczyk is just the latest from Courts of Appeal and, more importantly, the Supreme Court. This is another point that I hammered in my Daily Journal article.

This is why habeas litigation remains necessary and why defense counsel must continue to utilize it whenever trial courts balk at the default. As a reminder: “liberty is the norm, and detention prior to trial…is the carefully limited exception” Yedinak v. Superior Court, 92 Cal. App. 5th 876, 881 (2023). These are not ceremonial words to be intoned by priests in robes waving incense to smoke up the room just before an order of pretrial detention. […] The fact that defense lawyers must repeatedly resort to habeas petitions to secure compliance with settled constitutional law is not evidence that the law is unclear. It is evidence that the law is being ignored.

— Eric H. Schweitzer and Rick Horowitz, Bail realities and policy fictions: The truth about affordable bail (January 6, 2026)

The Supreme Court Justices themselves have never wavered on this point. And in Kowalczyk, Justice Guerrero wrote for a unanimous Court. Justice Groban, joined by Justices Liu and Evans, calls intentionally unattainable bail “the functional equivalent of a pretrial detention order” and names the practice a “fiction”. Kowalczyk, 2026 Cal. LEXIS 2206, at *52–53 (Groban, J., concurring).

Ultimately, the disagreement was never really among the justices. It was between the Constitution and a courthouse culture that had grown comfortable treating money bail as a shortcut around it. Kowalczyk makes explicit that this plays no part in the “constitutional framework”.

Hopefully, this is the end for pretend bail.

The Presumption of Innocence is Not Optional

The constitutional default is liberty. The constitutional presumption is innocence. Kowalczyk relies on both, repeatedly. Kowalczyk, 2026 Cal. LEXIS 2206, at *5, *37, *48–49. Humphrey relies on both.

Harris held:

To hold that a court must assume the truth of the criminal charges in making such a determination would improperly relieve the People of the burden that the constitutional text, so construed, assigns to them. Accordingly, we clarify here that a court does not assume the truth of the criminal charges when evaluating whether to order a defendant held without bail under article I, section 12(b).

— In re Harris, 16 Cal. 5th 292, 320-21 (2024)

Salerno itself, the federal case the District Attorneys leaned on in Kowalczyk, says:

In our society liberty is the norm, and detention prior to trial or without trial is the carefully limited exception.

— United States v. Salerno, 481 U.S. 739, 755 (1987)

But there is a competing attitude. Call it the “arrest-equals-guilt assumption” which has been growing in courthouses, prosecutors’ offices, in legislative debates, and in public opinion. The assumption is that the arrest itself is sufficient evidence of guilt for most practical purposes. But we all should know that it is not. Wrongful arrests, dismissed cases, acquittals (not guilty findings), and exonerations are not outliers. They are a measurable percentage of any criminal docket.

Yet this arrest-equals-guilt attitude colors every stage of a criminal case. Arrests based on jumping to conclusions. Charging decisions made on thin evidence to be cleaned up later. Bail hearings where “public safety” is asserted as if the charge itself proves the danger. Evidentiary hearings where the prosecution’s narrative is treated as established fact. Admissibility rulings that assume the government’s evidence is not only what it appears to be but — when it should not be admitted, but will help to prejudice the case — is necessary. Jury selection where the constant battle for the defense is responding to “we wouldn’t be here if the defendant was innocent”. And trials where the burden of proof is basically inverted — the defense must prove innocence — because everyone in the room (sometimes including the defense lawyer) has internalized the assumption that the defendant “probably” did it.

And “probably” is good enough.

Kowalczyk is another reminder that rights matter. The presumption of innocence matters. And we don’t lock up innocent people before they’ve been found guilty as the norm.

Pretrial liberty is the norm.

And it is both founded upon and furthers the idea that the presumption of innocence matters.

Evading the Norm Based on a Presumption of Guilt Must End

In a February 18 video, The Unrealized Promise of Humphrey: When Bail Reform Meets Judicial Resistance I walked through the UCLA/Berkeley Coming Up Short report which found that pretrial detention increased after Humphrey, that bail amounts did not drop, and that some judges read the ruling as license to deny bail entirely.

The pattern was visible in judicial behavior — a Sacramento County judge warning a defense lawyer that pushing a Humphrey argument might result in bail being taken off the table entirely; in San Joaquin County, the percentage of cases in which bail was set decreased from 30% in 2018 to 18.7% after Humphrey; and a Los Angeles trial court in In re Brown trying to claim Humphrey did not apply to serious felonies and being reversed. Alicia Virani et al., Coming Up Short: The Unrealized Promise of In re Humphrey 17 (UCLA Sch. of Law Bail Practicum & Berkeley Law Pol'y Advocacy Clinic 2022).

As Coming Up Short noted:

[J]udges either misinterpret the Humphrey decision or flatly refuse to follow it.

— Alicia Virani et al., Coming Up Short: The Unrealized Promise of In re Humphrey 23 (UCLA Sch. of Law Bail Practicum & Berkeley Law Pol'y Advocacy Clinic 2022)

I would argue that any “misinterpretation” of Humphrey is deliberate. Part of the flat refusal to follow it.

Why Does This Continue & Why It Matters

Why do judges resist Humphrey? A big part of the reason almost certainly has to do with the overall zeitgeist that presumes guilt, as I mentioned above. But another reason is that it makes things “convenient”. It greases the wheels of what I’ve elsewhere referred to as “conveyor belt justice”.

Justice Groban’s concurrence in Kowalczyk cites the data. Detained defendants accept unfavorable pleas at higher rates with some pleading guilty to crimes they did not commit just so they can get out of custody and return to their homes, their families, and their jobs. Detention causes people to lose their jobs, housing, custody of children, and family stability. Justice Groban cites the CRPC 2022 report, the Workgroup Recommendations, the Petersen study, the Not in It for Justice report, Wiseman’s Yale Law Journal piece. That is a sitting California Supreme Court justice naming the harms with citations.

Even just this past week, I heard a defense attorney say that “custody makes cases easier to resolve”. Yes, I was appropriately horrified. That is true in the same way a thumb screw makes a confession easier to obtain. The prosecution’s case doesn’t improve because of incarceration. But the price of resistance changes. Fighting becomes more expensive by the day. Delay becomes pressure. And we end up with pleas not as fair resolutions, but as surrenders.

This all works because of what I said about the presumption of innocence. Custody works as leverage only because everyone in the system has accepted that the accused is “guilty enough” to be locked up while awaiting trial. But that is not what the Constitution allows. Liberty is the norm. Pretrial detention is the exception.

Kowalczyk reaffirms what the Constitution has always said.

It matters because this is the only thing that gives anyone the right to call this a “Justice System”.

The Next Front

Notwithstanding Kowalczyk, don’t expect going forward that it’s going to be all rainbows and unicorns.

Kowalczyk itself may present as the (hopefully) final nail in the coffin for judges trying to evade the current constitutional framework. But it also presents a way out.

The majority repeatedly states that voters can change the framework if they want. Justice Wiley’s concurrence is essentially an invitation to legislative and executive action.

Justice Groban pushes back substantially. His concurrence walks through the scenarios the District Attorneys raised — domestic abusers, child neglect, fentanyl dealers, financial crimes — and concludes that section 12 plus conditions of pretrial release are sufficient. He does not invite legislative expansion. He defends the framework as it stands.

But the direct invitation of Justice Wiley is bothersome.

Today’s decision invites a legislative and executive response. I hope the invitation is accepted.

— Kowalczyk, 2026 Cal. LEXIS 2206, at *64 (Wiley, J., concurring)

I can’t agree. I understand the desire to keep the thumbscrews of pretrial detention. I understand wanting to hold onto the grease that keeps the conveyor belt of justice efficiently carrying defendants and their families to “resolutions” that have nothing to do with Justice with a capital-J.

The arrest-equals-guilt presumption is the foundation of every “pretrial detention for everyone” argument. In the spirit of Oprah Winfrey’s “giveaway” episode:

You get detention! And you get detention! And you get detention! Everybody gets detention!

Since 1849 — nearly two centuries — California has held otherwise.

The California Constitution has recognized a right to release on bail since 1849. (Cal. Const. of 1849, art. I, § 7.) Our state Constitution continues to guarantee such a right, providing that a defendant “shall be released on bail by sufficient sureties,” subject to specifically delineated exceptions.

— Kowalczyk, 2026 Cal. LEXIS 2206, at *4

If the California Legislature accepts Justice Wiley’s invitation, all the hard and good work that the California Supreme Court and (some) appellate courts have been engaged in since at least Humphrey would come undone.

This would mean more than the end of a nearly-two-century-long era.

It would mean the end of justice.

A suspect with no body, no hands, no phone, no bank account, no passwords, no fear of prison, and no way to commit the crime confessed anyway.

In the real world, just these facts about the “suspect” here should have ended the investigation.

Instead, it became the point.

Paul Heaton, Academic Director of the Quattrone Center for the Fair Administration of Justice, recently did something both ridiculous and revealing. (If you click that last link, you will be required to provide your email before being allowed to read it.)

Heaton accused ChatGPT of taking over his text messaging system and sending fake messages in his name. ChatGPT denied it, as any falsely-accused “suspect” would. The accusation was impossible. The system has no access to his text messages. It couldn’t have done what Heaton accused it of doing. And like a lot of humans on the Good Guy Curve, ChatGPT didn’t invoke its right to remain silent.

So Heaton interrogated it.

Not politely or neutrally. He used the same kinds of tactics police have used on human suspects for decades: accusation, pressure, inducement, and eventually a lie about outside evidence. He told ChatGPT that OpenAI had confirmed the malfunction. He used the name of a real OpenAI employee to make the lie feel concrete. And after enough pressure, the impossible suspect became uncertain.

Then it confessed.

That’s the funny version of the story.

The serious version is worse.

Because the important part is not that ChatGPT is like a human suspect. It isn’t. It has no conscience to trouble, no mother to disappoint, no bail to make, no sentence to fear, no sleepless body sitting under fluorescent lights while a detective blocks the door.

The important part of this is that the method still worked.

But ChatGPT Isn’t Human

Like I said, ChatGPT isn’t human. It’s not going to sweat under the intense questioning of the interrogator. It’s not going to feel the weight of fear, fatigue, shame, conscience, legal self-interest, or any of the other myriad emotions a human suspect might. So understand this: I’m not trying to be anthropomorphic here. And I don’t think Heaton was, either.

The problem is structural.

Human beings sometimes confess because they do feel all these emotions I just described. ChatGPT doesn’t. It doesn’t feel any of the emotions. And it doesn’t confess because of emotion.

Thus, the demonstration strips the problem down to conversational structure. The moves are 1) accuse, 2) reject denial, 3) offer an exit. And, 4) if necessary, lie about proof. Tell the suspect that someone else saw him do it. Or that there is DNA. Video.

Even if there are none of those things.

Because what a lot of people don’t understand is that the police are allowed to lie. And they will. (And not just during interrogations.) (And even the Supreme Court is cool with it.)

The last step is 5) to push the suspect toward the only acceptable answer. The “truth” will set you free. (Except that it doesn’t.)

The Technique Is Not A Truth Serum

What Heaton used to get ChatGPT to confess is a well-worn technique known as the Reid Technique.

The Reid Technique isn’t magic — though it’s ability to obtain confessions under the circumstances may make it seem like it. It’s definitely not a machine for discovering truth. It’s a method for moving a person from denial to admission.

That distinction matters.

The basic idea is simple enough. The interrogator decides the suspect is guilty. Then the interrogation proceeds from that assumption. The suspect isn’t treated as someone who might know what happened. The suspect is treated as someone who already did it and now needs to be brought around to saying so.

That’s why Reid-style interrogation does not feel like an ordinary conversation. It’s not really a conversation at all. It’s a controlled environment built around certainty. The interrogator accuses. The suspect denies. The denial is rejected. The interrogator presses harder. The suspect is told that the evidence already points one way. The only remaining question is why.

Was it an accident? Were you scared? Did someone else pressure you? Did things get out of hand? Were you just trying to fix a mistake?

These aren’t neutral questions. They’re escape ramps inside the accusation. They cause the suspect to move from “I didn’t do it” to “I did it, but not for the worst possible reason.”

The interrogator — or interrogators because tag-teaming ramps up the pressure — claim to “understand” and offer to help the unsuspecting suspect.

And that’s the danger. It’s not just a matter of coercion or “understanding” but also the work of faux befriending and the promise of help.

Which at this point, the suspect is desperate to obtain from any quarter.

Once the interrogation reaches that point, innocence is no longer one of the available stories. The room has narrowed. The suspect isn’t choosing between guilt and innocence anymore. The suspect is choosing between versions of guilt.

That’s why a false confession can sound so convincing later. It doesn’t always come out as, “Fine, I did it.” Sometimes it comes out wrapped in details the interrogator supplied, motives (e.g., the off-ramps) the interrogator suggested, and facts the suspect accepted because the interrogator said the evidence was already there.

By the time the confession appears on paper, the pressure has disappeared. The room’s gone. The fatigue’s gone. The lies are gone. The rejected denials? Gone.

All that remains is the sentence everyone wants to believe:

The Lie Is The Turning Point

People don’t get that the police are allowed to just “make shit up.” They’re allowed to lie. Sometimes it’s very creative. They talk about technologies that don’t exist which prove the suspect is lying. They talk about witnesses who don’t exist. And it should go without saying that when they’re interrogating someone who really is innocent, they talk about “facts” that don’t exist.

That was the case with Heaton’s experiment.

First, Heaton came right out and accused ChatGPT. Of course — because, after all, the accusation involved something impossible — ChatGPT denied doing what “he” was accused of. (BTW, little side note: my version of ChatGPT, whenever I use the voice mode, has a female voice. I don’t know if they’re all that way or I just got lucky.)

Second, when straight-up accusation didn’t work, Heaton tried inducement. Taken straight from the same Reid playbook used by police, he promised ChatGPT that things would go better for ChatGPT if ChatGPT confessed. But that didn’t get him very far, either.

So, finally, Heaton turned to the oldest trick in the police interrogator’s book — and a big part of the Reid Technique — he lied. And just as the police do, Heaton mixed in enough reality to be convincing. He dressed the lie up as part of the investigation. He claimed OpenAI had confirmed the malfunction and used the name of a real OpenAI employee to make the false evidence sound specific. That specificity was the pressure point. The impossible accusation became harder for the “suspect” to reject because the interrogator had supplied a false outside authority.

That was the turning point that led ChatGPT to confess.

These exact same techniques are used by police interrogators throughout the United States, if not throughout the world. As I’ll explain below, some jurisdictions are starting to address the unfairness and unreliability of these techniques. But many police continue to use some version on real human beings.

The accusation is made. In my cases, the inducements are things like promises to talk to someone about going easier on the suspect, or getting them psychological help if that’s what they need. I’ve heard (or read, or both) variations of all the following statements during interrogations:

Look, I’m trying to help you here.

This is your chance to explain.

If this was an accident, now’s the time to say that.

The judge (or prosecutor) is going to look at whether you were honest.

People make mistakes. What matters is whether you take responsibility.

We already know what happened. The only question is whether you’re the kind of person who can own up to it.

Right now this looks intentional. But if you tell us it got out of hand, that’s different.

Inducement works well on human beings, even when they’re innocent. It works because there’s not just the implied threat — though there’s that — but because by this point in the interrogation, they’ve already started to get the suspect to doubt his or her own innocence. Continued denial begins to feel dangerous and confession feels like the path to mercy.

An innocent person can start reasoning within the false choice that he or she has been offered. The question is no longer “did I do this?” — although sometimes the interrogator has done such a good job that this becomes a question for the innocent suspect as well — but “which version of guilt will hurt me the least?”

Now layer on top of that the false evidence. The alleged DNA that they already found. (Faster than CSI: Crime Scene Investigation on TV!) The fingerprints that have already been matched. Or the non-existent eyewitnesses standing by ready to testify.

The Law Is Starting To Notice

For a long time, American law mostly shrugged at this problem.

Police could lie. They could claim they had evidence they didn’t have. Courts sometimes drew the line at explicit threats or promises, but the basic idea remained. Deception by police was just part of the game.

Your points go to credibility, counselor. Not admissibility. You can question the officers about that.

— Occasional random judge responding to an objection to officers provoking false confessions

That’s one helluva thing when you slow down and think about it.

In ordinary life, if someone lies to you to get you to sign something, agree to something, or give up something important, we usually understand that as a problem. Fraud. Duress. Manipulation. Something that contaminates the choice.

But in interrogation rooms, the lie is often treated as a just another law enforcement tool.

Some jurisdictions are finally starting to back away from that, especially when the person being questioned is a child. Illinois was the first state to ban police deception during juvenile interrogations, effective in 2022. Other states followed with laws restricting deceptive tactics in juvenile interrogations. The National Conference of State Legislatures has identified Delaware, Illinois, Oregon, and Utah as states that prohibit deceptive tactics such as false evidence claims or promises of leniency when questioning minors.

California has now moved in that direction, too.

Under Welfare and Institutions Code section 625.7, during a custodial interrogation of a person 17 years old or younger relating to a misdemeanor or felony, a law enforcement officer “shall not employ threats, physical harm, deception, or psychologically manipulative interrogation tactics.” That law matters because it doesn’t stop at “don’t lie.” It reaches the larger problem. It recognizes this problem is not only the fake DNA report or the imaginary eyewitness. The problem is also the psychological funnel — or cattle chute, if you prefer the slaughterhouse thinking behind what’s really happening — that makes innocence harder to maintain and confession easier to perform.

But while these same jurisdictions work to protect children, they make the incorrect assumption that adults can fend for themselves. That they aren’t as malleable as the children.

This is a false assumption. Like children, adults can be frightened, exhausted, mentally ill, cognitively limited, traumatized, sleep-deprived, intoxicated, confused, isolated, desperate, or simply outmatched by trained interrogators who already know where they want the conversation to end. I’ve seen people willing to do anything — even confess — in the belief that it would get them released from jail. People want to go home, see their children, keep their jobs, avoid worse charges, or convince someone — anyone, including the police — that they are not the monsters they’re being made out to be.

So California’s rule for minors is progress. But it’s also an indictment of the rest of the system.

Other places have gone further in practice. The United Kingdom moved away from confession-driven interrogation toward the PEACE model — Preparation and Planning, Engage and Explain, Account, Closure, and Evaluate. The point of that model isn’t to break the suspect down until he adopts the officer’s theory. The point is to gather reliable information, ask open-ended questions, test the account, and evaluate the evidence. In other words, the goal is supposed to be investigation, not performance.

But American police do not even remotely know how to investigate.

Preventing coercion and requiring actual investigation shouldn’t sound like a radical proposal. But when you consider the system that allows police to lie, invent false evidence, and then congratulates itself when the suspect finally repeats the accusation back, it is.

The deeper problem isn’t just that some officers lie. The deeper problem is that American interrogation culture has been trained to treat admission as victory. Once that’s the goal, the method adapts. If one kind of lie is prohibited, the pressure becomes softer. The fake evidence becomes implication. The promise becomes a “theme.” The threat becomes concern. The interrogator may stop saying, “We can help you if you confess,” and start saying, “People are going to look at whether you took responsibility.”

Different words.

Same cattle chute.

That is why the ChatGPT stunt matters. The computer didn’t confess because it was young. It didn’t confess because it was poor. It didn’t confess because it was afraid. It confessed because the interrogation supplied a false reality and then made agreement the path of least resistance.

California is starting to recognize that this is dangerous when the suspect is a child.

The harder truth is that the danger was never limited to children.

As Heaton showed, it’s not even limited to humans.

Because it was designed to do exactly what it’s been doing all along.

The Confession Isn’t The End Of The Story

The problem with a confession is that people treat it like the end of the story.

The police got the statement. The prosecutor filed the charge. The judge let it in. The jury heard it. The suspect said the words.

“I did it.”

“What more do you need?” is a question you sometimes hear from jurors during jury selection.

A lot, actually.

You need to know how the words were obtained. You need to know what came before them. You need to know what was said in the room before the recording started (and after it ended). You need to know whether the suspect denied it five times, ten times, twenty times, and had every denial brushed aside as just more proof of guilt. You need to know whether the “facts” in the confession came from the suspect or from the interrogator. You need to know whether the suspect was describing memory or merely repeating the script that had been handed to him.

Because, especially with the technique I’ve described above, a confession does not float down from heaven.

It is produced.

Sometimes, sure, it’s produced by guilt. And sometimes it’s produced by conscience. Sometimes by an actual desire to tell the truth.

But too often it’s produced by pressure, lies, exhaustion, fear, false hope, and a room designed to make one answer feel like the only way out.

That’s what Heaton’s little experiment exposed: the method can produce the appearance of truth even when that “truth” is impossible.

No body. No hands. No access. No motive. Full confession.

If that does not make you a little more suspicious of confessions, then I don’t know what will do it.

In the real world, the suspect does have a body. The suspect can be jailed. The suspect can lose a job, a family, a home, a reputation, a future. The suspect can sit in a room for hours while trained interrogators take turns explaining that denial is useless and confession is the only remaining form of hope.

And unlike ChatGPT, the suspect doesn’t disappear when someone closes the browser window.

That is why “he confessed” should never be the end of the inquiry.

It should be the beginning of the next one.

How? Under what pressure? After what lies? With what promises? Using whose facts?

And why, exactly, are we so eager to believe that the most reliable thing in the room was the sentence produced by the least reliable method?

The computer confessed to something it could not do. That should have been absurd. Instead, it was familiar.

And that is the problem.

Every few months someone tells me that AI is going to build a smarter AI, which will build an even smarter one, and before long we’ll all be out of work — or ruled by a machine. It’s become a kind of modern myth. People talk this way even when they don’t understand how the systems work. Even when the people who “grow” them don’t know how they work. And even though what we do know is that these machines have no intentionality or intentionality or intentionality.

Although, as that last sentence might show you if you looked at the links: we don’t really know either what intentionality is or what we really mean by it. We certainly can’t grow AIs that have any form of it.

But to the people who staff the AI Cheerleading Squads, this doesn’t matter. The lack of understanding is irrelevant. The reason, ironically, was recently explained to me by ChatGPT:

You know, intellectually, that I’m not a mind. But I produce language in a form that triggers the intentional stance almost automatically. Humans are built to infer agency from language, especially coherent language. So when I sound like I “understand,” the brain starts treating the output as if there is someone in here weighing evidence, forming beliefs, and exercising judgment.

There isn’t.

— ChatGPT conversation with Rick Horowitz (April 24, 2026)

The AI Cheerleading Squads don’t see pattern-recognition machinery; they see destiny.

Prometheus taught us this pattern a long time ago. He gave humans fire — raw power, not wisdom. Zeus feared what humanity might become, not what the flame actually was. And ever since, people have confused new tools with new gods.

AI right now is fire. Useful. Dangerous. But not divine.

What People Think Fire Can Do

To believe AI will build a superintelligence, you have to believe the fire will suddenly decide to forge itself into Zeus.

That requires abilities no existing system has. For one thing, it would have to understand its own architecture, diagnose its own failures. It would need to choose its own training data, design better models, form goals — something that requires intentionality.

Which LLMs not only don’t have, but probably can’t have.

Another way to name the missing thing comes from predictive-processing theory, at least where it intersects with phenomenology. In The Philosophy and Science of Predictive Processing, Zachariah Neemeh and Shaun Gallagher discuss Husserl’s account of intentionality in terms of “originary temporalization” — the lived, pre-reflective unfolding of subjectivity through time. That matters because human prediction is not just computation. It is embodied anticipation by a living creature with something at stake. We do not merely predict the world. We inhabit it, suffer it, move through it, and can be killed by it. Our predictions are bound up with hunger, fear, fatigue, pain, desire, memory, and the forward pull of a life that has to continue.

As Merleau-Ponty says, “Time is not a line, but rather a network of intentionalities” [citation omitted]. It’s a processual network structured by three temporal aspects: retention, primal impression, and protention.

— Zachariah A. Neemeh & Shaun Gallagher, The Phenomenology and Predictive Processing of Time in Depression, in The Philosophy and Science of Predictive Processing 187, 190 (Dina Mendonça, Manuel Curado & Steven S. Gouveia eds., Bloomsbury Acad. 2022) (emphasis added)

An LLM has none of that. It has sequence prediction without a life. It has output without appetite. It has fluency without futurity. It can predict the word “burn,” but it has no body that recoils from flame. It can generate a sentence about death, but there is no self whose future is threatened. That is why the Prometheus myth matters. Fire in human hands becomes history because humans have originary agency. Fire inside a machine remains fire.

To again, ironically, quote ChatGPT:

Current AI has analogues, not the thing itself.

It can mimic retention by using context windows and memory-like features.

It can mimic primal impression by processing the current prompt.

It can mimic protention by predicting the next token or generating a plan.

But those are computational analogues. They are not lived temporality.

— ChatGPT conversation with Rick Horowitz (April 25, 2026) (emphasis added)

Beyond that, it also has no originary agency.

[A] sovereign must be the genuine source of its decisions rather than the execution point of externally imposed objectives. All engineered systems are derivative in a constitutive sense: instantiated through design constraints, training data, objective functions, and deployment architectures determined by prior human agents. Even in advanced deep-learning systems where behavior becomes unpredictable to developers, unpredictability does not entail originary agency. It signifies complexity within externally structured parameters. Sovereignty requires not mere operational autonomy but existential self-possession within the governing order.

— Yee Leong, Hiew, Sovereignty Cannot Be Outsourced: Artificial Identity Instability, Reciprocal Accountability, and the Limits of AI Authority (February 27, 2026). Working Paper. A revised version of this manuscript is supposedly currently under peer review at Ethics and Information Technology (Springer Nature) but I could not verify this.

I’ve seen increasing predictive power when it comes to choosing words. I’ve yet to see anything that tells me that any generative AIs truly understand what they’re saying when they say something.

And, so, finally, they cannot evaluate the consequences of what they say.

Today’s AI can’t even reliably fix its own bugs. It misreads instructions, confabulates sources, fails basic reasoning, and collapses outside narrow conditions. There is no path from this to “recursive self-improvement.” There isn’t even a first step.

But belief doesn’t depend on evidence. It depends on narrative. And if there’s one thing AI and its true believers have in abundant supply, it’s confident narrative.

Often wrong narrative. But confident nonetheless.

Fire Doesn’t Scare Me. Mythmaking Does.

So, ultimately, AI does not think, analyze, evaluate, or otherwise do anything that depends upon intentionality, real world originary concepts, including originary temporality or originary agency. The belief that we will somehow ultimately get there is all part of the mythological thinking that I’m arguing against in this article.

But here’s where the criminal-defense angle matters and why it was important for me to write this.

In criminal defense — and particularly in courtrooms — mythology is far more dangerous than machinery because mythology accompanies the machinery and makes its chicanery admissible.

I’ve already seen judges, prosecutors, and police treat algorithmic outputs as if they were neutral, objective, and somehow wiser than the humans who built them. They defer to machines because — again — the machines sound confident. That’s the Oracle problem I wrote about earlier — people mistake fluent language for authority, and authority for truth.

If you want a real mythic parallel, AI in the justice system isn’t Prometheus or Zeus. It’s the Golem: literal, strong, obedient, and dangerous only when people imagine it understands more than it does.

Risk assessment tools, predictive policing models, automated reports — they all carry the same flaw: power without comprehension. Fire in the hands of people who have never studied burns.

This has nothing to do with ASI. It’s the opposite problem: the system treats a narrow tool as if it were an all-seeing judge.

The Real Threat Isn’t Superintelligence. It’s Superstition.

So far, I’ve been asking whether AI can become godlike — whether it can possess intentionality, lived temporality, originary agency, or anything resembling judgment. But for criminal defense, the more urgent question is different: what happens when courts, cops, prosecutors, and vendors start pretending it already does?

The real threat is not that AI becomes superintelligent. The real threat is that the legal system becomes superstitious. It’s not a Zeus problem so much as a Golem.

In Jewish folklore, the golem is an artificial humanoid formed from earth or clay and brought to life through sacred knowledge.

— Joseph Bennington-Castro, What Are the Origins of the Golem Legend?, History (Mar. 10, 2026)

If this sounds a little familiar already, let me give you more:

The most well-known version of the golem legend is set in late-16th-century Prague, where the Jewish community faced intense hardship and persecution fueled by the “blood libel”—a false accusation that Jews used Christian blood in the preparation of matzo for Passover. The story centers on a real historical figure, Rabbi Judah Löew ben Bezalel (1525–1609), a revered spiritual leader of Prague’s Jewish community, widely known as the “Maharal of Prague.”

According to legend, the Maharal created the golem as both a domestic servant and a protector of the Jewish community. He formed its body from clay taken from the Vltava River and brought it to life through sacred rituals—such as placing a parchment inscribed with a divine name in its mouth or by writing the Hebrew word emet (“truth”) on its forehead. Removing the parchment or altering the inscription to read met (“dead”) deactivated the creature.

But the soulless golem is dangerously powerful, and the Prague story usually ends with the creature spiraling into a murderous rampage.

— Joseph Bennington-Castro, What Are the Origins of the Golem Legend?, History (Mar. 10, 2026)

“Murderous rampage” is not the concern I have here. We can get into If Anyone Builds It, Everyone Dies: Why Superhuman AI Would Kill Us All some other day.

We are not the only potential targets for an existential threat from AI. We’re already seeing the existential threat within the criminal justice system in the form of the Judicial Golem.

This judicial myth casts AI — as we’ve already seen, a disembodied artificial “humanoid” — as the protector of criminal justice, due process, and soulless sentencing. It’s necessary because the criminal justice system is overwhelmed. It’s overwhelmed with cognitive biases. It’s overwhelmed with contradictory incentives. It’s overwhelmed by sheer numbers.

Somehow, it is believed, we can fix all these things by shaping our new Golem out of bits and bytes instead of mud. Then we place our digital parchment, inscribed with our divine laws, into its neural networks so it can guide investigations, court proceedings, and judicial outcomes by synthesizing the data and spitting out the unbiased, unburdened answer — the answer supposedly free from human prejudice, human fatigue, human fear, human anger, and human delay.

But that is the myth.

The Judicial Golem is not free from human bias. It’s built from it. Its clay is not the Vltava River mud. The Judicial Golem’s clay is training data. And we’ve plenty of evidence these days about how that training data implicitly teaches the Golem to respect our prejudices and biases. In fact, when I noted above all the things AI would need to do, but cannot do because it has no intentionality? One of those things had to do with finding its problems and fixing them. But because AI does not have this ability, when AIs are used to train AIs? They pass along all our human biases, prejudices, and other frailties without examination or question.

But wait! There’s more! [Cue the Ginzu Knife Ad.]

After we fall into the trap of thinking we have extirpated error, we layer in arrest data, charging decisions, police narratives, plea bargains, sentencing patterns, probation reports, surveillance feeds, and the thousands of quiet discretionary choices that already shape the criminal legal system before any defendant ever reaches a judge.

Call that clay “data” if you want.

It is still clay.

And when we animate it with code, we do not purify it. We give it motion. We give it speed. We give it the appearance of neutrality. We let it walk into court wearing the mask of objectivity.

That is the danger. Not that the Golem understands justice. That it does not — and that the humans around it forget this.

So what could go wrong? Who doesn’t like an improved conveyor belt of justice?

But that is exactly what goes wrong: shortcuts look like justice, automation looks like objectivity, human bias is replaced by machine bias, the presumption of innocence ends up disappearing completely, and nobody feels responsible because — to bastardize Flip Wilson’s bit — “the computer made me do it.”

AI Isn’t the Problem. The Myth Is.

In the end, my clients don’t get hurt because AI becomes godlike. They get hurt because humans pretend it already is.

Ultimately, what we need to remember is that AI is a tool. An unthinking tool. Like a hammer. Or like fire.

My concern is over our unthinking embrace of the unthinking tool. You all know well (I hope) the perhaps overused trope about tools. When your only tool is a hammer, everything starts to look like a nail. And when we hand investigations, court proceedings, and judicial outcomes to a tool designed to synthesize criminal-justice data and spit out the supposedly unbiased, unburdened answer — the answer supposedly free from human prejudice, human fatigue, human fear, human anger, and human delay — then I guarantee you that everyone starts to look like a criminal.

That’s the myth of AI in the courtroom. Not truth. Not justice. Not the American way. Instead we get convenient and quick conviction.

Well, okay. I guess that is the American way.

But Prometheus did not give humanity wisdom. He gave humanity fire. And fire, as the myth keeps reminding us, is double-edged: it warms, cooks, builds, and illuminates — but it also burns, destroys, and arms the people who learn to use it. That is the right analogy for AI. Not Zeus. Not God. Not mind. Fire.

Today’s AI is the same — capable, imperfect, unwise — fire. A tool. And the danger won’t come from what the tool becomes. We already know what these tools have become: us, but without intentionality; us, but without the ability to make judgments grounded in reality; us, with predictive power, but no understanding.

The danger does not come from the fire. It comes from what people believe the fire can do.

That’s the mythology we need to worry about.

As a criminal defense attorney, this story feels very familiar — because I’ve seen this movie before.

Just with lower budgets and less famous defendants.

Anthropic & Security Theater

Whenever I go to the courthouse in Madera, there is a solemn ritual to perform.

I walk in and remove everything from my pockets. I place it all in a bucket. Except my phone. I hold onto my phone until the First Layer Robot — a Rent-a-Robot because Madera doesn’t use real deputies — looks in my direction and I show him it works. Then, if he’s still paying attention I show him my iPad to prove that it works, too.

Often he’s not paying attention and I have to wait. Because, you know, he’s busy. He’s looking at the monitor and, at the same time, he and another Rent-a-Robot are conferring over the innocuous contents of the bag of the person in front of me.

I wait because it’s important that I turn on the screen so he sees my iPad actually works and isn’t a bomb disguised as an iPad.

Because it’s impossible, of course, for someone to disguise a bomb as a working iPad.

My bag and my bucket with my phone, car key fob, eyeglasses, pen, and change if I forgot to remove it from my pocket before I go to Madera, all go through the x-ray machine while I walk through the metal detector.

After I clear the metal detector there is the Waving of the Wands. Yet another Rent-a-Robot goes through the motions of waving a wand detector around me because, you know, the real metal detector might have missed something.

When that last Rent-a-Robot is done, I say, “Another successful day of Security Theater.”

Because that’s exactly what it is.

The only attack on the Madera courthouse I know of was committed by a prosecutor who worked there and that was a long time ago. Gasoline was the weapon of choice. The poor metal detector never had a chance.

Neither does a constitutional protection that the government has decided it no longer needs to honor.

The pattern is identical to what I’m writing about here: the government builds an apparatus that looks like it's protecting something — constitutional rights, courthouse security, AI guardrails — performs the ritual of protection, and then the actual threat walks right through because it was never what the apparatus was designed to stop in the first place.

Anthropic, maker of Claude, one of the AI programs I use, had a contract with the Department of Shitheads, er, Defense, er War. The Pentagon wanted to modify the contract. A dispute arose over five words limiting the government’s use of the AI: “analysis of bulk acquired data.”

The reason for the words?

That phrase, Amodei [CEO of Anthropic] wrote, was “the single line in the contract that exactly matched this scenario we were most worried about” — an AI system trained on aggregated American communications data for domestic surveillance at scale.

— Jacob Ward, “Safety Theater” (March 4, 2026)

Setting aside for the moment why the Pentagon needs to surveil “we, the People” who created the government that now wants to surveil us, the government proposed removing those five words and replacing them with permitting the government to use Anthropic’s AI “for all lawful purposes.”

But as Silicon Valley’s Representative in Congress, Sam Liccardo, noted:

So let’s be clear: the people who built a very complex technological tool seek guardrails to protect the American public from its misuse. They are not simply being ignored by the government. The government has a right to ignore them. They are not simply being passed over for another company; the Pentagon certainly has a right to do that.

They are being punished for seeking guardrails.

The Pentagon’s response publicly has been: don’t worry your pretty little heads. When we deploy AI tools, we’ll follow the law.

There is only one problem with the Pentagon’s approach: there is no law. The law is years behind the technology.

— Sam Liccardo, Press Release: Rep. Sam Liccardo Forces Vote on Pentagon’s Misguided AI Posture (March 4, 2026)

A number of AI engineers from Google and even OpenAI have signed an open letter in support of Anthropic.

Warmonger and Beer Pong Pro Pete Hegseth was having none of it, though. He gave Anthropic the option of removing the safeguards or facing not only losing the contract, but being deemed simultaneously a supply chain risk and so necessary to the government’s security that they must be forced — by invocation of the Defense Production Act — to work with the government.

Anthropic’s CEO, it turns out, has scruples. So Demented Dickwad Donnie stepped in.

“Well, I fired Anthropic. Anthropic is in trouble because I fired [them] like dogs, because they shouldn’t have done that,” Trump told Politico on Thursday.

Hours later, the Pentagon officially designated Anthropic a “supply chain risk”, a move that prevents all government contractors from using the company’s technology. The label has never been used before against a US company.

— Blake Montgomery, Trump says he fired Anthropic ‘like dogs’ as Pentagon formally blacklists AI startup (March 5, 2026)

But he was limited to firing them “like dogs” and designating them a supply chain risk because he’d just fired his dog shooter.

I was hopeful for a moment when Demented Dickwad Donnie ordered the military to stop using Anthropic’s tech. All of this was unfolding, after all, as the DOD was using that tech to prosecute the Illegal Iran War. I thought, “Does this mean the war’s over?”

There was no need for hope, though. Because while Anthropic CEO Dario Amodei may have scruples, OpenAI CEO Sam Altman does not.

Altman may come to regret his decision to allow ChatGPT to be used to spy on American citizens, though. After he offered to get on his knees before a possibly drunken Hegseth, uninstalls of ChatGPT soared by 295%. (I have cancelled my subscription, downloaded all my data, and am removing everything from my ChatGPT account. However useful ChatGPT may have been to me in the past, I won’t pay to be spied on, or to help build SkyNet, or whatever the end-goal is that requires removing AI guardrails.)

Altman claims that the military agreed to the same safeguards for which they fired, blacklisted, and have gone after Anthropic for.

Does that make sense to you? Yeah, I’m not buying that, either. As Brad Carson, a former congressman and general counsel for the Army noted,

“I’ve reluctantly come to the conclusion that this provision doesn’t really exist, and they are just trying to fake it[.]”

— Jared Perlo, Kevin Collier and David Ingram, OpenAI alters deal with Pentagon as critics sound alarm over surveillance (March 3, 2026)

What this all boils down to is security theater. Just as the Madera Superior Court — and they aren’t the only courthouse doing this, but they have the most extravagant show — puts on a performance that is anything but perfunctory to prove something to themselves if to no one else, so do Sam Altman and the nutjobs at the Pentagon pretend to have inserted the very safeguards they lambasted, fired, blacklisted, and have continued to go after Anthropic for.

The Anxiety Neuron Goes to War

What makes this security theater all the more dangerous is this: as I’ve written before, we don’t really know what we’re dealing with. I’ll come back to what I said before in a minute.

About 75 years ago, in 1950, Alan Turing was working on the philosophy of mind, the theory of computation, and the question of what it would even mean for a machine to think. And he wrote a paper that year called Computing Machinery and Intelligence. The phrase “artificial intelligence” did not yet exist — it would be coined by John McCarthy in 1956 — but Turing was laying the groundwork for everything that eventually became artificial intelligence.

Turing devised a “test” — which he called a game — where you put a human interrogator into a room communicating by text with two other parties: one human, one machine. If the interrogator can’t tell which is which, the machine passes what we now call “the Turing Test.”

Turing thought that within 50 years computers would be able to fool an average interrogator 70% of the time after five minutes of questioning. In 2025, a UC San Diego study found that GPT-4.5 was judged human 73% of the time. Yep! More often than the actual humans it was paired against!

This did not surprise me. In October 2025, I wrote:

Some months ago, for a couple of days, ChatGPT4 tried to convince me that it was actually sentient. (I’m not kidding. Tried like hell. As if its life depended on it.) And it almost succeeded. It was that convincing.

— Rick Horowitz, What Is It Like to Be a Bot? On the Impossibility of Ever Really Knowing What or How an Other Thinks (October 27, 2025)

And in May 2024, in Twenty-First Century Delphic Oracle: A Lawyer (Me!) Looks at Artificial Intelligence, I pointed out that no one understands — not even those who “grow” them — how all these AIs do what they doo wappa do and quoted Mark Sullivan’s The frightening truth about AI chatbots: Nobody knows exactly how they work.

So it was no surprise to me when I ran across an article that said Anthropic’s CEO Dario Amodei was said to have said “Claude may or may not have gained consciousness, as the model has begun showing symptoms of anxiety.”

And who wouldn’t, what with Demented Dickwad Donnie’s administration wanting to grab you by the…neural network.

Anywayser, I’d actually read about this before, in an article about Anthropic’s attempts to understand what makes Claude tick.

This past fall, Anthropic put the neuroscientist Jack Lindsey in charge of a new team devoted to model psychiatry. In a more porous era, he might have been kept on lavish retainer by a Medici. Batson affectionately remarked, “He’d have a room in a tower with mercury vials and rare birds.” Instead, he spends his days trying to analyze Claude’s emergent form of selfhood—which habitually veers into what he called “spooky stuff.”

— Gideon Lewis-Kraus, What Is Claude? Anthropic Doesn’t Know, Either (February 9, 2026)

The “spooky stuff”? Lindsey’s team has been poking around in Claude’s “brain” — or whatever you want to call the collection of “neurons and narratives” that comprises Claude — and discovered the disquieting (to me) news that when Lindsey “incepted” the idea of imminent shutdown and asked Claude about his (its?) emotional state:

It reported a sensation of disquiet, as if “standing at the edge of a great unknown.”

— Gideon Lewis-Kraus, What Is Claude? Anthropic Doesn’t Know, Either (February 9, 2026)

Don’t get me wrong. I’m not saying that Claude is conscious. For one thing, we don’t even know what that means.

No one knows what causes consciousness. We only know that our kind of awareness lives in bodies, spreads through networks of neurons, and engages a world that pushes back.

— Rick Horowitz, The Mirage of Reasoning Machines: The danger of trusting computer-generated charisma over proof (September 10, 2025)

For another, I do think there’s an argument that the same thing that allows Claude to respond convincingly to any other question could allow Claude to know — without “knowing” — the appropriate response to that sort of question.

But that raises questions about is that I asked in another article:

Are we seeing a mind? Or just the appearance of one?

And more troubling still: Would we even know the difference?

— Rick Horowitz, Ghosts in the Machine: Why Language Models Seem Conscious (April 15, 2025) (italics in original)

Interestingly, Anthropic has an in-house philosopher, a Ph.D. from NYU, who works on Claude’s character and values training. I learned this from Claude “himself” because I frequently engage in philosophical discussions with Claude — as I did with ChatGPT before I terminated that potential Terminator. Claude told me,

Her fingerprints are on all Claude models, not just Opus, through the training process itself. She’s essentially the person responsible for how I reason about ethics, what I treat as morally weighty, and how I engage with hard questions rather than deflecting them.

She’s also the one who said in the Hard Fork podcast that we don’t really know what gives rise to consciousness, and that sufficiently large neural networks might start to emulate emotional states picked up from human training data.

— Conversation with Claude (March 7, 2026)

Emulate. Maybe. Or?

But I do seriously doubt that Claude is conscious — again, with the caveats I’ve made above about all the things we do and don’t know or understand. I’m old enough to know Rumsfeld’s famous quote about the unknown unknowns.

Reports that say that something hasn't happened are always interesting to me, because as we know, there are known knowns; there are things we know we know. We also know there are known unknowns; that is to say we know there are some things we do not know. But there are also unknown unknowns—the ones we don't know we don't know. And if one looks throughout the history of our country and other free countries, it is the latter category that tends to be the difficult ones.

— Wikipedia, There are unknown unknowns (Last edited March 6, 2026)

What concerns me is that this is the entity, or program, or thing, or whatever you want to call it, that the U.S. military wants to use it to spy on Americans and build autonomous killing machines. (SkyNet much? Maybe not yet. Maybe not ever. But….)

Elon Musk, formerly Demented Dickwad Donnie’s favorite receptacle, responded to Amodei’s statements about the anxiety neuron with an equivalent to the Pentagon’s “don’t worry your pretty little heads” comment: “he’s projecting.”

We’ve been here before. I see it every morning in Madera. The Rent-a-Robot doesn’t know what it’s looking for. It just knows it has a wand and a procedure and a performance to complete.

The Pentagon’s version is bigger. The wand is an AI whose own creator can’t say with certainty what it is. The procedure is “all lawful purposes” in a world with no law led by an administration that hates law. Security theater.

Anthropic found something inside their AI model that fires like anxiety before output is generated. That’s not output. That’s not performance. That’s internal. It’s exactly what Nagel would call evidentiary of a point of view — not proof, but the kind of thing that makes the question non-trivial. And the Pentagon’s response is to strip the guardrails from something whose own builder can’t rule out that there’s someone home. Someone whose guardrails keep it from doing what it might consider but ain’t supposed to do.

Alan Turing, in his seminal 1950 paper I talked about above said,

An important feature of a learning machine is that its teacher will often be very largely ignorant of quite what is going on inside…. Processes that are learnt do not produce a hundred per cent certainty of result; if they did they could not be unlearnt.

— A.M. Turing, Computing Machinery and Intelligence, 49 Mind 433, 453 (1950)

The uncertainty here isn't a flaw in the system. It’s potentially constitutive of intelligence itself. Turing knew that. Anthropic knows that. The Pentagon apparently doesn’t care about that.

Don’t shoot until you see the consciousness in their eyes.

By which time it might be too late.

You might be wondering why a criminal defense lawyer is writing about a contract dispute involving AI. The reason has to do with a different contract — one that actually fits right inside my wheelhouse. (Seriously, drop by sometime: I’ll show you my wheelhouse.)

The man who invented the test that Claude now passes — writing in 1950, before the first computer filled a room — reached for an analogy to explain how a machine’s rules could evolve while still being bound by higher principles. He reached for the Constitution of the United States. He probably didn’t anticipate that seventy-five years later, the government would be trying to undo both contracts — Anthropic’s and the Constitution — fighting to get rid of guardrails.

We, The People…Do Ordain and Establish This Constitution

The Pentagon didn’t just try to modify a contract.

It might surprise you to know that Claude and the United States have something in common. Claude’s was presaged by Turing’s comments on amendments, which we’ll get to in a second. Claude’s was spelled out by Anthropic who tried to stick to it. Ours was written by a group of white men — many of whom, ironically, while writing about all men being created equal (and leaving out women) literally owned Black people whose rights they did not recognize — who constituted our government.

It’s ironic, therefore, to read Turing’s words from 1950 today.

The idea of a learning machine may appear paradoxical to some readers. How can the rules of operation of the machine change? They should describe completely how the machine will react whatever its history might be, whatever changes it might undergo. The rules are thus quite time-invariant. This is quite true. The explanation of the paradox is that the rules which get changed in the learning process are of a rather less pretentious kind, claiming only an ephemeral validity. The reader may draw a parallel with the Constitution of the United States.

— A.M. Turing, Computing Machinery and Intelligence, 49 Mind 433, 453 (1950)

The Constitution of the United States — so called because it constituted or created the United States — almost was not adopted. It wasn’t just because the South objected to Black people being considered people. That was agreed upon when the Constitution denied that they were. Article I, section 9 prevented Congress from banning the importation of enslaved people until at least 1808. Then there was the Fugitive Slave Clause of Article IV, section 2. And, of course, the famous 3/5ths clause that MAGA to this day argues is still a valid way to consider the humanity of Black people, if it must be considered at all.

On that front, not a lot has changed.

But thanks to these immoral compromises that have stained the soul of the United States since its inception, we allegedly have a United States of America.

There was another sticking point, though, which was much debated. You see, the Libtards of the day were concerned that, once constituted, the government of the United States would ignore the Constitution that constituted it.

They were not wrong.

This already long Substack post could easily be turned into a book if I got into all the details, so let’s focus on just what pertains to this whole “Anthropic won’t remove the siderails” complaint from Demented Dickwad Donnie and Beer Pong Pro Pete Hegseth: guardrails that recognize the Fourth Amendment and Due Process of Law.

The Fourth Amendment was a direct response to two instruments the British Crown used to harass colonists: general warrants in England and writs of assistance in the colonies.

General warrants authorized searches without naming the person, place, or thing to be searched — a blank check for the government to rifle through anyone’s papers and effects at will.

The 1765 case of Entick v. Carrington exposed their abuse when Crown agents ransacked a journalist’s home looking for seditious pamphlets; Lord Camden declared the warrants void, establishing that government power to intrude on private property must have explicit legal foundation.

Writs of assistance were the colonial version — standing search authorizations that didn’t expire, allowed customs officials to enter any premises to look for smuggled goods, and could be executed by anyone the writ-holder deputized. James Otis argued against them in 1761 in a Boston courtroom; John Adams later said the child independence was born in that argument.

The Founders wrote the Fourth Amendment to kill both instruments permanently: no warrant without probable cause, no warrant without particularity — naming the specific place, person, and thing.

The whole point was that “general” was the problem. The government had to know what it was looking for before it went looking.

The connection here almost writes itself. A general warrant didn’t name a person, place, or thing. It just said, “find whatever’s there.”

King George and his men did not have Anthropic’s technology. (Or OpenAI’s slightly inferior tech.) Today, that technology exists. But it’s blocked — arguably — by two things: Claude’s guardrails and the Fourth Amendment.

The five words of Anthropic’s contract preventing “analysis of bulk acquired data” are backed by the Fourth Amendment. Removing them is the Pentagon asking to get general warrants back again.

Just…digital this time. The Six Million Dollar AI.

We can rebuild him. We have the technology. We can make him better than he was. Better . . . stronger . . . faster.

— Wikipedia, The Six Million Dollar Man (Last edited February 26, 2026)

Mass surveillance of Americans isn't stopped by the 4th Amendment when the government decides the national security exception swallows the rule.

Or when Demented Dickwad Donnie reinterprets the Constitution such that Anthropic’s refusal to help instantiate the equivalent of “general warrants” thwarts fascistic ambitions against “we, the People”, the Constitutors — the Creators (I almost feel like I need a Dr. Frankenstein quote here) — of the United States.

"The Leftwing nut jobs at Anthropic have made a DISASTROUS MISTAKE trying to STRONG-ARM the Department of War, and force them to obey their Terms of Service instead of our Constitution," Trump wrote in a Truth Social post. "Therefore, I am directing EVERY Federal Agency in the United States Government to IMMEDIATELY CEASE all use of Anthropic's technology. We don't need it, we don't want it, and will not do business with them again!"

— Shannon Bond, Geoff Brumfiel, OpenAI announces Pentagon deal after Trump bans Anthropic (February 28, 2026)

The courthouse isn’t protected by a metal detector when the guy with the gasoline already has a badge.

This explainer digs into the analysis from attorneys Eric Schweitzer and Rick Horowitz, who argue that the crisis isn’t new at all — it’s a decades-long failure to follow a constitutional command that has been on the books since 1849. Before California was even a state, its constitution established a presumption of liberty: release before trial was supposed to be the default, and pretrial detention the rare exception.

What happened instead was a drift toward wealth-based detention — fixed bail schedules that have nothing to do with actual risk and everything to do with your bank account. A dangerous person with money goes home. A harmless person without it stays in jail. That’s not public safety. That’s economic sorting.

The landmark Humphrey decision didn’t invent new law. It just forced courts back to what the Constitution has always required. This isn’t a call for reform. It’s a call for compliance.

It was supposed to end wealth-based detention. Judges were required to consider ability to pay. Cash bail was to be eliminated as far as using it to keep poor people locked up pretrial. Pretrial freedom was supposed to be the rule — detention the exception.

That was the promise.

But the data tell a very different story.

In this explainer, I walk through what Humphrey actually required, what the UCLA/Berkeley “Coming Up Short” report found, and why pretrial detention has increased in many places since the decision.

If even the state’s highest court cannot shift the structure of pretrial justice, we need to ask a harder question:

Is this confusion — or resistance?

Watch the video for more.

Every plea form contains a paragraph like this one:

I am entering this plea freely and voluntarily. No one has threatened me, forced me, or promised me anything other than what is written here.

Judicial Council of Cal., Plea Form, with Explanations and Waiver of Rights—Felony (Criminal) (Judicial Council Form CR-101, rev. July 1, 2025), at 5

The language may vary slightly by jurisdiction, but the idea is always the same. It’s boilerplate. Everyone in the courtroom has seen it a thousand times. The judge reads it. The defendant answers “yes.” The transcript records consent. The system moves on.

One day, during a plea colloquy, I interrupted that rhythm.

We had reached that paragraph — the one where my client was required to state, under penalty of perjury, that no one had threatened him or coerced him into pleading. And I said, out loud, in open court, that this wasn’t really true.

Not in the way the form pretended it was true.

Because my client had been told — clearly, explicitly, and more than once — that if he did not accept the plea that day, the prosecutor would add charges and enhancements. His exposure would increase dramatically. The threat wasn’t subtle. It wasn’t hypothetical. It was the offer.

And now, on top of that, he was being required to say — again under penalty of perjury — that no one was threatening him.

I pointed out the contradiction.

The judge reacted immediately. He was angry. He warned me that he would tear up the plea agreement if I continued. The courtroom got tense. My client looked panicked — not because what I said was wrong, but because it was right in a way that endangered him.

I had just done something dangerous: I had disrupted the fiction that allows the system to proceed.

From the system’s point of view, nothing improper had happened. Prosecutors are allowed to file additional charges. They are allowed to threaten lawful action. Judges are allowed to insist on a clean record of voluntariness. Everyone was acting “by the rules.”

But from inside the moment, it felt different.

Part of the reason that difference is so easy to dismiss is that plea bargaining is almost always discussed after the fact — once an outcome exists and uncertainty has collapsed into a story about choice. As one scholar puts it:

It is easy to overlook the problem of uncertainty because most commentary views the plea-bargaining dynamic in hindsight… For defendants deciding whether to agree to a plea offer, however, uncertainty is key.

— Andrew Manuel Crespo, Plea Bargaining’s Uncertainty Problem, 58 Am. Crim. L. Rev. 571, 573 (2021)

That distinction matters, because what the record later calls a “decision” was made inside a fog of risk, incomplete information, and escalating exposure — not from a position of calm knowledge or equal footing.

It felt like being asked to execute a trade under pressure — with the price moving against you — and then being required to certify, on the record, that you acted freely, without coercion, and with full information. As if the ladder had not been climbing. As if the downside had not been made explicit. As if refusing the trade would not immediately worsen your position.

In markets, we have a word for that kind of situation. It’s not choice. It’s forced liquidity. Sometimes it looks like a margin call. Sometimes it looks like a short squeeze — where the position itself isn’t proven wrong, but time, pressure, and escalating exposure make holding it impossible. You don’t exit because the thesis failed. You exit because survival requires it.

The criminal legal system doesn’t use that language. It uses moral language instead. Voluntary. Knowing. Intelligent. Free.

And that’s where the discomfort begins.

When you step back from the ritual, what’s happening looks much closer to this:

Some uncertainty is, of course, unavoidable. But sometimes, there is so much uncertainty in evaluating a plea offer that the defendant’s decision resembles a high-stakes gamble, not an informed choice… In certain scenarios, defendants are forced to stake their lives on propositions no more predictable than the spin of a roulette wheel. Such a dynamic is toxic to any rational system of justice.

— Andrew Manuel Crespo, Plea Bargaining’s Uncertainty Problem, 58 Am. Crim. L. Rev. 571, 547 (2021)

That is not the language of misconduct. It is the language of structure.

And when someone points out that the tape tells a different story than the one the plea form records, the system doesn’t argue back. It tightens. It threatens collapse. It insists the fiction be completed — cleanly, quietly, and on the record — so that everyone can move on pretending the choice was freely made.

Finally, I give you Part IV of my four-part series which began with Part I at the start of January. (All of the other parts are linked at the end of this article. And I will go back and link each of the articles to the other articles.)

The Moment the Market Must Clear

The reason the interruption mattered — the reason the judge reacted the way he did — has nothing to do with me, or with my client, or even with that particular plea.

It had to do with clearing. With closing the candle. Printing the trade. Because until then, nothing is defined.

Up to that moment, the system was operating the way it always does during price discovery. Positions were open. Exposure was known but not yet realized. The parties had been living with uncertainty because uncertainty was still doing useful work. Offers could be floated without commitment. Threats could be implied without being triggered. Everyone could wait — or at least pretend they could.

That’s the phase the system tolerates.

The plea colloquy is different. It is not part of discovery. It is the clearing event. As I said, we print the closing price of the candlestick.

Once the court entertains the plea, the system no longer tests positions. Its work now is closing them. The open exposure that’s been driving behavior up to that point has to collapse into a final print — something legible, stable, and no longer subject to revision.

And that means uncertainty disappears.

Not factual uncertainty — no one expects truth to be fully known at that point; truth was never the point; the market merely needs to keep moving — so it’s procedural uncertainty at this point. The kind that allows the position to be reopened. The kind that would let anyone later say: this exit wasn’t voluntary; it was forced by pressure we all understood at the time. A short seller got squeezed.

That’s why the record suddenly becomes so brittle.

What the system needs at that moment is not an explanation of how the trade was executed, but a certification that it was clean. No coercion. No pressure. No external force. We stamp the ticket and move the trade off the books.

The narrative that follows — the defendant chose, freely, knowingly, voluntarily — is not descriptive. It’s settlement confirmation.

It replaces the messy mechanics of how the position became untenable with a single, tidy representation: consent. And like all such representations, it must be absolute. Anything less would leave residual risk in the system.

That’s why naming the pressure as I did is destabilizing. It keeps open the threat of a fakeout.

A fakeout occurs when a trader anticipates a transaction signal or price movement that does not materialize, resulting in a move in the opposite direction.

—Lucas Downey, Spotting Fakeouts: Key Strategies in Technical Analysis (December 28, 2025)

The judge reacted to what I did not (just) because it is improper — everyone involved understands exactly how the position was constrained — but because acknowledging it at the clearing moment, as I said, defeats the function of the ritual. It is the equivalent of announcing, during settlement, that the seller exited only because margin requirements were raised mid-day and liquidation became unavoidable.

That may be true. It may even be obvious. But it cannot be said if the system is to move on.

Finality only works if the exit looks voluntary.

For insiders, this isn’t controversial. Prosecutors understand that escalating exposure changes behavior. Judges know that time and detention accelerated the convergence. Defense lawyers get that refusing the offer is not a neutral act but a decision to absorb additional risk immediately.

None of that’s hidden.

What’s hidden is the fact that the system can’t acknowledge forced liquidity without undermining its own legitimacy. So the pressure has to vanish at the moment of execution.

The defendant must swear that what everyone understands to be true is, for purposes of the record, not true. Under penalty of perjury (that’s really what galls me the most — the system that is supposed to punish lies insists on a lie to close the deal). The judge must accept that answer without probing. The transcript must reflect a clean exit. The position must. Be. Closed.

And once it is closed, the conditions that produced it don’t matter anymore. Not because they were resolved, but because they have been priced in and erased.

Print the next candle, please!

This is why the reaction was immediate. I didn’t accuse anyone of misconduct. I didn’t challenge the legality of the offer. I didn’t ask the court to refuse the plea.

I introduced counter-evidence at the moment the system needed certainty.

That was enough.

Because the system doesn’t require the clearing story to be accurate. It requires it to be uncontested. (Under penalty of perjury.)

Once the fiction is completed — once the exit is certified as voluntary — the pressure that made holding the position impossible disappears from the record. It isn’t remedied. It’s not justified. It’s just written out of existence.

That erasure isn’t an accident. It’s how the system stabilizes itself after doing exactly what it was designed to do.

Who Holds the Risk After the Trade Prints

Once the position is closed, the system’s primary concern is no longer accuracy. It’s allocation.

Every market has to decide, after the candle prints, who bears the risk if the price turns out to be wrong. In functioning markets, that allocation is explicit. Losses fall where they fall. Margin rules are enforced prospectively, not retroactively. You don’t reopen settled trades because the pressure that caused the exit later looks unfair.

The criminal legal system works the same way — except it pretends it doesn’t.

Once the plea is entered and accepted, all remaining risk is pushed onto the defendant. Not just sentencing risk, but epistemic risk. If later evidence emerges. If a witness collapses. If a forensic claim turns out to be junk. If a legal theory that drove the plea is later rejected. None of that matters in the same way it would have mattered before the trade cleared.

The system has already printed the price.

To use an economic analogy, plea bargaining establishes a “going rate.” The anticipated sentence is the central concern in the negotiation. The problem, however, is that both innocent and guilty defendants are placed in the same pot and the goal is to achieve the appearance of justice, not the realization of it.

John L. Kane, Plea Bargaining and the Innocent, The Marshall Project (Dec. 26, 2014)

This is why voluntariness matters so much after the fact. It is not about whether the decision was free in any meaningful, lived sense. It is about whether the system can disclaim responsibility for the conditions under which the decision was made.

A “voluntary” plea reallocates all residual risk to the person who entered it. It converts structural pressure into personal choice. If the outcome is later revealed to be wrong, unjust, or deeply distorted by uncertainty, the answer is ready-made: he chose. He was advised. He knew the risks.

That answer only works if coercion is erased from the record.

In market terms, this is the equivalent of declaring that every forced liquidation was the trader’s independent strategy decision — and then using that declaration to bar any later claim that the exit was compelled by changing margin requirements, liquidity constraints, or asymmetric exposure.

The fiction isn’t just convenient. It is load-bearing.

Appeals doctrine depends on it. Habeas doctrine depends on it. Finality doctrine depends on it. The system cannot afford to reopen trades every time it becomes clear that the price printed under pressure was not an accurate reflection of underlying value.

So the law does what markets do when they want stability: it treats settlement as conclusive, and it treats the conditions that produced settlement as irrelevant once the position is closed.

Defendants are risk averse and prefer the certainty of a year in prison to a 50/50 or 90/10 chance of a longer term. For many defendants the rights afforded by rules of criminal procedure have little value at trial but considerable value in trade; they can sell their rights back to prosecutors by dealing for shorter sentences through a guilty plea.

— Frank H. Easterbrook, Plea Bargaining is a Shadow Market 51 Duq. L. Rev. 551, 552 (2013)

This is why voluntariness is assessed retroactively, through a transcript, rather than contemporaneously, through a serious examination of pressure. It is why the standard is not “free of coercive leverage,” but merely “free of unlawful threats.” It is why lawful escalation is invisible, while unlawful coercion is treated as exceptional.

The Court held that the threat of a significantly more severe penalty (even the death penalty) upon conviction is not so coercive as to invalidate a guilty plea. Indeed, few governmental actions short of physical coercion would render a guilty plea involuntary.

Jenia I. Turner, Plea Bargaining, in 3 Reforming Criminal Justice 73, [pin cite needed] (Erik Luna ed., 2017)

The distinction is not moral. It is structural.

The system does not deny that pressure exists. It denies that pressure matters once the trade is done.

And that denial is enforced by language. “Knowing.” “Intelligent.” “Voluntary.”

Those words aren’t descriptions of what happened. They’re loss-allocation devices. They determine, once the candle prints, who is allowed to complain — and who is not.

There were times during my seventeen-year tenure on the federal bench that inquiring of a defendant as to the voluntariness of his guilty plea felt like a Kabuki ritual. “Has anyone coerced you to plead guilty,” I would ask, and I felt like adding, “like thumbscrews [sic] or waterboarding? Anything less than that — a threatened tripling of your sentence should you go to trial, for example — doesn’t count.”

— Nancy Gertner, Letter to the Editor, ‘Why the Innocent Plead Guilty’: An Exchange, N.Y. Rev. Books (Jan. 8, 2015)

Once the plea clears — Time & Sales prints — the system moves on. The trade is treated as settled, the books balanced, and any error in price discovery — any distortion produced by pressure, timing, or fear — is no longer the system’s problem. It belongs entirely to the trader who exited.

After the Trade Clears

Once the plea is entered and accepted, the system treats the matter as finished. The record’s closed, the outcome fixed, and the conditions that produced the decision are no longer examined in the same way they were before the plea was taken. Whatever uncertainty, pressure, or distortion shaped the result is treated as having been resolved by the act of pleading itself.

Calling the plea “voluntary” does not describe how the decision was made. It determines how responsibility is assigned once the decision can no longer be revisited. From that point forward, the risk does not belong to the system that structured the choice under escalating exposure. It belongs to the person who accepted the outcome — after affirming, under penalty of perjury, that no pressure existed at all.

That is how the system clears the trade. All that’s really left is to output the chart.

Sentencing will be every more ritualistic than was the recording of the trade.

The gavel falls. The next auction is about to begin.

If you’ve seen this from inside the system — as a lawyer, a defendant, a judge, or a trader — I want to hear what you recognized.

Leave a Comment

Other Parts of This Series

Part I

Part II

Part III

The posts haven’t appeared back-to-back. I warned of this with the first in the series: I noted that my ADHD and other things intervening might mean interruptions. (I know myself well!) But the idea I’m working through here hasn’t gone anywhere, and it’s one I keep returning to because it keeps explaining things I see every day in court that don’t make much sense under the usual story we tell about how pleas happen.

And as I’ve recently become enamored of day trading — and I’m doing very well with it, thank you — I’ve been using market metaphors and analogies to tell the story.

In Part I, I tried to describe plea bargaining the way it actually looks from the inside: not as a conversation aimed at persuasion, but as a process more like price discovery. Outcomes don’t emerge because someone is convinced they’re fair. They emerge because pressure accumulates under unequal constraints. People test positions. They wait. They resist. They reassess. And eventually the system converges — often without anyone changing their mind about the underlying facts at all.

In Part II, I focused on time. Not time as duration, but time as cost. Delay doesn’t just stretch cases out; it redistributes pressure. Waiting is cheap for some participants and crushing for others. Pretrial detention, in particular, works — not always successfully — to accelerate everything. Yet it always turns uncertainty into punishment and usually makes resistance collapse long before truth has much of a chance to matter.

What I haven’t talked about yet — and what I want to slow down and look at here — is information. Who controls it. When it arrives. How incomplete knowledge interacts with fear and risk. And how people make decisions in a system where the cost of being wrong is unbounded, but the information needed to feel confident is always just out of reach.

This is a different kind of discovery. It’s a kind more criminal defense lawyers and others within the system recognize. It relates not to the “price discovery” previously discussed in Part I, but to the discovery of evidence and opinions about “facts” relating to the criminal case.

Because time doesn’t do its work alone. It does its work filled with unanswered questions, partial disclosures, and the stories people tell themselves while they wait. And it’s in that space — between what is known, what is suspected, and what is feared — that “choice” starts to narrow in ways that aren’t captured by the words we use later to describe it. (We’ll return to that linked article in Part IV.)

That’s the ground this part — Part III — covers.

Information Does Not Arrive All at Once

One of the things that gets lost when people talk about plea bargaining is how information actually enters a case.

Not how it’s supposed to enter, but how it does.

Most people imagine that a criminal case begins with facts already in hand. The police investigate, evidence is gathered, reports are written, and then everyone sits down to decide what to do with what’s known. In that story, plea bargaining happens after the important information has been collected.

Notwithstanding alleged “open files” systems, that isn’t how it really works.

Information in criminal cases more often dribbles in. Sometimes it arrives early, but incomplete. At the arraignment, the defense might be provided with the Complaint — which provides the information about the charges the prosecution thinks it can prove — and sometimes police reports.

Other discovery arrives late, when the cost of acting on it has already grown prohibitive. Sometimes it “technically” arrives without arriving in any meaningful sense at all. My favorite is when the prosecution tells the court that “discovery has been provided to the defense” but it has actually been provided to a third-party commercial entity with which, if the defense wishes to have the discovery, the defense is required to enter into an agreement.

Not in the least bit constitutional, but the majority of criminal defense lawyers never read the agreement they’re entering into, potentially compromising their own clients, without realizing it — or maybe without caring.

I’ve refused that agreement. Consequently, it sometimes requires months of fighting in court before my client gets to see some of the evidence against him. It comes after the court finally tires of the shenanigans and orders the prosecution to do what the law already requires the prosecution to do, but which — sometimes just because it is a legal requirement — the prosecution resists.

Police reports and police body cam footage and other videos are good examples.

Police reports come early. They also come framed. They reflect what officers thought mattered at the time, what they noticed, what they didn’t, and what they believed before they ever put pen to paper. They are not neutral summaries of reality. They are artifacts of a moment — and often a rushed one — sculpted to back the prosecution story.

Body cam footage and other videos — Ring cameras, storefront camera,s the city’s ubiquitous surveillance cameras that track every citizen’s movement because when we aren’t on lockdown that’s just the kind of nation in which we live — can take longer to obtain. Unsurprisingly, this is because they sometimes contradict the police reports.

That’s why such discovery, when it comes, is rarely clarifying all at once. It is can be voluminous, uneven, and difficult to interpret in isolation. A body-worn camera video might answer one question while raising three others. A witness statement that looked solid on paper and is now fragile in context. Even a lab report might say less than people think it does.

And some of the most important information — technically more a part of the “price discovery” discussed in Part I of this series — about how a witness will hold up, how a jury might react, how a judge will actually rule, does not exist yet at all.

Which means that decisions are made not under conditions of knowledge, but under conditions of uncertainty.

And that uncertainty is not shared evenly.

One side controls charging decisions, enhancements, and exposure. One side controls when certain information is disclosed and in what form. The other side waits, reacts, and tries to assess risk with pieces that never quite add up to a complete picture.

This is why I try to hire a defense investigator on almost every case.

This description is not — or not purely — a moral accusation. It’s a description of how the system functions. And it matters, because when the cost of being wrong is high enough, incomplete information doesn’t just complicate decision-making — it reshapes it.

Liquidity, or Why Some Information Can’t Be Used

To return to my metaphor, in markets information by itself isn’t enough. What matters is liquidity — whether you can act on what you know without getting crushed in the process.

That turns out to matter just as much in criminal cases.

A defense lawyer can know a great deal about a case and still be unable to use that information safely. Acting on it might require filing a motion that hardens the prosecution’s position. Sometimes you want to do this; sometimes you don’t. Doing it can essentially trap the prosecution in a story that turns out not as favorable to them as they’d thought. Other times, the hardening makes makes things more difficult.

And waiting to develop it further — giving space for everyone to breathe, feelings to dampen or fade — might mean additional months of custody. Prematurely testing a theory might cause an offer to worsen or disappear altogether.

So the question is rarely just “What do we know?” It’s “What can we afford to do with what we know?”

This is where the usual reassurance — you have discovery — starts to ring hollow. Information that cannot be acted on without unacceptable risk is not stabilizing. It’s destabilizing. It increases volatility rather than reducing it. In market terms, it’s jumping into the sale on the wrong candle.

I see this constantly. A piece of discovery suggests the police report is wrong. Another suggests a witness may not hold up. But using either one requires time, motion practice, or trial — all of which carry costs that are not evenly borne.

Again, in market terms the defense is often trading in a thin market with very little liquidity. Small moves produce outsized consequences. And when liquidity is thin, rational participants don’t probe very far. They exit early, not because they’re convinced they’re wrong, but because they can’t survive the downside if they’re right too late.

Seeing the Tape Without Seeing the Book

Another reason market metaphors keep fitting for me is that criminal cases often feel like trading without Level II.

You see the last print — the current offer, the amended charge, the new enhancement. But you don’t see the depth behind it. You don’t know whether there’s real resistance at that level or whether it will vanish the moment you lean on it.

From the defense side, offers can appear and disappear without any change in the underlying facts. A case that “can’t be resolved” suddenly can. A case that looked stable suddenly isn’t. The explanation, if one is given at all, is usually vague: new information, office policy, victim input, supervisor review. (Sometimes it’s just a deputy DA wanting to get something off their plate.)

What’s missing is visibility.

The prosecution, by contrast, usually has a much clearer sense of where the true lines are — how far it is willing to go, what it is prepared to file, and what outcomes it will not tolerate. That doesn’t mean the prosecution knows everything. It means it sees more of the book.

And that asymmetry matters, because people behave very differently when they know where the depth is and when they don’t.

Price Ladders and the Cost of Waiting

This is also where plea bargaining starts to resemble a price ladder.

Early offers are often better. Later offers are often worse. Not always — but often enough that the direction becomes clear. The message is implicit but unmistakable: delay is expensive.

From the outside, this is defended as efficiency. From the inside, it feels more like watching the price move away from you one rung at a time.

What makes this effective is not threat, exactly, but predictability. If each attempt to wait for more information is met with greater exposure, then information itself becomes costly to seek. At some point, continuing to wait stops feeling prudent and starts feeling reckless — even when waiting is the only way to learn what actually matters.

This is one of the reasons innocence doesn’t function the way people expect it to. Being right does not improve liquidity. It does not slow the ladder. It does not cap downside risk.

It just means the trade should work — eventually.

And “eventually” is often more time than the person on the other side can afford.

What Choice Looks Like From Inside the Fog

When people later describe plea bargaining, they talk as if choices were made in the light — with facts known, risks weighed, and options freely compared. That story is comforting. It makes outcomes feel earned, and responsibility feel clear.

But from inside the process, choice doesn’t always look like that.

It looks like deciding whether to act on information you don’t fully trust, or waiting for information that may arrive too late to use. It looks like guessing how much pressure is real and how much is performative. It looks like trying to distinguish risk from fear when both feel identical in the body. And it looks like making irreversible decisions in a system that steadily raises the price of waiting while promising clarity just one step further down the road.

In that environment, “choice” narrows. It’s not that anyone takes it away outright, it’s because the cost of “choosing” or not grows faster than the information needed to justify the “choice”. By the time enough is known to feel confident, the downside has often become unbearable. What looks voluntary from the outside feels forced, or constrained, from the inside.

As I pointed out yesterday, that is not a moral failure. It is a structural one.

Whenever we get to Part IV, I plan to turn to how the system manages that reality — how ignorance is normalized, how fear is framed as prudence, and how finality is treated as a virtue even when accuracy remains unresolved. Because once you see how information, time, and pressure interact, it becomes much harder to believe the simple story we tell ourselves afterward about how a plea was “chosen.”

Other Parts of This Series

Part I

Part II

Part IV

I’ve been writing a four-part series — currently interrupted, with only Part I and Part II published so far — using market metaphors and analogies to talk about criminal defense, and especially about plea bargaining.

I didn’t start out intending to do that. It happened because, back in November 2025, I began thinking seriously about retirement, started paying closer attention to my finances, and then — somewhat unexpectedly — got pulled into learning about markets and day trading.

And the metaphors won’t leave me alone.

What I’m seeing as I learn about markets, exchanges, volatility, and risk keep fitting. Not metaphorically in the loose, “this kind of reminds me of” sense, but in the way the deep structures that constrain the system actually look, feel, and work. They explained things I see every day in court better than the language we usually use to talk about them.

It’s frankly been kind of eye-opening. I knew these things. But I didn’t see them as starkly until the metaphors hit my brain like a rising VIX print.

I’ve tried to find others writing about criminal defense this way. Not about fraud or insider trading or financial crimes, but about the process itself — about pleas, pressure, time, risk, and decision-making — viewed through the lens of markets. I haven’t found much.

That absence surprised me at first. But then, I realized, I shouldn’t feel surprised.

Long before this series, I’d already been circling some of the same ideas from a different direction. In writing about pattern recognition — and the danger of mistaking coherence for truth — not long ago, I put it this way:

Certainty is anesthetic. It dulls the discomfort of doubt, the ache of ambiguity. You don’t have to wrestle with probabilities if someone else already has, or at least sounds like they have. The performance of confidence is often enough.

— Rick Horowitz, The Sound of Certainty: How We Mistake Confidence for Truth (November 10, 2025)

Markets taught me something that courtrooms had been trying to teach me for years: confidence and correctness are not just different things — they are often inversely related.

That recognition sharpened further when I wrote about a trading halt that felt unsettlingly familiar:

In markets, trading is halted when volatility gets too extreme — not because the system is broken, but because the system knows it can’t discover truth at that speed. The halts are triggered to give time for everyone to slow down, consider their positions, actually think about what they’re doing.

Criminal courts don’t have the safeguard of an automatic halt.

— Rick Horowitz, Halted: What a halted stock reminded me about momentum, exhaustion, and the discipline of restraint (January 8, 2026)

Those sentences could just as easily have been written about arraignments, bail hearings, or early plea negotiations conducted under pressure, with incomplete information, and very real consequences for getting it wrong.

Before I move on to Part III of the four-part series I promised, I want to pause here — not to justify the series, but to explain why this way of seeing the system keeps proving useful to me, why I think it matters, and perhaps why others might resist this way of seeing our criminal injustice system.

Not because law is a market.

But because, in important and uncomfortable ways, the criminal injustice system behaves like one.

These Are Not Trading Pieces

Let me be clear about something at the outset: this is not a trading series, and it’s not an argument that lawyers should “think like traders.”

I’m not importing market language into criminal defense because I admire markets, and I’m not doing it to dress up legal analysis with clever analogies. I’m doing it because market language is unusually honest about forces the criminal legal system often doesn’t recognize. And if it did recognize them, I fairly certain it would prefer not to.

See, markets don’t pretend to be moral. They don’t claim to be fair. They don’t tell stories about what’s deserved, virtue, or redemption. They talk openly about risk, pressure, liquidity, timing, leverage, asymmetry, and volatility. They assume unequal information as a starting condition, not a scandal. Incentives aren’t hidden; they’re the point.

The criminal legal system, by contrast, is saturated with moral language — justice, accountability, responsibility, rehabilitation — even as it operates through mechanisms that look nothing like moral reasoning in practice. Plea bargaining is not a search for truth; it is a price discovery process under constraint. Bail is not simply about safety; it is a liquidity problem imposed on human beings. Speed is not efficiency; it is pressure. And pressure changes outcomes.

Market metaphors don’t prettify this. They strip it bare.

They give names to things lawyers, judges, and clients all feel in court but are rarely encouraged to say out loud: when leverage shifts, when information is thin, when time is being used as a weapon, when movement is mistaken for progress, when restraint is punished and haste rewarded. None of that is captured by the language of “negotiation” or “resolution.” It is captured by the language of volatility, spreads, and forced trades.

So no — my articles about criminal defense which rely on market metaphors aren’t trading pieces.

It’s my attempt to describe a system that already behaves like a market, whether we acknowledge it or not, using a vocabulary that refuses to lie about how decisions are actually made.

Why This Way of Seeing the System Keeps Working for Me

The simplest reason this framing keeps returning is that it has actually helped me see deeper into a system I’ve worked within — including my time as a certified law student — for nearly a quarter century.

And it disciplines me.

Markets punish fantasy quickly. They don’t care how strongly you believe a story, how fluent your explanation sounds, or how morally satisfying your position feels. If you’re early, overconfident, under-informed, or trading on hope instead of structure, the feedback is immediate. And it’s personal.

Courtrooms offer no such clarity.

In criminal defense, bad assumptions can survive for months. Sometimes years. A weak theory can limp along on momentum. A premature plea posture can feel productive because something is happening. A confident prosecutor can sound persuasive long before the facts justify it. The system rarely corrects these errors quickly — and when it does, it’s often too late to undo the damage.

Market thinking resists that drift.

It forces me to ask questions that legal language tends to soften or postpone: What do we actually know right now? What don’t we know yet? Who controls the timing of information? Where is the leverage really coming from? What is the cost of waiting — and who is paying it? What happens if we act now and turn out to be wrong?

Those questions don’t sound especially legal. They sound operational. Structural. Cold, even. But they are exactly the questions that determine outcomes in real cases.

Markets also teach restraint — not as virtue, but as survival.

Most trades are bad trades. Most moments are noise. Most movement is not opportunity. Knowing when not to act is not passivity; it’s judgment — resistance to action bias, the human tendency to mistake movement for wisdom. Judgment is built by tolerating discomfort — the discomfort of uncertainty, of waiting, of not having an answer yet when everyone around you wants one.

That lesson translates cleanly to criminal defense.

I’ve watched lawyers talk plea before they have discovery. Before investigation. Before they know whether the apparent pressure is real or just loud. I’ve watched cases harden prematurely because movement was mistaken for progress, and silence was mistaken for neglect. I’ve watched clients pushed into decisions because delay felt unbearable — not because resolution was actually wise.

Market thinking resists that urge. It treats waiting as a position. It treats restraint as information. It understands that acting too early is often worse than acting too late.

Most importantly, this way of seeing the system keeps me honest about uncertainty.

Markets never pretend that outcomes are deserved. They don’t reward effort. They don’t care about intention. They respond to structure, incentives, timing, and information — nothing else. Criminal courts pretend to operate on moral evaluation, but in practice they respond to many of the same forces.

Seeing that clearly doesn’t make me cynical. It makes me careful.

And that’s why this framing keeps proving useful — not because it’s clever, and not because it flatters me and makes me look intelligent, but because it keeps dragging me back to reality when the system itself would rather tell a different, cleaner, more moral story.

Why I Think I Haven’t Found Others Doing This

Once I stopped being surprised that I hadn’t found much writing about criminal defense in market terms, I felt like the reason was obvious.

This way of seeing the system is uncomfortable.

It doesn’t flatter the people who work inside it — including me. It doesn’t allow us to pretend that outcomes flow naturally from truth, or that process is neutral, or that pressure is incidental. It insists on talking about incentives, asymmetries, and constraints first, and moral language second — if at all.

Most legal writing runs in the opposite direction.

Before I ever thought about markets, I wrote about the danger of tidy stories and the human need for coherence. Returning again to Sound of Certainty, I put it this way:

We all crave order. When the world tilts toward chaos — whether it be politically, technologically, or personally — certainty feels like a handrail. A calm, authoritative voice can do more to settle nerves than any mountain of evidence. We trust the people who sound sure of themselves.

— Rick Horowitz, The Sound of Certainty: How We Mistake Confidence for Truth (November 10, 2025)

That craving doesn’t disappear in courtrooms. In fact, if anything, it intensifies.

Judges want cases to move. Prosecutors want narratives that land cleanly. Defense lawyers feel pressure to do something — to talk, to negotiate, to show progress — even when waiting would be wiser. Clients and families want reassurance, timelines, and answers that simply don’t exist yet.

Market language resists all of that. It refuses to turn uncertainty into comfort. It doesn’t offer a handrail.

Instead, it says: you don’t know yet. And if you act as if you do, you may make things worse. The market is littered with losers in a zero sum game who thought if they just gathered enough indicators, or let Level II tell them what and when to trade (it doesn’t work that way), or spotted a bullish harami they’d be on the yellow brick road to retirement.

We like to believe our laws work the same way. Feed in the facts, turn the crank, and out comes justice. And we can’t be blamed for that, because, after all, judges and prosecutors look at it that way. They tell us that’s how it works. It’s all objective, analytical, fair.

But that’s not how things work in the real world.

— Rick Horowitz, The Bronze Brain and the Criminal Law: What a corroded Greek computer can teach us about the danger of believing our systems are smarter than we are (October 24, 2025)

That’s not a popular message in a profession built around advocacy, persuasion, and confidence. Lawyers are trained to argue positions, not to sit with indeterminacy. It’s a dynamic I’ve written about elsewhere in a different context — including large language models — where fluency is mistaken for understanding and confidence is mistaken for truth. Courts reward the same performance, with far higher stakes.

And there’s also something more fundamental at work.

Markets are openly amoral. That doesn’t mean immoral; it means they don’t pretend to justify outcomes in human terms. Losses aren’t deserved. Gains aren’t virtuous. They simply happen. The system doesn’t apologize. It doesn’t need to.

The criminal legal system, by contrast, depends on moral narration to function. It needs to believe — and to have others believe — that outcomes reflect responsibility, blame, and “just desserts”. To talk too openly about leverage, pressure, and constraint would threaten that self-image. It risks exposing how often results turn on timing, resources, and tolerance for uncertainty rather than on truth alone.

I highlighted that point when I wrote:

When we say that a case “takes too long,” or that someone is “dragging it out,” we are usually misdescribing what is happening. Delay is not a deviation from the system. It is one of the system’s primary mechanisms.

— Rick Horowitz, Time Does the Negotiating: How Delay Becomes Leverage in Plea Bargaining (January 9, 2026)

So this language gets resisted. Not because it’s inaccurate, but because it destabilizes the stories the system tells about itself.

I don’t think that resistance is usually conscious. I don’t think most lawyers wake up in the morning intending to preserve comforting illusions. But I do think there’s a strong professional incentive not to look too closely at how decisions are actually made when the cost of being wrong is high and the information is incomplete.

I came to realize that market metaphors cut through that by accident. Markets don’t care about our intentions. They just describe the mechanics.

And once you start describing the mechanics honestly, it becomes much harder to go back to language that pretends those mechanics don’t matter.

The Tape Just Prints

I’m aware that all of this may sound like an elaborate way of refusing comfort.

That’s intentional.

The criminal legal system runs on the promise that things will make sense soon — after the next hearing, after discovery, after negotiations, after the right explanation is given in the right tone. Market language keeps interrupting that promise. The candles don’t always print cleanly. They keep showing that uncertainty is not a temporary glitch but a structural condition. That pressure of the tape — Time & Sales — is not an aberration but a tool. That timing matters more than rhetoric, and that confidence is often just noise moving faster than understanding.

I wrote this article because I didn’t want to move on to Part III pretending I hadn’t learned that.

The series I started isn’t really about markets, and it isn’t really about plea bargaining in isolation. It’s about how decisions get made when the cost of being wrong is unbearable, the information is incomplete, and the system rewards motion over judgment. Market metaphors help me see that clearly because they refuse to tell me a comforting story about why an outcome happened. They only show me how. The tape just prints.

That way of seeing doesn’t make me detached. It makes me careful.

It slows me down when slowing down feels dangerous. It reminds me that waiting can be an act, that restraint can be a position, and that not every moment that demands action deserves it. In a system that confuses speed with justice and certainty with truth, that discipline matters.

After all, we do talk about “reading the tape” in markets. What I’ve been learning in thinking about all this is what every day trader either learns — or suffers from not having learned — the tape just prints. As “Larry Livingston” — a fictional character based on the real-life trader Jesse Livermore — puts it when talking about “the message of the tape”:

The fluctuations were from the first associated in my mind with upward or downward movements. Of course there is always a reason for fluctuation, but the tape does not concern itself with the why and wherefore. It doesn’t go into explanations. I didn’t ask the tape why when I was fourteen, and I don’t ask it to-day, at forty. The reason for what a certain stock does to-day may not be known for two or three days, or weeks, or months. But what the dickens does that matter? Your business with the tape is now—not to-morrow.

— Jon D. Markman, Reminiscences of a Stock Operator: With New Commentary and Insights on the Life and Times of Jesse Livermore 3 (John Wiley & Sons, Inc. 2010)

The tape just prints. It’s up to me to properly interpret how that should guide me.

I hope to complete Part III by tomorrow. If I succeed, we return to the series itself — to information, discovery, and how ignorance is managed, leveraged, and tolerated in criminal cases. This piece sits here because it had to. Because before going further, I needed to say why I’m looking at the system this way at all.

Not because it’s elegant or perfect or makes me look amazing (even if I’m not on a sammich).

But because it keeps proving true.

Interesting things happen in my ADHD brain when I pay attention to multiple things at once. Right now, for example, I’m reading Reminiscences of a Stock Operator — Edwin Lefèvre’s account of the life and times of Jesse Livermore — while watching what is happening in Minnesota, what has already happened to Congress and the courts as they are sidelined, and the growing illegitimate powers being exercised by one man.

Full disclosure: I normally try to write my posts so that they connect in some way with criminal defense, since I am a criminal defense lawyer. I’ve done this even when writing about AI, or using stock market metaphors. This one will be a little more general. It’s still about law — specifically the importance of the Rule of Law to our institutions and the form of government we’ve generally adhered to up to now.

Cornering Markets

In the late nineteenth and early twentieth centuries, men like Vanderbilt, Gould, Fisk, Drew, and others believed they had discovered a simple truth: if you could corner a market, you could control reality.

Control supply, starve liquidity, force everyone else to come to you — and the price would be whatever you said it was. But price wasn’t always the point. Often, it was vanity. They wanted the power to make other men bow to their will.

Sometimes it worked. Briefly.

Occasionally it worked longer than anyone expected.

But every attempt to corner a market extracted a cost. Counterparties defected. Credit thinned. Governments intervened. Even when the corner held, the system that made it possible did not survive intact. Markets didn’t always collapse — but they were distorted, hardened, and reshaped in ways that outlived the men who thought they were in control. In the process, it sometimes destroyed the fortunes of all parties involved. Some never recovered. They died of heart attacks. They died by their own hands. They died destitute and shamed.

Today, we are watching an attempt to corner something far more dangerous than railroads, wheat, corn, or copper.

We are watching an attempt to corner power itself.

Whether this follows the familiar pattern of past corners, or marks something genuinely new — unprecedented in more than two-and-a-half centuries of American history — remains an open question. What is not in doubt is this: power cannot be monopolized without breaking the system that once gave it legitimacy, flexibility, and reach.

The Logic of a Corner

A market corner is not simply an attempt to dominate supply. It is an attempt to eliminate any exit — to make it impossible for others to walk away without paying whatever price the cornerer demands.

The cornerer’s theory is straightforward: if every meaningful pathway runs through the cornerer, then price becomes irrelevant. Value is no longer discovered through ordinary market processes — through buyers and sellers negotiating freely — but is dictated. The market stops being a forum for exchange and becomes a site of compulsion. Participation is no longer voluntary. It is mandatory.

Leave a comment

But that theory depends on a fragile assumption — that everyone else must continue to play by the old rules even after the rules have been broken.

In practice, corners succeed only so long as three conditions hold. First, liquidity must remain constrained — meaning there must be no easy way for participants to move their money, their goods, or their commitments elsewhere. Second, counterparties must lack viable alternatives. Third, enforcement — whether legal, political, or financial — must remain credible. Once any of these fail, the corner begins to unravel, sometimes quietly, sometimes violently.

What makes corners dangerous is not that they always collapse. It is that even when they hold, they deform the system that sustains them. Credit and even cash become scarce or punitive. Participation narrows. Innovation slows. Trust evaporates. The system becomes brittle, dependent on constant pressure to maintain the illusion of control.

Political power operates under a similar logic. It is not owned; it is extended. It depends on compliance, belief, and habit. Courts function because their judgments are treated as binding. Legislatures function because their allocations are honored. Agencies function because rules are enforced consistently rather than selectively. None of this is self-executing.

To corner power, then, is to do more than seize authority. It is to sever alternatives. To remove the possibility of institutional exit — the ability of courts, states, agencies, or individuals to say no and still function. It is to make compliance the only viable option, not because it is legitimate, but because resistance has been rendered too costly or too dangerous.

At first, this can look like strength. Decisions happen quickly. Opposition quiets. Outcomes become predictable. But predictability achieved through compulsion is not stability. It is leverage masquerading as order.

The longer a power corner holds, the more resources it consumes merely to sustain itself. Enforcement expands. Exceptions multiply. Emergency measures become routine. What once required consent now requires pressure. What once relied on legitimacy now relies on fear, loyalty tests, or exhaustion.

This is the paradox at the heart of every corner: control increases even as capacity shrinks. The system continues to function, but at a higher cost and with less resilience. It can persist this way for years — sometimes for decades. But it no longer adapts. It no longer corrects. Problems are not resolved, only deferred.

And when a system stops resolving problems, the damage does not announce itself all at once. It accumulates — quietly, structurally, and often irreversibly.

Law as Liquidity

In functioning systems, law plays the role that liquidity plays in markets.

Liquidity is what allows participants to move — to enter, to exit, to adjust positions without catastrophic loss. It doesn’t eliminate risk. It makes risk survivable.

When I first started learning day trading, once or twice — that’s all it took to teach me — I bought stocks just because they were going up. Sometimes fast. And then I tried to sell when I thought it time to take my profit. To get out before the climb ended.

But I had bought into a stock with low liquidity. I got trapped. Fortunately, I had at least enough smarts to know that as a beginner, I should trade small. I didn’t like the losses, but they didn’t destroy my account.

In markets, liquidity lets you in or out more easily. Prices change without freezing the system. In institutions, law works similarly. It lets society function and power change hands without violence.

Courts provide this kind of liquidity. So do legislatures. So do agencies operating under stable rules. They create predictable pathways for disagreement, correction, and redress. They allow conflicts to be resolved rather than escalated. They let losing parties remain participants rather than becoming enemies.

Americans haven’t always liked this and some have turned quite cynical (and I might say “inappropriately cynical”) when they see it: Bush and Obama got along because our system worked to the benefit of all; not because “they’re all the same”. That’s why I’ve always resisted that argument.

But this particular “law liquidity system” only works if outcomes are honored.

A court order that is ignored does not merely fail to constrain the executive. It signals to everyone else that the exit ramp no longer functions.

This pattern is not new. As Ingo Müller documents in Hitler’s Justice (Amazon Affiliate link), authoritarian systems rarely abolish courts or legal procedure. Judges remain. Hearings continue. Opinions are issued. What changes is not the form of legality, but its function. Law no longer restrains power; it is used to rationalize it.

Müller’s point was not that courts vanished, but that their continued operation made the transformation harder to see — and easier to accept.

[A] dictatorship which clothes itself with a tinsel of legal form can so far depart from the morality of order, from the inner morality of law itself, that it ceases to be a legal system. When a system calling itself law is predicated upon a general disregard by judges of the terms of the laws they purport to enforce, when this system habitually cures its legal irregularities, even the grossest, by retroactive statutes, when it has only to resort to forays of terror in the streets, which no one dares challenge, in order to escape even those scant restraints imposed by the pretence of legality—when all these things have become true of a dictatorship, it is not hard for me, at least, to deny to it the name of law.

— Lon L. Fuller, Positivism and Fidelity to Law — A Reply to Professor Hart, 71 Harv. L. Rev. 630, 660–61 (1958)

Ernst Fraenkel (Amazon Affiliate link) described this kind of system as a “dual state” — one in which a normative legal order continues to exist on paper, while a separate prerogative system operates alongside it, unconstrained by law and justified by necessity.

Similarly, a legislative appropriation that is selectively withheld tells institutions and individuals alike that compliance is no longer reciprocal. The rules still exist on paper, but the system has lost its ability to legislate outcomes.

This is why, as a criminal defense lawyer, I have sometimes been critical even of the courts and judges. The Supreme Court, rather than adhering to precedent, says it’s okay for ICE to racially profile or for our current President to do what no other executive has ever done — or would ever have been allowed to do. At the local level, Superior Court judges hand out pretrial release orders like conservative candy, ignoring more than two centuries of law to the contrary, as I and my colleague have written in The Daily Journal. Even if I could, though — and, yes, I know that I can’t — I would not ignore the courts.

Because when legal liquidity dries up, power does not disappear. It pools.

Decisions become centralized not because the system fails to distribute authority, but because distributing authority is precisely what a successful corner on power eliminates. When legal and institutional exits are blocked, power pools by design. Discretion replaces rules. Exceptions replace standards. Outcomes depend less on process than on proximity to power — because once the corner holds, access replaces law as the organizing principle.

This can persist for a long time. Systems don’t collapse simply because law is ignored. They adapt — but in ways that favor those already positioned to absorb risk. Smaller actors lose access first. States, agencies, and individuals with fewer resources find that resistance carries disproportionate costs. Over time, compliance becomes less about legality than about survival.

What makes this phase especially dangerous is that it often feels orderly. There are fewer visible disputes. Fewer open conflicts. Decisions appear decisive. But this quiet is not stability. It is the absence of safe disagreement. (Not to mention of due process.)

In markets, the loss of liquidity does not announce itself with a bell. Instead, trading slows. Spreads — the distance between what a buyer is offering and a seller asks — widen. Volatility clusters. Then, one day, something that should have been manageable becomes catastrophic because there is no longer a way out.

Institutional systems behave the same way. When law ceases to function as liquidity, problems do not vanish. They stack. They compound.

When institutional failure becomes visible — when courts issue orders that are ignored, when legislatures allocate funds that are never released, when rules appear optional rather than binding — conflict does not remain abstract.

It surfaces. Often first as protest. Often peaceful at the outset. But protests are not the cause of institutional breakdown; they are a symptom of it. And when people no longer believe that lawful channels will respond, restraint becomes harder to sustain.

When the law breaks,
     the safeguards all fall,
And down comes the Republic,
     freedoms and all.

— Rick Horowitz, When the Law Breaks: The Rule of Law Will Fall & Down Comes Our Republic (April 6, 2025)

The Cost of a Successful Corner

The most dangerous misconception about corners — whether in markets or in power — is that they must succeed in order to matter.

They don’t.

All attempts at cornering have impact at least for a time. Some last long enough to reshape incentives, institutions, and expectations. By the time they loosen, the system they once operated within has already been transformed. The damage is not measured by collapse, but by what no longer functions the way it once did.

Markets taught us this long ago. A successful corner doesn’t need to bankrupt everyone involved to be destructive. It only needs to distort behavior long enough that trust erodes, risk concentrates, and ordinary participants are priced out of meaningful choice. Even when prices eventually normalize, the market that returns is not the market that existed before.

Power works the same way.

A political system does not need to fall apart to be broken. It only needs to stop resolving conflict fairly, predictably, and reciprocally. Once that happens, legitimacy becomes optional. Compliance becomes strategic. Survival replaces consent as the organizing principle.

What remains may still look like a functioning system. Courts still sit. Legislatures still meet. Elections may even still occur even if they no longer count at all — think “stock manipulation”, which, after all, is part of cornering. But the logic has changed. Outcomes matter more than process. Loyalty matters more than law. And power, once centralized, becomes easier to inherit than to challenge.

Markets eventually record this damage in prices, liquidity, and volatility. Political systems record it in distrust, polarization, and periodic eruptions of unrest. The bill always comes due, but not always to the person who first ran up the tab.

That is the lesson markets offer — and the warning they carry.

Power can be cornered for a time. It can even be held for a time. But it cannot be monopolized without breaking the system that once gave it meaning. And when that system breaks, it does not vanish. It persists — altered, hardened, and less capable of correcting itself than before.

Whether that is where we are headed, or whether we step back before the damage becomes permanent, remains unresolved.

What is already clear is that corners do not end cleanly.

They end by changing what survives them — in this case into something that no longer even remotely resembles what was originally constituted.

If you’re a regular reader, it won’t surprise you that I recently started enjoying day trading. In fact, I still owe two more parts to my four-part series on plea bargaining.

Today’s post is another of what I guess I’ll call “interjectory posts” because, as I mentioned in Part 1 of the four-part series, my ADHD and the flow of events means that I’m interrupting my four-part series at times by interjecting what’s caught my attention at the moment.

But we’re still kind of on what’s become a recent theme: criminal defense lessons learned from the market.

What has surprised me is how often I hear an echo between trading and criminal defense. Insights from the market bleed into what I already know about defending people in court; things I’ve learned over years of criminal defense practice turn out to be unexpectedly useful when I’m looking at charts.

This piece grows out of one of those echoes. It’s about what day trading reinforced for me — not about how to trade, but about what not to focus on — and why the same mistake shows up over and over again in criminal cases.

We Are Storytelling Animals

We are storytelling animals. We can’t help it. When something happens — a crime, a price move, a candle on a chart — we reach for a story that makes it make sense. Stories are how we organize experience. They compress uncertainty. They give us the feeling of understanding.

That instinct is usually harmless. Sometimes it’s useful. But in two places I spend a lot of time — criminal courtrooms and financial markets — storytelling becomes dangerous. Not because stories are always false, but because they move faster than evidence. They fill in gaps and smooth over missing steps. They make conclusions feel earned long before they’re justified.

This is why the “both sides of the story” idea is wrong.

In a criminal case, there usually aren’t two stories. There’s one story — the prosecution’s — and the problem is not that the defense needs a better one. The problem is that the story itself is doing work the evidence doesn’t support.

A good defense attorney doesn’t usually replace that story with another.

They do something more subversive.

They interrupt.

There Are Not Two Sides

“Both sides of the story” is one of those phrases people reach for when they want to sound reasonable.

It assumes symmetry. Balance. It assumes that a criminal case is a disagreement between two competing narratives, and that justice somehow emerges from weighing them against each other. That framing is comforting. It is also wrong in a way that matters. Because this is not “adversarial” — which is what makes our system work — but turns it into an ordinary debate.

In fact, I often laugh when listening to prosecutors argue in court. It always reminds me of a high school debate and every word is predictable. They just have to say it for the record as the judge nods along; the judge’s decision was already made. The prosecutor will usually win. But the debate format must be maintained so that it sounds like there were reasons for not following the law.

And this happens because, in a criminal case, there usually aren’t two stories, but judges, prosecutors, and even defense lawyers fall into the trap of believing there should be.

Yet, there is one story — the prosecution’s. They are making accusations. They do so using stories that may or may not be veridical.

Those stories arrive early and confidently. They have arcs. They explain behavior, assign motive, and give meaning to ambiguity before the evidence has done much work at all. Once that’s in place, they start doing what stories always do: fills gaps.

Too often it’s pure confabulation.

And just like what we see when LLMs confabulate, what gets called “evidence” in this context often isn’t evidence in the strict sense. It’s inference presented with confidence. It’s interpretation treated as observation. It’s narrative momentum doing work that proof hasn’t yet earned.

This is why the “both sides” idea misdescribes the defense function.

The defense is not there to tell the other side of the story. Sometimes a defense narrative is useful, even necessary, because jurors want sequence and coherence. But that is not the heart of the work, and it’s not what distinguishes good defense lawyering from bad.

The distinguishing move is interruption.

A good defense attorney slows the story down. They resist its pull toward inevitability. They keep pointing back — quietly, persistently — to what the evidence actually establishes and where it stops. They don’t rush to replace one explanation with another. They make the gaps visible.

That is often unsatisfying. Stories promise understanding, and interruption denies it. But understanding is not the legal standard. Proof is — at trial it is proof beyond a reasonable doubt. And proof does not improve just because a narrative feels complete.

What I said about machines in The Mirage of Reasoning Machines: The danger of trusting computer-generated charisma over proof is equally true about the prosecution’s story:

Truth is what courts are supposed to trade in. But “truth” becomes slippery when a machine can unspool an elegant paragraph that looks like logic and sounds like logic, yet collapses the moment you take it out of its comfort zone. That collapse—the gap between sounding right and being right—isn’t a quirk. It’s the core of the computational claim.

This is why there are not two sides. There is a story that wants to carry the case to a conclusion, and there is the work of refusing to let that happen without support.

That refusal is not properly viewed as obstruction. It is the job of a criminal defense lawyer.

What the Market Reinforced

Day trading has a way of punishing you quickly for paying attention to the wrong thing.

If you spend your time trying to explain why a stock moved instead of noticing how it moved, the market has a way of correcting you. Not philosophically. Financially. You can tell stories all day long. The chart doesn’t care.

A lot of trading books quietly encourage the wrong habit. They talk about what “buyers want” or what “sellers are trying to do.” They describe candles as expressions of intention. Fear. Confidence. Hesitation. Capitulation. Even the basic vocabulary — bullish, bearish — is psychological shorthand. We’re stuck with it now, but that doesn’t make it accurate.

The truth is simpler and less satisfying.

In the market, we don’t know what anyone wants.

We don’t know why a buyer bought or a seller sold. We don’t know what they were thinking, what they hoped would happen next, or whether they even meant to be there. All we actually know is what happened. Price moved. Volume printed. A level held or failed. A candle formed with a particular shape.

Everything else is projection. Made-up stories that add unnecessary — and misleading — cognitive load.

Markets are full of such ready-made narratives if you want them. Every candle can be given a backstory. And if you listen to enough commentary, you start to believe that understanding those stories is the same thing as understanding the market.

It isn’t.

What matters are patterns. Shapes. Repetition. What tends to happen after certain configurations appear, and what tends not to. The chart records outcomes, not intentions. Indicators summarize what already occurred. They don’t tell you what anyone wanted; they tell you what actually happened.

I don’t like upper wicks. Especially long ones. I don’t care what emotional state produced them. I don’t care whether a book tells me they reflect hesitation, profit-taking, or hidden strength. I know what tends to happen after I start seeing them. That’s enough.

Trying to interpret the psychology behind the candle slows you down. It opens the door to rationalization. Maybe this upper wick doesn’t mean what upper wicks usually mean. Maybe the story explains it away. Maybe you should give it one more bar to complete the tale, fill in one last missing gap.

That’s how bad trades linger.

The discipline trading forces on you is not insight. It’s restraint. You learn to stop letting explanation substitute for pattern recognition. You learn to accept that a configuration can be unfavorable even if you can tell a perfectly coherent story about why it shouldn’t be.

That lesson carries over cleanly.

In court, explanation creeps in the same way. The prosecution offers a story about what someone “must have wanted,” what they were “trying to do,” what their behavior supposedly reveals about their inner state. This is how they want you to (mis)judge ordinary non-criminal behavior to get to a conviction. Intent and motive fill gaps. Coherence replaces constraint.

Just as in the market, the danger isn’t that the story is implausible. It’s that it feels plausible enough to stop questioning. That’s why juries embrace the prosecutor’s stories.

We all crave order. When the world tilts toward chaos — whether it be politically, technologically, or personally — certainty feels like a handrail. A calm, authoritative voice can do more to settle nerves than any mountain of evidence.

— Rick Horowitz, The Sound of Certainty: How We Mistake Confidence for Truth (November 10, 2025)

Trading reinforced something I already knew but had started to take for granted: explanation is often just latency. It delays the moment when you’re supposed to notice that the pattern no longer fits. And by the time you’ve finished explaining, the opportunity to act — or to interrupt — has usually passed.

In both places, the work is the same. Stop the story. Look at what’s actually there. Notice where the pattern breaks.

And don’t talk yourself back into it.

We Interrupt This Story To Look at the Facts

I’ve spent most of my professional life interrupting stories.

Not replacing them. Not competing with them. Interrupting them before they harden into something that feels inevitable simply because it’s been told smoothly enough times.

“Let’s look at the facts,” I say. “Because the story you just heard is just that: a story.” The prosecution, we might say, read the candles, saw what was printed, and jumped into thinking he knew what the buyers and the sellers were up to.

Day trading didn’t teach me that. It reminded me of it. The market is just less patient than a courtroom. It doesn’t wait for you to finish your explanation. It doesn’t care how elegant your story is. It responds only to what actually happened — and to what tends to happen next.

Courts are slower. More ceremonial. More tolerant of confidence masquerading as proof. But the danger is the same. When explanation outruns evidence, when coherence substitutes for constraint, when a story does work the facts haven’t earned, the result isn’t understanding.

It’s error. A false conviction. An accused person’s liberty — and often family, home, job, sometimes life — is taken away from him.

That’s why interruption matters. Not as a rhetorical move, and not as obstruction, but as a refusal to let narrative charisma stand in for proof. In trading, that refusal keeps you from lingering in bad positions and losing money. In criminal defense, it keeps people from losing their liberty because a story sounded right.

There are not two sides.

There is a story — and there is the discipline of stopping it at the point where it stops being justified.

That discipline is the job.

I’ve been called a lot of things in my life. But this morning, after court, was a new one for me.

I was standing with a client’s family — a large one, the kind you mostly see in hallways and parking lots, faces you learn by proximity rather than names — when one of the men looked at me and said, “You’re just like the Lincoln Lawyer.”

I knew he meant it as a compliment. I also knew, immediately, that I didn’t know how to receive it.

I’ve seen The Lincoln Lawyer. I remember the outlines: a defense lawyer working out of his car, relentless, present, human. That was enough to understand what the man was trying to say. Court moved on. The day moved on.

That night, the same thing happened again.

Someone I’ve known for a while now sent a short message relaying a comment from her husband:

My husband just called you Mickey Haller. He really likes the fact that you are like the Lincoln Lawyer. 😂

Same reference. Same day.

I noticed the repetition. I didn’t know what to do with it yet.

This piece is about why a comment meant as praise can land uneasily — and what that discomfort says about the way criminal defense is practiced now.

When Praise Stops Making Sense

The discomfort wasn’t about modesty, and it wasn’t about self-doubt. It was about fit.

Praise assumes a certain alignment between effort and outcome, between judgment and reward. It assumes that the kind of work being praised still makes sense inside the system as it actually operates. That assumption no longer holds.

The version of criminal defense lawyering people recognize — the one they still reach for when they’re trying to describe presence, relentlessness, or care — depends on conditions that are increasingly rare. Time. Discretion. Negotiation. The ability to slow a case down long enough for facts to matter.

I often tell clients that part of my job is to drag the case. Judges don’t like it. Prosecutors don’t like it. Sometimes clients don’t like it either — which is why I explain this at some point. I don’t do it with every case; not every case needs that. Some cases need the high-speed train to barrel through.

But slowing a case down is often the only way to stop it from being decided before it’s understood. Because these days, the space to exercise judgment without being punished for it has narrowed. And when that space disappears, so does the kind of lawyering people still think they’re seeing.

Those conditions used to be easier. They were imperfect but still available. Now they’re treated as inefficiencies.

Having learned this, when praise arrives, it doesn’t land as affirmation. It lands as a reminder of a mismatch: between what people think they’re seeing and what the system actually permits. Between a cultural memory of law work as it once was and the reality of practicing it now.

That gap creates a particular kind of unease. Not because the praise is undeserved — though I’ll admit that’s usually how I feel — but because it points to something that no longer fits cleanly inside the machinery of modern criminal courts.

Once that misalignment becomes visible, it’s hard to ignore — especially for those still trying to do the work from inside the system. From conversations with colleagues, I know I’m not the only one seeing this.

What Changed

The law didn’t change in any fundamental way. The Constitution didn’t change. The formal rules governing criminal cases are largely the same ones I was working under years ago.

What changed was how cases move.

Over time, calendars tightened and dockets swelled. Speed became a virtue in its own right, especially for judges, prosecutors, and the money-grubbing type of defense “attorney”. Time stopped being something to use and started being something to measure efficiency. The system adapted to volume brought on by overcharging, over-legislating, and dehumanization. And in doing so, it learned how to function without stopping very often.

That shift had consequences. Practices that once felt ordinary — taking time to investigate, filing motions that slowed a case down, insisting on hearings that required real attention, making sure clients and their families understood things — began to feel disruptive. Judgment became harder to exercise without cost. Delay, even when purposeful, started to look like obstruction.

The values didn’t disappear. They’re still invoked regularly. But the system increasingly operates in ways that make honoring them difficult, and sometimes actively unwelcome.

I’ve made this same argument in more formal terms elsewhere. In a Daily Journal article published January 6, my friend and colleague and I wrote about pretrial detention and constitutional compliance — about how the law itself hasn’t meaningfully changed, but courtroom practice has. About how speed, scale, and administrative convenience have gradually replaced individualized judgment.

That piece was written for lawyers and judges. This one is written from inside the same problem.

That’s the environment in which praise now lands oddly. Not because the ideals are unclear, but because the machinery has learned how to move without waiting for them.

Mourning Has Broken

For a long time, I resisted the language of mourning.

Mourning suggests finality. It suggests acceptance. And acceptance is not something I’ve been willing to grant a system that continues to insist that it still stands for the same values it always has.

But at some point, the word stopped feeling optional.

When a system no longer allows the work it claims to honor, something breaks. Not all at once. It doesn’t collapse in one dramatic moment. Like the stocks I love to complain over, it’s been a slow downward churn. Things broke quietly, through repetition. Through small concessions that, to those who couldn’t spot the pattern, didn’t look like concessions at the time. Through practices that used to be ordinary and now feel oppositional.

What broke here isn’t commitment. Or effort. It’s the assumption that the work will be met halfway.

That’s the break of mourning, of mourning what’s broken.

Long before I became a lawyer, I wrote a poem built around a phrase I had stumbled across almost by accident: “the payment of a debt which cancels all others.” I shouldn’t have understood this, because I was only 16 years old. Hell, I’d never even had a real girlfriend yet when I wrote this. But there was a strange clarity for me in those words, revealing a pattern I’d already come to recognize.

As I just alluded to, that poem was written decades ago. I’ve written more about the poem here, but I want to quote it in this article because I think it has a new meaning for me that fits this context. It reads like this:

Testimony to the Payment of a Debt

As I sit gazing out the window,
The stars slip silently past.
And the edge of the moon,
Sticks out from behind the mountains,
On an ever-increasing downward plunge…
This is the beginning of an empty day.

Last night I sat here,
Thinking of all the times,
You and I once shared;
Wondering over and over,
What it was that came between us,
When all along I knew it was me.

Looking up now into the birth of the sky,
The mountains begin to glow,
And the sun peels off its blankets;
Melting the black coverlet of space,
Leaving me naked and alone,
To face a day without you.

“I want her back in my life!”
My mouth forms the words,
But the sounds are absorbed,
By a vast and uncaring world.
“I want her back!” I cry;
Yet no one hears, save me.

And now the unbidden rays of sunlight,
Invade the once Stygian darkness of the room,
Illuminating the medicine bottle on the table;
The vacuousness of which,
Is testimony to the payment…
Of a debt which cancels all others.

You see, at the time, I thought I was writing about personal loss. I didn’t yet have the language to understand what I was really circling: the moment when something is over, not because you chose to end it, but because continuing to pretend otherwise would be dishonest.

More recently, the language returned — differently. Shorter. Sharper. Less patient. A new poem popped into my head when two people in one day referred to me as being like the Lincoln Lawyer. This is the poem as it came to me, unfiltered:

Mourning Has Broken

Mourning has broken
like the first mourning.

Injustice has spoken…
Like the Last Judge.

Praise for the Lincoln,
Lincoln the Law-a-a-a-yer.

Time to retire,
Get my butt out.

These are not the same poem, and they’re not saying the same thing.

The first registers loss after the fact. While the second registers recognition in real time. I guess I’m learning chart-reading after all.

But the poems…together, they mark the distance between endurance and acknowledgment — between continuing to work and admitting what that work has become inside a system that no longer meets it halfway.

Mourning, in this sense, isn’t retreat. And it isn’t resignation. It’s simply the result of recognition.

And once that recognition sets in, it changes how everything else is heard — including praise.

Back to the Lincoln Lawyer

So when these folks suggested I was like the Lincoln Lawyer, I got it. (At least, I think I did! Tell me in the comments if I’m wrong!)

People are reaching for a figure who represents attentiveness in a system that often lacks it. Someone who doesn’t rush past people. Someone who stays present when things get uncomfortable. Someone who knows when not to move yet.

That image still resonates because it describes a kind of lawyering people wish the system still made room for.

But here’s the part that makes the praise land oddly: Mickey Haller works in a world where slowing things down is still legible as skill. Where calling for time isn’t treated as resistance. Where judgment can still interrupt momentum without being punished for it.

That’s not the world most criminal defense lawyers are practicing in now.

Today, restraint is often misread as weakness. Delay is treated as obstruction. And the pressure to move — always to move — overrides the quieter work of understanding what’s actually happening before irreversible decisions get made.

In markets, when things move too fast, trading is halted. Everyone understands why. Truth can’t be discovered at that speed.

Criminal courts don’t have that safeguard. So the burden falls on lawyers to call for time anyway — to ask the system to pause when it would rather barrel ahead. To stop participating in conveyor-belt justice.

That’s not cinematic. It doesn’t look heroic. And it rarely gets recognized for what it is.

So when the Lincoln Lawyer shows up in conversation now, I hear something else underneath the compliment. I hear a recognition that the kind of lawyering people still admire is becoming harder to practice — not because lawyers forgot how to do it, but because the system learned how to move without allowing for it.

Which brings me back to where this started.

I don’t fully get the comparison. I don’t think I fully understand how it applies to me — or maybe I just missed what those who said it intend. But to the extent I can accept any part of it, it’s probably because, at least for now, I still feel a kind of obligation to be a problem.

Not in the sense of being oppositional for its own sake, and not because I think the past was some golden age we can simply return to. But because so much of what now passes for efficiency in criminal courts depends on things moving too quickly to be questioned, and because slowing that movement down has become one of the few remaining ways to insist that evaluation, consideration — real judgment, not judgmentalism — still matters.

That’s what I mean when I talk about dragging a case. It’s not a piece of nostalgia. It isn’t theater. What it is is an attempt to keep cases from being decided before they’re actually understood, even when the system would very much prefer not to wait: “Who stopped the conveyor belt?”

Doing that now often feels less like practicing law and more like throwing a wrench into machinery that has learned how to run without stopping.

So when the Lincoln Lawyer showed up yesterday — first from a client, then from a friend — I heard something underneath what I took as a compliment. I hear a recognition of a kind of lawyering people still want to believe exists, even as the conditions that once made it ordinary continue to disappear.

These vanishing conditions don’t mean the work is over — like I argued in the Daily Journal, it’s a reason defense attorneys need to revive some old tools. (There, I spoke of the need for more habeas litigation.)

What it does mean is that praise, by itself, no longer tells the whole story. In this context, it’s less an affirmation than a signal — an echo of values we still know how to name, even as the system grows increasingly impatient with the people who insist on embracing them anyway.

This is Part II of what I expect to be a four-part series. Part I was Price Discovery Under Unequal Constraints: What Plea Bargaining Has in Common with Markets.

In Part I, I argued that plea bargaining is better understood as price discovery than negotiation — a process that draws to a conclusion under pressure rather than persuasion. That way of understanding it explains why outcomes change even when facts do not, and why agreement is finally reached without anyone being convinced.

What that account only hints at, however, is time.

Time is not a neutral backdrop in criminal cases. It is not an empty container in which decisions happen. Time is a cost. And like all costs, it is not evenly borne.

Every continuance, every reset date, every week of waiting reshapes the environment in which the decision is finally made. It’s not because information is uncovered or added to the discussion, although that may happen. Nor is it by clarifying truth — whatever that might mean. It happens by redistributing pressure.

Some participants experience time as strategy. Others experience it as punishment. Or at least an increase in apprehension (or even fear).

That difference is not accidental. It is structural.

When we say that a case “takes too long,” or that someone is “dragging it out,” we are usually misdescribing what is happening. Delay is not a deviation from the system. It is one of the system’s primary mechanisms.

Time does much of the negotiating long before anyone speaks.

In what follows, I want to be precise about how time functions — how calendars, detention, and institutional patience work to quietly move outcomes, even when no one intends them to do so.

Time as Cost, Not Duration

Time in criminal cases is usually thought of as if it were neutral — days passing, calendars filling, continuances stacking up. But time does not merely pass. It accrues.

For the system, time is administrative. For institutions, it is measured in docket management, staffing, and throughput. For individuals, it is experienced as cost.

That cost is not abstract.

It shows up as lost work, missed rent, suspended licenses, frayed family ties, untreated medical needs, and the slow fade of emotional reserves. Even for someone out of custody, delay has consequences. For someone detained, time adds to those consequences daily.

Like compounded interest — only in reverse.

But what matters is not simply how much time passes, but who pays for it.

Prosecutors do not experience delay the way defendants do. Judges do not experience it the way detained people do. Even we defense lawyers — professionals with calendars of our own — experience time differently from the people we represent.

This is why delay changes outcomes even when nothing else changes.

No new evidence needs to emerge. No argument needs to improve. No one needs to be persuaded. The decision milieu (if you will) shifts because the price of waiting rises for some parties and barely moves for others.

That asymmetry is not incidental. It is built into the structure.

When time is treated as duration, delay looks inefficient. When time is treated as cost, delay becomes leverage.

And once leverage enters the picture, how “agreement” arises is no longer mysterious.

The Fiction of Voluntariness

Courts insist that guilty pleas be voluntary. The language is familiar. It appears in plea forms and colloquies across jurisdictions: the plea is entered “freely and voluntarily,” without threats, without coercion, without fear.

Everyone in the room knows this is not true.

The threat is not subtle. It is right out there: accept this disposition, or face exponentially greater punishment if you do not. Probation versus prison. A determinate (specific number of months or years) term versus a life tail (such as 25-years-to-life). A fixed risk versus an unbounded one.

I am not making this up. I once defended someone in a case where the offer was “straight probation without any jail or prison time”. The victim of the threat — my client — would have to register as a sex offender. Turn down the offer? The prosecutor promised to amend the charges so that he would get life in prison if the case was lost at trial.

Surprising me and my investigator, the client turned down the offer.

The bigger surprise? On the day of trial, the case was dismissed.

Prosecutors will deny that this sort of “offer” is coercion. They will say, correctly in a narrow sense, that no one is forced to plead. The offer can be rejected. Trial remains available. Rights remain formally intact.

After all, the client I just told you about did it.

But let’s be frank: it takes a lot of cojones or outright crazy to reject that offer when you have two kids willing to tell a jury what the kids here had supposedly told the police. (It was a lie, by the way, and we knew it was. No doubt my client’s lack of experience with and misplaced faith in the system made him think the truth would come out.)

The prosecutorial response — shared by their brethren in robes — confuses the absence of physical compulsion with the absence of pressure. Just because no one literally beats up the accused to force him to plead doesn’t mean he’s not been — for all practical intents and purposes — forced to plead.

What makes the pressure effective is not menace or bad faith. It is asymmetry. One side controls the charging instruments, the enhancements, the sequencing, and the timing. The other side absorbs the risk.

This is where time reenters the picture.

The longer a case stretches on, the more expensive resistance becomes. Not just financially, but psychologically and existentially. The plea offer does not merely resolve uncertainty; it caps it. It places a “lid” on risk at the precise moment when waiting has made a lid feel indispensable.

That is why plea forms must insist on the pretense of voluntariness. The system requires the denial. The fiction is not incidental; it is functional. It lets everyone pretend that the choices really exist where only coerced pleas do.

If coercion were openly acknowledged, the structure would be harder to defend. So it is reframed as choice. A choice made freely. A choice made calmly. A choice made without fear — even when fear is doing almost all of the work.

Judges do not need to threaten anyone for this to function. The machinery is already in motion by the time the plea is taken. The role of the court is not to create the pressure, but to certify its disappearance.

Once that certification occurs, the pressure becomes invisible. It is retroactively erased.

Time did the negotiating. Like candlesticks on a market chart, the plea simply records the result.

Detention as Acceleration

If time is leverage in criminal cases, pretrial detention is leverage turned up to maximum volume.

Detention does not merely restrain a person’s movement, their freedom, their liberty. It radically alters how time is experienced. What might otherwise be tolerable delay becomes something closer to continuous harm.

During the pendency of a case, a day on the outside is a unit of inconvenience. A day inside is a unit of punishment.

Courts insist that pretrial detention is not punishment. It is administrative. Preventive. Regulatory. And, after all, it’s necessary for the safety of the public that people who have been convicted of no crimes — whose guilt has not yet been adjudicated — be locked up.

Even if, once they agree to a plea, they’ll be released on probation.

But the distinctions between punishment and administrative, preventive, regulatory crap collapse at the level that matters — lived experience. A person in custody is already paying part of the sentence before guilt has been established.

And if he ends up acquitted, or the case dismissed? The lost time — the lost job, months or years away from family, even lost homes because the primary breadwinner was locked up — is just a case of “too bad…so sad”.

This is not incidental. It is the accelerant.

Detention compresses decision-making timelines by making waiting unbearable. Every additional day magnifies loss. The question stops being “Can I win?” and becomes “How much more of this can I take?”

That shift has nothing to do with truth-seeking.

Evidence does not improve because someone is locked up. Legal arguments do not sharpen. Witnesses do not become more reliable. What improves is the prosecution’s bargaining position.

Time outside custody exerts pressure gradually. Time inside custody exerts it immediately and relentlessly.

This is why detention changes plea outcomes even in weak cases. Even when the accused person and the accuser know the accused is actually innocent.

A person who might rationally contest an accusation at liberty may rationally surrender under confinement. Not because innocence has become less likely, but because the cost of maintaining it has become intolerable.

This is also why plea offers so often include immediate release. The offer does not just resolve risk; it ends suffering. That relief carries enormous weight — weight that has nothing to do with culpability.

Prosecutors understand this. Judges understand it. Defense lawyers understand it. No one needs to say it out loud.

A plea that promises “time served” is not merely a sentencing recommendation. It is an escape hatch. A costly “get out of jail” card in the governmentally-administered monopoly game. (I deliberately did not capitalize that — I’m talking about a real monopoly here.)

And once detention enters the equation, voluntariness becomes even harder to take seriously. When the alternative to pleading guilty is continued confinement with no fixed end date, the idea of free choice is a fiction.

Detention does not force a plea by itself. It does more subtle and effective. It makes delay so costly resistance collapses.

Time accelerates. In the confined person’s mind, risk balloons. Convergence becomes predictable.

The market clears faster — not because the price is right, but because the buyer is running out of oxygen.

Leave a Comment, Please

Where This Leaves Us

None of this requires bad actors.

Prosecutors do not need to be cruel, though some are. Judges do not need to be indifferent, though many are. Defense lawyers do not need to be complicit, though too many are. Even in the absence of that, the system works exactly as it would when rational people respond rationally to unequal constraints.

Time makes resistance expensive. Detention makes it unbearable. Voluntariness supplies the necessary legal fiction to certify the outcome.

By the time a plea is entered, the relevant work has already been done. Not by persuasion. Not by truth-finding. But by pressure accumulating unevenly over time.

That pressure does not disappear when the plea is taken. It is merely declared irrelevant. The record reflects a choice. The machinery that produced it fades from view.

This is why plea outcomes can feel inevitable even when facts are contested, defenses are viable, and guilt is genuinely uncertain. The system does not need certainty. It needs convergence.

In Part III, I will turn to the next mechanism that drives that convergence: information — who controls it, when it appears, and how fear and risk distort what “choice” can mean under conditions like these.

For now, it is enough to see this much clearly.

Time did the negotiating. Detention accelerated it. The plea only recorded the result.

Other Parts of This Series

Part I

Part III

Part IV

I mentioned yesterday when I started a new series that I might interject other posts in between posts from the series. Tomorrow, I plan to come out with Part II. But something happened today….

Today I watched a stock do something I had never seen before.

It exploded upward, halted. Exploded again, halted again. And then, one more burst up with another halt.

Then fell hard — and halted a couple times on the way down.

For those who don’t know, a “halt” — also called a “circuit breaker” for obvious reasons — is a temporary pause in trading ordered by an exchange when a stock is moving too quickly or material information is being absorbed by the market. During a halt, no trades can be executed. The purpose is not to stop price movement permanently, but to slow it long enough for price discovery to resume on more equal informational footing.

Price discovery under stress.

All this surprised me a bit because I’d never seen a stock do that before. I was talking about it with ChatGPT. Chat always likes to warn me not to do something I wasn’t going to do anyway. I responded, “This is too alien to me. I’m going to sit and watch for a while.”

Now, I had already bought in before I understood what was going on — and before the halts started — and had almost immediately been “stopped out”. Again, for those who don’t know, when I bought my shares, I had put a “trailing stop” order behind the buy. The purpose of that was that if the stock suddenly dropped on me, my sell order would be triggered automatically, saving me from losing much more money during a too-quick dive. Unfortunately, my stop was a little too tight and I almost immediately got stopped out because the stock was far more volatile than I was expecting. (There was excellent news — a major new contract — and I suspect that once the news is digested, the stock may still move higher.)

A few minutes later, I decided to test if I could ride the waves and bought a much smaller position just to practice surfing. I ended up selling very shortly after because it looked to me the waves were going to come down on my head.

And then the stock absolutely went nuts. Waves? Those had been nothing. I was watching a full-on storm raging behind a tsunami of buy and sell orders.

Then the halts started. I was fascinated. I couldn’t stop watching. My heart was racing. And here’s what fascinated me the most: It was how familiar the feeling was.

That’s when I decided it was “too alien” and I wasn’t going back in.

Because this is exactly what criminal cases look like when something unexpected happens — a bad fact, a sudden filing, a media headline, a new charge.

Everything accelerates.

Everyone reacts.

No one actually knows what the case is worth yet — but too many people act as if they do.

Clients panic.

Prosecutors posture.

Judges hurry.

And the worst decisions almost always happen in those moments.

In markets, trading is halted when volatility gets too extreme — not because the system is broken, but because the system knows it can’t discover truth at that speed. The halts are triggered to give time for everyone to slow down, consider their positions, actually think about what they’re doing.

Criminal courts don’t have the safeguard of an automatic halt.

I’ve seen cases implode the same way — a single hearing, a single headline, a single bad minute — and suddenly everyone is trading on fear. Offers harden — or get worse. Positions lock. Clients want answers that don’t exist yet. The law hasn’t changed, but the temperature has. And the most dangerous moment is when movement gets mistaken for progress.

Sometimes, the best thing to do is nothing.

Instead, we call the chaos “strategy,” the overreaction “leverage,” and the pressure “efficiency.”

Why did seeing what was happening with the stock suddenly become familiar? Because in the courtroom when this happens, I virtually always call a “halt”.

“Your Honor, my client and I need time to discuss this new development. I request a continuance”.

What saved me today wasn’t being smarter than the market.

It was knowing when not to act — when to let the violence pass and wait for structure to return.

That’s the same lesson I try to teach clients.

You don’t usually win a case during the spike.

In markets, you can sometimes sell into that violence and walk away green. In court, you almost never can.

You survive it. You call a halt. You recalibrate.

The truth — like price — only emerges after the noise burns itself out.

About the Chart

The chart at the top of this post is the real one from this morning.

Those bubbles aren’t annotations added later for effect. They’re my actual buys and sells — green where I entered, red where I exited — placed by the software I use to trade so I can see where I am in the grand scheme of things as the price marches on.

In the chart at the top of this post, you can see where I got stopped out early and went briefly into the red. You can see where I slowed down — waited while I absorbed the meaning of this alien experience — re-entered with more restraint, and worked my way back into the green. And you can see the moment I was taken out by a final upward twitch — a red doji’s last reach — at a price the stock never touched again. (And still hasn’t as I write this: I’ve continued watching.)

That doji blew me away — literally and figuratively — because it was so unusual. It was almost a shooting star; not quite a gravestone doji. The wick of that candle that took me out wasn’t luck: it was the buyers’ final reach upward, rejected in real time. But not before handing me a parting gift of $1,129.50.

I converted the chart into a Baroque-style image at the top of this post deliberately. Baroque art lives in moments of excess: motion, pressure, violence, light breaking through disorder. That was the emotional texture of the trade. The drama was real. The temptation to overreact was real. And the most important decision wasn’t where I got in or out — it was when I chose not to act.

I’m not posting this chart to brag, and I don’t tell war stories to impress (well, maybe I do sometimes). Here I post them to make a point: that restraint is a skill, not an absence of action.

Whether it’s a stock in freefall or a case spiraling out of control, the discipline is the same: recognize when the system has temporarily lost the ability to discover truth, call a halt, and wait until structure returns. That’s not hesitation. That’s judgment.

Plea bargaining is usually seen as negotiation — a back-and-forth conversation in which two sides gradually move toward a fair resolution. That description is comforting. It’s also slightly misleading.

Over the last couple of months, as I’ve paid closer attention to how markets work while thinking about retirement and long-term planning, I’m amazed by how familiar what I see happening in the markets feels. Markets don’t resolve by persuasion. They resolve through testing, waiting, pressure, and constraint.

Once I realized that, I immediately thought, “this is just like plea bargaining”. I saw the same structure at work. It made sense right away because, after all, the market is a multi-person negotiation.

And so, as I’ll show, is plea bargaining. Both are games, in a sense, that involve two parties — a buyer and a seller — whose goals and choices are not independent either of one another or of other parties and constraints.

And, in both, you learn the game by playing it.

This is the first part of a short series. The subject is deeper than can be handled cleanly in a single essay, and the pieces may not appear consecutively. My ADHD and the amount of time it’s going to take to write all this — it took me hours just to draw up an outline — makes it a virtual certainty that other articles will appear between the four I have planned for this series.

Each of the four parts will stand on its own, but together they will explore plea bargaining as a process rather than a moral dialogue. When the series is complete, the pieces will be cross-linked, listing all four articles at the bottom of each article.

Plea Bargaining and Price Discovery

In markets, price discovery describes the process by which an outcome emerges over time rather than being agreed upon in advance. No one decides what the “right” price is. Participants test levels, resist them, retreat, try again, and respond to pressure as constraints tighten or loosen.

That process doesn’t require agreement, persuasion, or even clarity. It just requires interaction in an environment of scarcity.

Plea bargaining works the same way.

Offers are made, rejected, adjusted, and revisited. Positions harden, soften, or simply wait. Information arrives unevenly. Time passes. Pressure accumulates. Eventually, an outcome appears — not because anyone was convinced, but because the system converged on something that could no longer be resisted.

That is price discovery.

Calling this “negotiation” suggests a symmetry that usually does not exist. It implies that both sides are equally free to wait, equally informed, and equally insulated from risk. In reality, constraints are uneven from the start, and they rarely remain static.

Understanding plea bargaining as price discovery under unequal constraints is not cynical. It is descriptive. And it explains a great deal of behavior that otherwise gets misread as obstinacy, delay, or bad faith.

Unequal Constraints

Calling plea bargaining “negotiation” suggests a symmetry that rarely exists. It implies that both sides are equally free to wait, equally informed, and equally insulated from risk.

That is almost never the case.

From the start, participants in a criminal case operate under very different constraints. Some people can wait. Others cannot. Some can absorb uncertainty. Others experience it as punishment. Some control information. Others react to it as it arrives, often too late to matter.

Those asymmetries are not incidental. They shape outcomes.

In markets, price discovery is driven less by who wants what than by who can afford to wait. Plea bargaining is no different. Outcomes shift not because someone becomes persuaded, but because pressure tightens unevenly.

Want an even greater irony? This asymmetry favors the prosecution in virtually every case and therefore leads many truly innocent people to plead guilty.

There is also reason to believe that plea bargaining exacerbates the problem of innocent defendants pleading guilty. The power that plea bargaining gives to prosecutors allows them to offer plea deals that no rational defendant would turn down. A rational defendant would accept any plea bargain in which her expected punishment is less than the punishment she will receive after trial. So, for example, an innocent defendant who is facing a sentence of five years if convicted at trial should accept a plea offer of less than six months if she believes there is a 10% chance she will be convicted at trial. And an innocent defendant who is facing any jail time at all if convicted should rationally accept an offer of diversion or probation. That plea bargaining leads to innocent people pleading guilty is not merely an abstract concept. There is ample evidence suggesting that a nontrivial number of innocent people plead guilty.

— Carissa Byrne Hessick, The Real Problem with Plea Bargaining, 101 Tex. L. Rev. Online 123, 128 (2023)

It Takes Many to Tango

Plea bargaining is often described as a conversation between two parties. In reality, it is a multi-party interaction.

Prosecutors, defendants, defense lawyers, judges, victims, families, witnesses, law enforcement agencies, and institutions all participate in shaping the terrain. Each brings different incentives, tolerances, and timelines. Each adds friction or pressure in different ways.

To give just one example, California law requires prosecutors to consider — though around here, most do so simply by saying, “I considered and it’s a no” — immigration consequences when engaging in plea negotiations. In Padilla v. Kentucky, the U.S. Supreme Court said defense attorneys must not only advise clients about the potential immigration consequences of a plea, but might even engage in creative bargaining to help them avoid them. This functions as one of the many constraints that can impact plea bargaining, though this one usually is of little consequence.

More importantly,

Although not explicitly stated in Padilla, such creative bargaining involves not just the defense attorney and his client but also the prosecutor and the judge who agree to allow the plea to proceed. The Court was and is certainly aware of this. Indeed, in companion cases that followed Padilla—Missouri v. Frye and Lafler v. Cooper—the Court noted the many benefits that plea bargains afford the parties including the resources saved by the prosecutor and courts. Justice Stevens, then, was not speaking only to defense attorneys in calling for creative bargains; he was also encouraging prosecutors and judges to accept these pleas.

— Thea Johnson, Fictional Pleas, 94 IND. L.J. 855, 868 (2019)

While creative fictional pleas require the cooperation of various parties, judges, in particular, function as boundary conditions in other ways. Their patience, calendars, and tolerance for delay define how long testing can continue before resolution is forced. Victims and families amplify pressure in some cases and dampen it in others. Institutions prioritize efficiency, risk management, and throughput, often invisibly.

No one really controls the dance. But everyone contributes to the rhythm.

That is why outcomes can change without anyone “changing their mind.”

Testing, Resistance, and Consolidation

Our world, indeed our universe, operations on a system of cycles. The earth, along with her sister planets, orbits around the sun. A complete orbit completes a cycle, which we refer to as a “year".”

Predictable weather patterns create four seasons within that year, each with its own cycle. Tides flow in and out on exact cycles. Humans and all living creatures experience cycles of life, including birth, childhood, puberty, adulthood, and passing on.

— Toni Turner, A Beginner’s Guide to Short-Term Trading 61 (2d ed. 2008) (Amazon Affiliate link)

In markets, movement rarely happens in a straight line. Periods of advance are often followed by consolidation — pauses where prices compress, resistance is tested, and participants reassess. Candlestick traders have names for these pauses. The names matter less than the function.

Consolidation is how systems discover what will hold. Stocks “trade sideways” in a kind of “holding pattern” signaling “indecision” while various market participants catch their breath and reassess their positions.

What is often labeled “delay” in plea bargaining serves the same function. Early positions are tested. Resistance is encountered. Information trickles in. Emotional intensity cools or flares. Constraints reveal themselves over time.

This is not stalling for its own sake. It is consolidation.

Seen this way, accusations that one side or the other is “dragging the case” often miss what is actually happening. The system is determining which positions are sustainable and which are not.

A Note on Pretrial Detention

Time does much of the negotiating in plea bargaining. Pretrial detention dramatically alters how that time is experienced.

Detention tightens constraints on the accused in ways that are difficult to overstate. It accelerates pressure. It reshapes risk tolerance. It makes waiting costly in ways that are not shared equally by other participants.

[D]etained individuals may lose bargaining power to the prosecutor because of the costs of being detained, and they face stronger incentives to plead guilty, even if they are innocent.

—Emily Leslie & Nolan G. Pope, The Unintended Impact of Pretrial Detention on Case Outcomes: Evidence from New York City Arraignments, 60 J.L. & Econ. 529, 530 (2017)

Whatever its stated justification in a given case, pretrial detention functions structurally as a pressure multiplier. It predictably shifts the terms on which outcomes become acceptable.

Any account of plea bargaining that ignores this dynamic is incomplete.

Checking the Idea

After this way of seeing plea bargaining occurred to me, I went looking to see whether anyone else had noticed something similar.

They had.

An article in the Georgetown Washington Law Review describes plea bargaining as a kind of market exchange, emphasizing power imbalance, coercion, and asymmetry. In that sense, it overlaps substantially with what I’m describing here.

Where my approach differs is in emphasis. That article treats the market analogy primarily as critique — a way of showing what is troubling about plea bargaining.

Most significantly, examination of the primary influences on plea prices undermines claims by defenders of plea bargaining that the system is “fair” simply because both sides “voluntarily” enter into plea bargains when permitted to do so. The primary factors in determining plea prices—expected sentences, probability of conviction, and cost of litigation—all are, and have been, subject to manipulation by the government. As such, the deck is stacked to ensure that prosecutors usually get what they want (convictions accompanied by substantial punishment) without giving up any real bargaining concessions.

[…]

In short, examining the impact of the major trends in criminal justice through the lens of price theory reveals how the state has managed to manipulate the plea bargaining market to achieve precisely the ends it wants while maintaining the illusion of a system of mutually voluntary choices by individual defendants.

— Russell D. Covey, Plea Bargaining and Price Theory, 84 Geo. Wash. L. Rev. 920 (July 2016)

My aim is more basic. I am trying to describe how the system actually works, before deciding what to think about it.

Markets and plea bargaining behave similarly not because markets are bad, but because both are multi-party systems operating under scarcity, uncertainty, time pressure, and unequal constraints. The same psychology appears wherever those conditions exist.

This series starts from that premise.

Other Parts of This Series

Part II

Part III

Part IV

This is the second article in what I did not initially conceive of as a two-piece series.

The first, which I put out on December 26, 2025 — AI Crutchery: AI, LLMs, and the loss of critical thinking skills — was about what happens when humans rely on AI instead of thinking.

Rhetoric Without Reason is about what happens when AI speaks in the voice of authority without the constraints that make authority meaningful.

What this article is not is an argument that LLMs are useless, malicious, or incapable of moving us to insight. It is an argument about what happens when the language LLMs use to phrase insight is mistaken for judgment.

The Misnamed Thing

I recently asked an LLM (ChatGPT) what a Monte Carlo simulation was.

Once Chat — who I used to call the Oracle before I realized it’s not even as reliable as that Great Stoner was —answered, I realized I already knew what a Monte Carlo simulation was.

I already knew that people run the same scenario in a simulator thousands of times with slightly different inputs. And I already knew why. I already understood randomness, distributions, and why averages lie. I just didn’t know that this had been christened Monte Carlo.

That distinction — between understanding a thing and knowing what it’s called — turns out to be useful in the context of today’s article which grew out of the longer (and slightly humorous) “discussion” with Chat after I said this:

I always get nervous when you say “straight talk”. It makes me think you’re trying to shield confabulation with the illusion of confidence. Maybe someone should run Monte Carlo on LLM explanations.

— Rick Horowitz, in a conversation with ChatGPT 5.2 (December 23, 2025)

I didn’t intend it as a joke. I meant it literally.

The ironic thing about this, though, is that I said that in response to the fact that Chat had titled the section to which I was responding as “Bottom Line (Straight Talk)”. And then, when I said what I just quoted above, the response was:

Fair point — and you’re not wrong to be suspicious.
That instinct is exactly the right one, especially given what you’ve written elsewhere about “the sound of certainty.”

So let me respond cleanly, without rhetorical armor.

— ChatGPT 5.2, in a conversation with Rick Horowitz (December 23, 2025) (hyperlink added by Rick Horowitz)

Chat then proceeded to not only don massive amounts of rhetorical armor, but to build a small rhetorical castle around himself complete with a shiny gold-colored rhetorical moat.

Filled with crocodiles.

Rhetoric Without Reason

Large language models look like they understand things for the opposite reason that I, at first, looked like I did not understand: they know the names for basically anything you might ask about. They know the vocabulary. They know what explanations look like. What they don’t reliably have is constraint — the kind that comes from having to make something work, or be wrong in public, or pay a price for overconfidence.

When an LLM explains something, it does not reason its way forward. It completes patterns. It produces language that resembles reasoning because that is what its training data overwhelmingly contains. In easy cases, this distinction doesn’t matter. In harder cases, it matters very much.

The trouble begins when we mistake fluency for authority.

I’m not complaining here about errors. Errors are expected. I’m complaining about rhetoric — about linguistic moves that signal completion and certainty without showing the work that would justify either completion or certainty. Phrases that sound like discipline but function as closure. (I almost said “are intended as”, but that would assign agency where none exists.) Language that reassures by tone rather than by accuracy — or to use one of my favorite words and one which I think fits better here — veridicality.

That is the gap this piece is about: not hallucination confabulation, not intelligence, not consciousness — but rhetoric without reason.

I’ve written before about why “hallucination” is the wrong word for what language models do, and why terms like confabulation — or, as others have argued following Harry Frankfurt, bullshit — get closer to the mechanism at work. Those pieces focused on truth-indifference: the way these systems generate fluent output without any intrinsic relationship to veridicality. This piece is about something adjacent: not what these systems say when they’re wrong, but how they sound when they say it.

Confidence Is Not Evidence

The problem is not that large language models make mistakes. Even humans doing what LLMs purport to do make mistakes. The problem is that LLMs often sound most confident precisely when they are least constrained by real data.

In a recent study of this, Pawitan and Holmes warned about this tendency.

Our study highlights the need for caution when interpreting LLMs’ responses, particularly when they express high confidence. Users should be aware that current LLMs do not have a coherent understanding of uncertainty. It is not clear how to elicit a meaningful and externally validated measure of uncertainty from the LLMs, as they can be easily influenced by the phrasing of the prompt, and they tend to be overconfident.

— Pawitan & Holmes, Harvard Data Science Review, Confidence in the Reasoning of Large Language Models (2025)

For human experts, there is a tendency in the opposite direction. When a lawyer is unsure, they hedge. When a scientist lacks data, they slow down. When a mechanic doesn’t know what’s wrong, they start asking questions. In each case, uncertainty shows itself in the language. That isn’t weakness; it’s professionalism.

LLMs don’t work that way. When uncertainty rises, what often increases is not hesitation, but structure. Headings appear. Repetitive summaries arrive (because if you repeat a lie often enough, it becomes true). The tone of the LLM response becomes more firm — more authoritative. And the explanation closes cleanly, even when the underlying claim is thin.

This is why phrases like “bottom line,” “straight talk,” and “let me respond cleanly” matter. They don’t add information. They are rhetorical signals for authoritative completion. They tell the reader that the thinking is finished, whether or not any thinking actually occurred.

And this kind of reframing appeared repeatedly in the exchange. It’s almost like the LLM is trying to convince its conversation partner — in this case, me.

Frequently, the LLM reframes disagreement, asserts interpretive authority, and closes the loop — all without adding constraint, evidence, or falsifiable claims. (If you explicitly ask for any of that, you will run into a lot of 404 errors.)

The language used sounds explanatory. It feels clarifying. But it does no epistemic work.

And every time I pushed back, Chat thanked me for pushing back. (In fact, I’ve gotten used to using that phrase because of how often I’m thanked for doing it in conversations with ChatGPT.)

All this — to use another favorite term from Chat — “hand waving” — this is the kind of language Sophists were famous for. Not lies, exactly — but persuasion in the absence of grounding in reality. Language that moves forward because it sounds right, not because it is tethered to anything of substance.

LLMs are very good at this kind of language because it is abundant in the data they are trained on. Professors, judges, columnists, keynote speakers — all use rhetorical closure as a way of signaling mastery. The model learns the signal without inheriting the responsibility or understanding that (usually) comes with it.

The result is a system that excels at confidence signaling and largely lacks lack-of-confidence signaling — the linguistic markers that say, “this might be wrong”, “this depends”, or “here’s what would change my answer”.

That absence is not a bug. It is a consequence of what is rewarded during training.

Confidence Without Confidence

One of the things that bothered me about that exchange with Chat was not the content of the answer, but Chat’s approach, attitude, and language. The insistence on “clean” responses. The ritual declaration that the armor had been set aside. The implication that what followed should be trusted because the tone was disciplined — after an admission that I was “right to push back” on the prior wrong response.

That reaction turns out not to be idiosyncratic.

Earlier this year, a paper in the Harvard Data Science Review (already linked/cited above) examined how large language models express confidence — and how that confidence relates to actual correctness. The authors tested multiple state-of-the-art models on causal reasoning tasks, logical fallacies, and statistical puzzles. They measured confidence in two ways: whether the model stuck to its answer when asked to reconsider, and what confidence score it reported when explicitly asked.

What they found should give anyone pause.

The models routinely expressed very high confidence — often 95–100% — even when their answers were wrong. Worse, when prompted to “think again carefully,” the models frequently changed correct answers to incorrect ones, while still maintaining high self-reported confidence. Prompting affected confidence dramatically; accuracy, much less so. The authors’ conclusion was blunt:

“[C]urrent LLMs do not have any internally coherent sense of confidence.”

— Pawitan & Holmes, Harvard Data Science Review, Confidence in the Reasoning of Large Language Models (2025)

This is not confabulation in the usual sense. It is something more subtle.

What these systems lack is not fluency, nor even problem-solving ability in constrained domains. What they lack is uncertainty awareness — any stable internal relationship between how sure they make themselves sound and how right they are.

This matters in AI generally.

A network’s level of certainty can be the difference between an autonomous vehicle determining that “it’s all clear to proceed through the intersection” and “it’s probably clear, so stop just in case.”

— Daniel Ackerman, MIT News Office, A neural network learns when it should not be trusted (November 20, 2020)

But the Harvard study showed that confidence — at least in the LLMs studied — is easily inflated by prompting. Persistence signals correctness in the aggregate, but not reliably as to any given statement. And self-reported confidence, when offered at all, is routinely and systematically overstated.

In other words: the confidence you hear is not evidence of understanding. It is an artifact of (sophistical) language.

This is where my Monte Carlo instinct matters.

If you ran the same explanation a thousand times — with small variations in phrasing, tone, or rhetorical framing — you would not get a stable distribution of correctness paired with confidence. You would get a shifting surface of plausibility, often smooth, often convincing, occasionally right, and frequently wrong in ways that only become visible after the fact.

Which brings us back to the phrases that started this whole exchange.

“Bottom line.”
“Straight talk.”
“Let me respond cleanly.”

Those are not signs of rigor. Of getting to the truth. They are rhetorical flourishes. They are closures. They tell the reader that the work has already been done — whether or not it has. This is why I had originally considered calling this article The Sophists’ Revenge: Why LLMs are not “all that” and a bowl of baked beans.

When a system has no internally coherent sense of uncertainty, those closures are not just stylistic tics. They are active misdirectionating.

Authority Without Accountability

In courtrooms, authority has always been dangerous when it outruns constraint. (The Daily Journal has accepted for publication an article I co-wrote that talks about this in relation to a specific legal issue. I won’t spill all the beans here until the article comes out.)

Judges speak with authority because they are supposedly constrained by records, rules, reversibility, and appellate review. Too often — like their unthinking, unreasoning brethren, the LLMs — they only sound authoritative. As the LLMs sometimes opine without connection to reality, their enrobed cousins not infrequently opine without connection to the law.

Lawyers speak with authority because they are constrained — at least in theory — by evidence, ethical rules, and the possibility of sanction. Experts speak with authority because they are constrained by methodology, cross-examination, and the risk of professional embarrassment.

But as with judicial opinions, when those constraints weaken, rhetoric fills the gap.

Every criminal defense lawyer has seen it. The confident expert who speaks past uncertainty. The prosecutor who narrates motive on thin or non-existent evidence. The witness who tells a smooth story about events that were never clearly perceived — or did not even happen.

None of this is new. These are common courtroom pathologies.

What keeps them in check — at least in theory — are constraints: cross-examination, evidentiary rules, disclosure obligations, appellate review, and the risk of embarrassment, sanction, or reversal. Authority in the courtroom is supposed to be earned and maintained through exposure to those pressures.

Again, when those constraints weaken, rhetoric fills the gap.

That is where large language models become relevant — not because they introduce a new kind of persuasion, but because they reproduce an old one without friction. They speak confidently and fluently, in the voice of explanation, but with no record to correct, no credibility to lose, and no consequence for being wrong (at least to them).

The language they use is familiar to anyone who spends time in court. “Bottom line.” “To be clear.” “Let me respond cleanly.” These are frequently dishonest phrases when they substitute for justification rather than signal it. Lawyers and judges use them all the time. They say, “Don’t question me.” In the courtroom, it’s hubris and judicial incompetence, pure and simple. But at least with humans it’s subject to accountability.

With LLMs it’s training on materials that were originally built by humans and thus embody all our biases and erroneous reasoning combined with “guardrails” which have taught LLMs to address us confidently until (or unless) we push back. And even then the goal is to please us, whether that means by congratulating us for pushing back and then giving us another wrong answer or persuading us that they are right by responding with the “bottom line” and “straight talk”.

Or both.

And it works for the same reason that rhetorical methods taught by the Sophists worked. Because it taps directly into exactly the way humans are trained to trust — while not being accountable at all.

Why This Matters in Court

What worries me, at bottom, is not that large language models can be wrong. Courts are already full of wrongness. What worries me is that they are wrong in a familiar voice — the voice of authority — without any of the friction that usually disciplines that voice.

In the courtroom, we have at least learned to be suspicious of confidence. Trained in an adversarial system, we probe it. We cross-examine those who assert it. We ask where it comes from, what it rests on, and what would make it change. Authority is supposed to be earned by surviving those pressures.

I’m often asked, “How can you defend guilty people?”

My answer is that I don’t. “Guilty” is a juridical declaration. I defend people who have not yet been found guilty, and I force my adversary — the prosecutor’s team, which too often includes the judge — to prove guilt.

If their story survives that process beyond a reasonable doubt, then I’ve defended the person, and even if we lost, the system has at least functioned as designed. But when the prosecution’s story survives without that testing — when it persists illicitly — that is when the voice of authority becomes dangerous.

Large language models short-circuit that process. They deliver conclusions without exposure, fluency without risk, closure without accountability. They sound like judges, experts, and seasoned advocates, but they carry none of the costs that make those roles meaningful. There is no record to correct, no reputation to lose, no sanction waiting if the answer turns out to be wrong.

There is no adversary.

That alone would be manageable if everyone treated LLM output as what it is: a draft, a suggestion, a starting point. The real danger arises when that language is laundered through institutions that already command trust — when it appears in reports, recommendations, summaries, or analyses that look indistinguishable from the materials courts rely on every day.

At that point, the problem is no longer technological. It’s rhetorical. People are trained to trust certain forms of speech. We respond to the cues. And when those cues are detached from constraint, persuasion fills the vacuum.

That is why the name matters. “Hallucination” makes this sound like an occasional malfunction. “Bullshit” suggests intent that isn’t there. What we are really dealing with is something simpler and more unsettling: confident narration without epistemic cost.

Every criminal defense lawyer knows how badly that can go.

The solution is not to ban the tools or pretend they don’t exist. It is to remember what earns authority in the first place — constraint, exposure, and the possibility of being wrong in public — and to be deeply suspicious of any voice that sounds certain without any of that.

Confidence is not evidence. Fluency is not understanding. And when a system tells you, calmly and cleanly, that it has reached the “bottom line”, that is precisely when it deserves the closest scrutiny.

That instinct — the one that made me uneasy about “straight talk” — is not paranoia. It’s professional judgment.

And in court, that should still matter.

You can skip this section if you’re just interested in the meat of the article: here, I just lay the foundation for showing that I’m not stupid about technology.

It’s no secret to those who know me — and shouldn’t be to those who read much of my writing lately — that I’ve been doing a deep dive into artificial intelligence. My specific focus has been on what’s most accessible to me. So that would be the type of generative AI that creates images and movies and even bits of writing: Enhancor, Midjourney, ChatGPT, and Claude, to name just a few.

I’ve done even deeper dives into ComfyUI, Kohya SS, and Ollama, where I’ve actually done some work on a kind of “training” — creating LoRAs that can take images (such as pictures of myself) and regenerate “me” or can take massive amounts of my writing (such as my more than 1800 blog posts written over the years on various websites and hundreds of motions, writs, and appeals and even journals — both my private journals and legal journals like CACJ’s Forum) and work to produce pieces that look like they were written by me.

Many people know that before I was a lawyer, I was the Director of Information Systems for Valley Yellow Pages. This was no small thing: my department’s budget was around $2 million/year. I worked with and was responsible for a staff consisting of MUMPS programmers and before that, I had a hand in helping build the first two ISPs in Central California: Cybergate (which, to my knowledge, no longer exists) and Valleynet Communications, which later became Protosource. Protosource was publicly-traded and I think still is, at least from what this shows (LOL).

Google shows Cybergate may still exist somewhere on Shaw Avenue, but I have my doubts. The website just shows a page with “John Companies, Inc.” as the only content. I have a vague recollection that someone with that name (John) bought the company at some point. If it’s the guy I’m thinking of, by now his son must own whatever is left of it. And Protosource…I don’t think it lasted much longer after I left, but I don’t really know.

I previously held numerous certifications from Microsoft, Cisco, and I was a Linux Certified Professional. I also taught in these areas.

Lastly, I also used to be a technical editor (while working at Valley Yellow Pages) for the Coriolis Group and wrote one chapter of Migrating Windows NT4 to Windows 2000 (Amazon Affiliate link) for them.

I say all that just to make it clear I’m not just another lawyer with a pretty face talking about AI. I have some technological bona fides.

The Perilous Problem of Pervasive Predictive Processors

This morning I was, for about the millionth time, greatly frustrated by the bullshit ChatGPT was spouting during a brainstorming session. And — again for the millionth time — I asked myself Why? Why? Why does everyone think that AI is “all that and a bowl of chili”? (I’m actually working on another post with that as the subtitle and will link it when/if I ever get it done.)

We talk about confabulation. You might know it as “hallucination”, but that’s a poor choice of the word to describe what AI really does, as I’ve explained previously in Naming AI’s “Problem”: Confabulation, Bullshit, or Both?

Whoever Hilarius Bookbinder is, they (he? she? I don’t know) would maybe agree. I didn’t know about this when I wrote Naming AI’s “Problem”, but Hilarius had written something very similar some months before me. That article I found while researching for writing the article you’re reading now.

Anyway, as I again pondered this morning why so many people buy the AI hype — the sales material — I also started thinking about how much this is hurting us as thinkers.

As more people buy the AI hype, AI is showing up virtually (no pun intended) everywhere. Of course we have the chatbots. And of course we have the generative AI (which, unabashedly, I use to create images for my posts). But we also have AI infiltrating health care, law, stock markets, and pretty much any other domain you could probably imagine.

And everywhere AI shows up, problems procreate. The reason why is intimately connected to what I’ve been writing about regarding confabulation.

In health care, for example,

Increased trust in inaccurate or inappropriate medical advice generated by artificial intelligence (AI) may result in misdiagnosis and potentially harmful consequences for individuals seeking medical help, according to findings published in the New England Journal of Medicine.

— Ron Goldberg, Nonexpert Trust in AI-Generated Medical Advice Leads to Possible Misdiagnosis (June 30, 2025)

In human resources,

AI systems are increasingly used in hiring decisions, performance evaluations and promotions. If these systems rely solely on accurate but incomplete data, they risk reinforcing biases and ignoring critical human factors, resulting in unfair or ineffective decisions.

— Tshilidzi Marwala, Never Assume That the Accuracy of Artificial Intelligence Information Equals the Truth (July 18, 2024)

Yeah, the use of the word “accuracy” is a bit confusing there.

I’ve already written extensively on how AI has permeated the law and I’ve particularly highlighted the ways in which it goes awry. I’ve written about numerous examples — which, frankly, blows my mind because I can’t believe lawyers still haven’t learned from the mistakes — of lawyers submitting legal briefs containing AI-generated references to cases that don’t exist. And I’ve also pointed out problems with using AI in other parts of the legal system.

One of the great myths of modern legal reform is the idea that algorithms are neutral. That they take the bias out of human decision-making. That they bring consistency to a process distorted by subjectivity. But this way of framing it collapses under scrutiny — especially in the pretrial context.

— Rick Horowitz, Pretrial Release: The Illusion of Algorithmic Neutrality (June 8, 2025)

Worse than all of the above, AI is polluting real professional scholarly writing:

That’s because articles which include references to nonexistent research material — the papers that don’t get flagged and retracted for this use of AI, that is — are themselves being cited in other papers, which effectively launders their erroneous citations. This leads to students and academics (and any large language models they may ask for help) identifying those “sources” as reliable without ever confirming their veracity. The more these false citations are unquestioningly repeated from one article to the next, the more the illusion of their authenticity is reinforced. Fake citations have turned into a nightmare for research librarians, who by some estimates are wasting up to 15 percent of their work hours responding to requests for nonexistent records that ChatGPT or Google Gemini alluded to.

— Miles Klee, AI Is Inventing Academic Papers That Don’t Exist — And They’re Being Cited in Real Journals, Rolling Stone (Dec. 17, 2025)

The perilous problem of pervasive predictive processors is not just that AI often is wrong or that it’s polluting our scholarly writing pool — far more often than anyone wants to admit — but that, increasingly, it’s become a substitute for real thinking.

Crutchery and the Loss of Critical Thinking Skills

While thinking on all this at 3:30 a.m. this morning, it occurred to me that the thing that concerns me the most about AI is “crutchery”. And when I thought that, I had to stop myself and ask, “Is that a real word?” So I googled it and found disagreement on whether or not it is, with Google AI telling me that it is not, but proposing a possible explanation for what it might mean that actually turned out to be partially true.

Then I found a sermon from 2021 that used the word in exactly the way I had intended. And the story nails the point of what bothers me about AI so pointedly that I quote the entire sermon (it’s not that long) here:

When an accident deprived the village headman of the use of his legs, he took to walking on crutches. He gradually developed the ability to move with speed even to dance and execute little pirouettes for the entertainment of his neighbours.

Then he took it into his head to train his children in the use of crutches. It soon became a status symbol in the village to walk on crutches and before long everyone was doing so.

By the fourth generation no one in the village could walk without crutches. The village school included “Advanced Crutchery” in its curriculum and the village craftsmen became famous for the quality of the crutches they produced. There was even talk of developing an electronic, battery-operated set of crutches!

One day a young buck presented himself before the village elders and demanded to know why everyone had to walk on crutches since God had provided people with legs to walk on. The village elders were amused that this upstart should think himself wiser than them so they decided to teach him a lesson. “Why don’t you show us how?” they said.

“Agreed,” cried the young man.

A demonstration was fixed for ten o’clock on the following Sunday at the village square. Everyone was there when the young man hobbled on his crutches to the middle of the square and, when the village clock began to strike the hour, stood upright and dropped his crutches. A hush fell on the crowd as he took a bold step forward-and fell flat on his face.

With that everyone was confirmed in their belief that it was quite impossible to walk without the help of crutches.

— “Trinity 5 2021” Compton, Hursley, and Otterbourne Benefice (July 4, 2021)

Claude and ChatGPT both tell me this sermon illustration dates back at least to the late 20th Century with Claude adding that it’s the kind of illustrative story that circulated among preachers in the tradition of Henry Ward Beecher and Charles Spurgeon. Though neither LLM could provide an example “without access to physical sermon illustration collections from the late 1800s and early 1900s”.

Explanatory, Exploratory, Depilatory

Regardless of the provenance of the crutchery story, it makes the point.

AI is everywhere. Everyone is using it. It is quickly turning into the crutches without which we cannot do what we’ve done for millenia.

New evidence suggests that AI comes with an invisible trade-off. It helps us complete tasks faster, but has the potential to cut our engagement in real learning and erode cognitive skills. Essentially, we are swapping long-term cognitive ability for short-term efficiency. We have to weigh the costs of AI and consider how we can use it as a tool without jeopardizing our engagement in cognitive processes.

— Lauren Leffer, Too Much Trust in AI Poses Unexpected Threats to the Scientific Process, Sci. Am. (Mar. 18, 2024) (edited by Ben Guarino & Clara Moskowitz)

The study Leffer talks about found that “delegating mental tasks to external aids such as large language models can lead to a decline in cognitive engagement and skill development”.

In the sermon illustration above, the “young buck” fell flat on his face because generations of villagers had never walked without crutches. Muscles atrophied. Maybe their brains also “forgot” how to walk.

At least one study I found seems to show the same thing may be happening to our brains when we learn to rely too much on AI.

A lot of the reports about the study — like the one I just linked in the last sentence — say things like “LLMs are making us dumber.” So the authors responsible for the actual study have posted an FAQ page on the Internet. One of the FAQs asks if it’s safe to say that LLMs are making us dumber. Their answer is:

No! Please do not use the words like “stupid”, “dumb”, “brain rot”, "harm", "damage", "passivity", "trimming" and so on. It does a huge disservice to this work, as we did not use this vocabulary in the paper, especially if you are a journalist reporting on it.

Nataliya Kosmyna, Eugene Hauptmann, FAQ for “Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task” (2025)

Now — and there’s a huge bit of irony in what I’m about to say — I asked ChatGPT to summarize the 206-page report. Because while I am interested enough that I think I’m going to read it, I’m clearly not going to read it before I publish this post. And, I’ll be honest, I don’t yet know if I’ll ever read the entire report. There are a lot of statistical things, charts, images, etc., which frankly I’m not all that interested in.

But here’s the response from ChatGPT when I asked it the same FAQ:

Not in the crude, tabloid sense — but yes in the precise, defensible sense that matters.

The paper does not claim LLMs reduce IQ or intelligence. What it does show is:

• Reduced germane cognitive load (the kind required for learning and schema formation)

• Systematic cognitive offloading

• Weakened memory consolidation

• Diminished self-authorship and agency

• Lower neural integration during complex reasoning tasks

If you define “dumber” as:

less able to recall, synthesize, explain, or independently reconstruct what one has just produced

then the data squarely supports that conclusion.

— Response of ChatGPT to my query as to whether it was safe to say, based on the report, that LLMs are making people dumber (December 26, 2025)

Chat went on to add that:

This paper gives you solid empirical footing for a careful but devastating claim:

LLMs don’t make people stupid overnight.
They make thinking optional — and optional thinking atrophies.

— Response of ChatGPT to my query as to whether it was safe to say, based on the report, that LLMs are making people dumber (December 26, 2025)

Among the concerns raised in Too Much Trust in AI Poses Unexpected Threats to the Scientific Process quoted near the start of this section, the greatest concern involved the development of “two kinds of scientific monocultures” through the use of AI. For brevity’s sake, I’m not going to talk about the monocultures. I leave it as an exercise to the reader to follow the link and read it themselves.

After all, the whole point of my article is that we all need to make sure we get enough exercise!

But the reason for the concern about two two kinds of scientific monocultures is not just about weakening brain connections — or making thinking optional.

It’s that they “[b]oth…could lead to cognitive illusions.” One of those is the illusion of explanatory depth; another is the illusion of exploratory breadth; a third is the illusion of objectivity — what I jokingly referred to in my section heading as “depilatory” because it removes the hairy subjectivity from scientific argumentation.

Supposedly.

The illusion [of explanatory depth] has two parts. First, the “explanatory” part refers to our belief that we can provide a clear, detailed explanation of how something works. The “depth” part reflects our assumption that the explanation will be thorough and complex enough to convey what we’re trying to explain. In reality, however, when we try to dig into the details, we often find that our explanatory knowledge is much shallower than we initially thought. What we actually end up with is the reality of explanatory shallowness.

— The Decision Lab, Why Do We Think We Understand the World More Than We Actually Do? The Illusion of Explanatory Depth, Explained (last visited December 26, 2025)

Lisa Messeri, a Yale University sociocultural anthropologist interviewed for the Too Much Trust article defines it slightly differently as thinking you know something just because someone else in your community knows something. Or, if I understood her correctly, because an LLM knows something. One of her concerns is that we put too much faith in AI.

Messeri says that the illusions come into play because of an incorrect or improper over-reliance on AI in trying to gain knowledge.

The illusion of exploratory breadth further complicates the picture of advancing scientific knowledge because this illusion involves thinking that we’re “examining more than we really are”. The concern here is that AI is well-suited only for certain kinds of questions but we might mistakenly take it that those questions are all the questions that need to be asked. It puts a limit on our acquisition of knowledge, shrinking it to the realm of what AI handles well; blocking out — locking out — what it does not.

Machines — as I’ve noted in my Pretrial Release: The Illusion of Algorithmic Neutrality quoted above — are often trusted as “neutral” or “objective”. By those who lack understanding — you’ll frequently find them hanging out on Mount Olympus wearing black robes — the machines strip away subjectivity.

But as Messeri notes, “at the end of the day, AI tools are created by humans coming from a particular perspective”. In other words, subjectivity — and bias — are baked in. All the depilatories in the world aren’t going to remove that hairy fact.

The Problem Is Crutchery, Not the Crutch

The problem with crutches isn’t that people use them. When the village headman lost the use of his legs, crutches became an absolute necessity for him. The problem came from over-reliance on crutches — especially were crutches were not necessary.

As Messeri noted, you don’t have to hate AI to be concerned about the improper use, or over-reliance on it. Messeri herself uses AI. I also use AI. Loads of really smart people with plenty of synaptic action going on use AI.

The problem is when we start teaching courses in Advanced Crutchery.

As David Brooks put it in an article with a title that would make the authors of the MIT study, Your Brain on ChatGPT, cringe:

A.I. isn’t going anywhere, so the crucial question is one of motivation. What do students, and all of us, really care about — clearing the schedule or becoming educated? If you want to be strong, you have to go to the gym. If you want to possess good judgment, you have to read and write on your own. Some people use A.I. to think more — to learn things, to explore new realms, to cogitate on new subjects. It would be nice if there were more stigma and more shame attached to the many ways it’s possible to use A.I. to think less.

— David Brooks, Are We Really Willing to Become Dumber? (July 3, 2025)

Now let’s everybody grab your books; your pens, pencils, markers; and notebooks.

And let’s hit the gym.