On November 20, 2023, the D.C. Circuit heard over two hours of oral argument in Donald Trump’s appeal of the district court’s order restricting his statements about court staff, prosecutors, and witnesses. The length and breadth of the oral argument underscored the difficulty of the First Amendment issues presented in the case. The D.C. Circuit—and perhaps the Supreme Court thereafter—will have little choice but to create brand-new law governing how the First Amendment applies to Trump’s statements. That’s because Trump is such a unique figure that existing case law doesn’t begin to provide guidance.
You all did the reading, right?
In Part I, I made the following points:
The gag order prohibits Trump from “targeting” the Special Counsel and his team, the judge’s staff, and “any reasonably foreseeable witness or the substance of their testimony.” So no more “Deranged Jack Smith & his team of Thugs” and no more referring to Mike Pence as a “coward.”
If Trump wasn’t a criminal defendant, the gag order would be unconstitutional. Trump’s statements are constitutionally protected unless they’re “intended to produce” and “likely to produce” “imminent disorder.” That standard isn’t satisfied.
But Trump is a criminal defendant. There’s Supreme Court case law holding that (1) courts can take pre-trial prophylactic measures to prevent juries from being influenced by publicity, including (2) punishing lawyers for publicizing inadmissible evidence during press conferences.
But but, there’s no case law on whether (1) courts can gag the criminal defendant over his own objection, or (2) whether courts can gag the criminal defendant for reasons other than preventing jurors from being influenced by publicity.
In Trump’s case, the district court issued its order based on its finding that Trump’s statements “pose a significant and immediate risk that (1) witnesses will be intimidated or otherwise unduly influenced by the prospect of being themselves targeted for harassment or threats; and (2) attorneys, public servants, and other court staff will themselves became targets for threats and harassment.” There’s zero case law coming close to deciding whether this is a sufficient basis to gag a criminal defendant.
There’s no case law, time will tell.
It’s boring to say “there’s no case law, time will tell.” Does the First Amendment permit the gag order or not?
Well, I don’t know. Rather than offering a definitive answer here—which we’ll get in the next week or two from the D.C. Circuit anyway—I’d like to probe some of the novel questions raised by the district court’s order, and then explain how they could have arisen only in a case involving Trump.
What types of risks justify the gag order?
At core, the district court issued the gag order because it concluded there was a risk that bad things would happen if Trump kept saying the things he was saying. But a risk of what, exactly?
Suppose we say that before issuing a gag order, a court must make a predictive judgment that Trump’s statements will cause X to happen, with probability Y. What’s X and what’s Y?
I can think of at least four possibilities for X:
Bad things that will occur even if third parties behave, but that don’t affect the trial’s integrity. Trump’s statements might cause prosecutors, court staff, and witnesses to worry that someone’s going to harass them. Maybe they’ll have to pay for extra security. Even if no one actually does harass them, and even if their concerns do not affect the trial, it’s bad for public servants and witnesses to live in fear.
Bad things that will occur even if third parties behave, and that do affect the trial’s integrity. Even if third-party harassment never materializes, the fear of harassment wrought by Trump’s statements might influence how trial participants behave, affecting the trial’s integrity. For example, witnesses might alter their testimony or refuse to cooperate because they fear Trump’s rages and third-party harassment. The prosecutors may pull punches to avoid being referred to as thugs.
Third parties doing bad things that don’t affect the trial’s integrity. On August 5, 2023, 43-year-old Abigail Jo Shry called Judge Chutkan’s chambers and left a vile and threatening voicemail with a racial slur. Ms. Shry, described in media accounts as an alcoholic and drug addict without stable residence, was indicted in the Southern District of Texas and was placed in an in-patient drug rehabilitation facility pending trial. Ms. Shry’s actions will not affect the trial’s integrity: it is extremely unlikely that Judge Chutkan, an experienced federal judge, will allow her rulings to be influenced by a crazy person leaving a threatening voice message. Similarly, even if Trump’s statements induce future Abigail Shrys to leave threatening voice messages, the experienced public servants in this case will ignore them and stay the course. However, it is still bad for crazy people to be leaving threatening voice messages.
Third parties doing bad things that do affect the trial’s integrity. The nightmare scenario is that a third party becomes radicalized by Trump’s statements and physically harms a witness or public servant. Less nightmarish, but still bad, would be third parties harassing witnesses in a manner that causes them to change their testimony.
So which is it? Does the court have to predict that all of these things will happen? Some of them? Only one of them?
Presumably there is some probability that all four of these scenarios will arise, but some are more likely than others: throwing up numbers randomly, maybe there’s a 30% chance that trial participants will be nervous but a 1% chance that a witness will change his testimony because of Trump’s statements on Truth Social.
To return to where we started: Suppose we say that before issuing a gag order, a court must make a predictive judgment that Trump’s statements will cause X to happen, with probability Y. As outlined above, it’s unclear what X is. It’s also unclear what Y is. How probable do any of these bad events have to be to justify a gag order? Words like “significant” or “substantial” don’t help. Is it 50%? 5%? 0.5%? Does it depend on which bad event we’re talking about?'
Additional questions abound. Should the analysis vary from person to person? Some witnesses may be more susceptible to influence than others. One would suspect that former Vice President Pence has thick skin and can take public criticism. Maybe other witnesses have thinner skin. Should this be considered in deciding what speech is permissible and what isn’t?
How should we think about causation here? Trump’s statements are part of a broader ecosystem of Trump supporters criticizing prosecutors, judges, and witnesses. If Trump is silenced, those criticisms will continue. Does the court have to make a predictive judgment that Trump’s statements will be the but-for cause of potential or actual harassment or trial disruption?
Does the court have to consider the counterfactual? Suppose the gag order is upheld. Trump undoubtedly will rage about the Unconstitutional Gag Order; such a statement would not itself violate the gag order. Thousands of Trump supporters, who are not bound by the order, will respond by upping the ante on their verbal attacks of the judge, prosecutors, and witnesses. Suppose the judge predicts that third-party harassment is just as likely to occur in this scenario as without the gag order, except that the etiology of the third-party harassment will change: without the gag order, people will be radicalized by Trump’s statements, and with the gag order, people will be radicalized by Trump supporters complaining about the gag order. Is this a basis to deny the gag order?
Any risk analysis is going to be unscientific. It is impossible to accurately assess the probability of any particular outcome. Even if we come up with a specific legal standard for assessing whether a gag order is permissible, there is a risk of a garbage in, garbage out phenomenon—the risk estimates will be arbitrary, so the output of the specific legal standard will also be arbitrary. Still, it’s hard to see how any meaningful legal analysis could be done without developing such a standard.
What evidence should the court rely on in making predictions about Trump’s behavior and its effects on third parties?
So far as anyone knows, no one has (yet) harassed the Special Counsel or any potential witnesses. The pathetic Ms. Shry is the only person we know of that has harassed the judge, and that harassment occurred before Trump started attacking the judge. Standing alone, that’s a pretty thin record on which to find a likelihood that Trump’s statements will induce third parties to behave badly.
To fill this lacuna, the Special Counsel’s appellate brief asserts that the episode with Ms. Shry “was part of a pattern, stretching back years, in which people publicly targeted by the defendant are, as a result of the targeting, subject to harassment, threats, and intimidation.” The brief offers the following examples:
Trump criticized election officials and workers, who then testified before Congress that they were harassed and threatened on the Internet.
A state judge needed additional police protection after Trump criticized him.
Trump criticized the New York judge’s law clerk, after which the presiding judge said: “Since the commencement of this bench trial, my chambers have been inundated with hundreds of harassing and threatening phone calls, voicemails, emails, letters, and packages.”
To my eye, this is still pretty thin. It’s all hearsay and we don’t know the details of the harassment and whether it can be linked to Trump’s statements. Even weaker is this statement from the Special Counsel’s brief:
One of the defendant’s supporters recently described this dynamic. JA.296 (the defendant “knows exactly what to say and what not to say” and “won’t go up to someone and say, ‘I want you to kill someone,’” but will instead “send someone, to tell someone, to kill someone”)"
The citation appearing at JA.296 is a YouTube channel entitled “60 Minutes Australia” displaying an episode entitled “Donald Trump Spills Secrets.” In that episode, an audio clip is played, with dramatic music in the background, of Australian businessman Anthony Pratt offering his views about Trump. (You can watch the video on YouTube here; the quoted statement appears at the 7 minute mark). No context is offered for Pratt’s statement, nor does Pratt state the basis for his views. It’s just a dude, who knows Trump, opining on what he thinks about Trump. Immediately after this statement, the episode shifts gears to salacious allegations about Trump’s relationship with Melania. I recognize the Rules of Evidence don’t apply to a pre-trial hearing of this nature, but this type of evidence—an Australian YouTube channel playing a context-free recording of an Australian businessman offering his personal impressions of Trump—is so unreliable that it shouldn’t be considered in any judicial proceeding, much less a proceeding imposing a prior restraint on a criminal defendant’s speech.
And yet. Judges live in the world of reality, and it is obvious that when you have millions of people reading Trump’s statements, some of them might act on them. And, on the back of everyone’s minds are the events of January 6, 2021. Under one interpretation of the events that day, Trump’s statements incited his supporters to commit acts of violence. At least arguably, the gag order is justifiable to prevent this outcome from recurring. Of course, Trump has not yet been tried for his actions related to January 6, and the court cannot assume he is guilty when entering a pretrial order.
What does the government have to prove about Trump’s mental state?
Some Supreme Court cases hold that a person can’t be punished for particular types of speech unless the person had a particular mental state while speaking. For example:
Under Brandenburg v. Ohio, 395 U.S. 444 (1969), and Hess v. Indiana, 414 U.S. 105 (1973), incitement is constitutionally protected unless the speaker intends to provoke imminent disorder.
Under New York Times v. Sullivan, 376 U.S. 254 (1964), a public figure cannot recover for libel unless the speaker knew the statement was false or recklessly disregarded whether it was false or not.
Most recently, in Counterman v. Colorado, 143 S. Ct. 2106 (2023), the Supreme Court held that a person cannot be punished for uttering “true threats” unless the person “had some subjective understanding of the threatening nature of his statements,” but “a mental state of recklessness is sufficient.”
Both the theoretical and the historical justifications for these cases are open to debate. Justice Barrett’s dissent in Counterman, which advocates for an objective standard on true threats, is particularly well done in my opinion. Still, these cases raise the question whether the government must prove that Trump knew, or was reckless, that his statements would incite violence.
I don’t think the government has to make that showing for Trump’s pre-gag order statements. Trump isn’t being punished for those statements; instead those statements are being used to justify a forward-looking order intended to protect public safety. But what about Trump’s post-gag order statements? If Trump violates the gag order, he will be engaging in contempt of a court order, which is a crime. For Trump to be criminally liable, what should the government have to prove regarding his mental state?
Should the government have to prove that Trump intended for his statements to incite violence or intimidate witnesses, as in Brandenberg?
Should the government have to prove that Trump was merely reckless as to the risk his statements would incite violence or intimidate witnesses, as in Counterman?
Should the government merely have to prove that Trump was aware he was violating the court order, irrespective of whether he intended to cause any additional harm?
Should the government merely have to prove that Trump knew he was speaking, such that if the statements objectively violated the court order, Trump is liable?
Not only is it challenging to determine the appropriate mens rea standard, but it’s challenging to assess Trump’s mental state. Criminal defendants are generally not like you and me, and so it is often hard to understand what was going through their mind. But Trump is especially not like you and me (I am skeptical that Trump is a regular reader of this particular Substack), and so assigning conventional legal labels like “intent” and “recklessness” to Trump’s mental states is uniquely taxing.
Is it relevant that Trump’s statements are political speech?
In case you have not heard, Trump is running for President. As the Supreme Court has emphasized many times, “the First Amendment has its fullest and most urgent application precisely to the conduct of campaigns for political office.” FEC v. Cruz, 142 S. Ct. 1638, 1650 (2022). Most, perhaps all, of Trump’s statements are material to his presidential campaign:
Anticipated witnesses, such as Mike Pence and Mark Meadows, are former political figures. Trump wants to call those political figures cowards and liars. Political figures attacking other political figures is quintessential political speech.
One of Trump’s core themes is that he’s a victim of a widespread conspiracy initiated by Crooked Joe Biden, and that only he can Drain the Swamp. Bashing the Special Counsel—who, despite being Special, is ultimately part of the Department of Justice—is part of that theme.
More broadly, all of Trump’s public statements are relevant to his fitness for office. For our fellow Americans who have not yet formed an opinion on Trump’s fitness for office—say, people who recently emerged from lengthy comas, or people who recently returned to Earth following alien abductions, like at the end of Close Encounters of the Third Kind—reviewing Trump’s statements is surely relevant in deciding whether he’s the type of person who should be President or not.
On the one hand, Trump shouldn’t get special rights merely because he’s running for President. On the other hand, millions of Americans shouldn’t be deprived of hearing from a major political candidate because of a piddling thing like a federal indictment. It’s a pickle.
Is the gag order vague, and what happens if it’s impossible for the order to be clearer?
The gag order bars Trump from making “any public statements” that “target” the Special Counsel and his staff, the court’s staff, and “any reasonably foreseeable witness or the substance of their testimony.” It exempts “statements criticizing the government generally, including the current administration or the Department of Justice; statements asserting that Defendant is innocent of the charges against him, or that his prosecution is politically motivated; or statements criticizing the campaign platforms or policies of Defendant’s current political rivals, such as former Vice President Pence.”
One can postulate interpretations of this gag order that render it either blatant viewpoint discrimination or absurdly overbroad. Under one possible interpretation of the phrase “target any reasonably foreseeable witness,” Trump can’t say anything negative about, e.g., Mike Pence, but can only say positive things. This would be pure viewpoint discrimination that would be quite hard to justify.
Under another possible interpretation of the phrase “target any reasonably foreseeable witness,” Trump can’t say anything about major political figures in his administration who aren’t “current political rivals,” which would be overbroad. Likewise, under a possible interpretation of “target … the substance of their testimony,” Trump can’t say anything about the events following the 2020 election, because the “substance of their testimony” will include testimony about those events. That, too, can’t be right.
Likewise, what does it mean to say that Trump can’t “target” the Special Counsel? What if Trump publicly expresses disagreement with one of the arguments in the prosecutor’s legal filing and mentions the Special Counsel’s name. Is that “targeting”?
I very much doubt that the judge would interpret the gag order in this manner. But the fact that the order is even susceptible to such interpretations introduces a vagueness problem. The Supreme Court has expressed particular concern about vague laws in the First Amendment context, where there is a risk that ambiguous prohibitions will deter constitutionally protected speech. That is a problem here—even if the judge would not actually punish Trump for a relatively anodyne statement criticizing the Special Counsel, there is a risk that Trump will be deterred from making such statements by virtue of the order’s ambiguity.
But it is hard to blame the district court. It’s impossible to enact a gag order with clearly defined boundaries in this context. Perhaps the language of the gag order can be clarified slightly, but significant ambiguity is inevitable.
The problem is that Trump’s statements routinely straddle the line between political speech and witness intimidation. When Trump calls Mike Pence a liar and coward, he is criticizing a political official for being a liar and coward while also criticizing a witness for being a liar and coward. Any gag order will produce a significant number of edge cases merely because of the … edgy nature of Trump’s public statements.
The fundamental question here is whether the inevitable vagueness of a gag order justifies not having a gag order. Which is worse, an ambiguous order deterring political speech or an anything-goes policy authorizing witness intimidation?
It’s because it’s Trump.
In summary, there are a lot of unresolved questions in this case. Unresolved questions are common in our legal system. That’s why courts exist—to resolve them. But these are no ordinary unresolved questions. They are completely different from any First Amendment issues that have arisen before. And that’s because Trump is completely different from any candidate that has arisen before. To overstate the point a little bit, Trump breaks existing First Amendment doctrine.
I’d like to point to three examples of this phenomenon.
Low-value political speech. Content-based restrictions on speech are generally justified on the theory that some speech has such little value that it’s OK to ban it. As the Supreme Court recently explained in Counterman, “historically unprotected categories of speech” are of “such slight social value as a step to truth that any benefit that may be derived from them is clearly outweighed by the social interest in their proscription.” Conversely, for purposes of the First Amendment, political speech is considered the highest-value type of speech—per Counterman, it is at the “First Amendment’s core.”
What are we to make of Trump’s statements such as “IF YOU GO AFTER ME, I’M COMING AFTER YOU”? In one sense, this speech is utterly valueless as “a step to truth”—it’s a threat of vengeance against people who participate in the justice system’s pursuit of truth.
But in another sense, this speech is extremely valuable. Trump is running for President. He’s telling us what he’s going to do as President—wield revenge against his perceived political enemies. Shouldn’t we want to know what a presidential candidate will do if he’s elected? It’s hard to imagine a more useful type of political speech than a politician’s own statements about how he will exercise his powers.
This difficult issue arises because of Trump’s uniquely atavistic presidential campaign, in which statements like “IF YOU GO AFTER ME, I’M COMING AFTER YOU” are inextricably intertwined with Trump’s platform. His statements are valueless political speech—a heretofore unrecognized category. First Amendment doctrine was not developed for presidential candidates with platforms like this.
The numerator problem. The Supreme Court’s standard for incitement liability— advocacy is protected under the First Amendment unless the speech is both intended to, and is likely to, yield imminent disorder—was not built with Trump in mind.
Why did the Supreme Court develop such a speech-protective standard for incitement liability? Fundamentally, I think the Court was skeptical that abstract advocacy really could lead to the type of social unrest that would warrant banning it. The type of rabble-rousers who advocate violence typically don’t have widespread followings. The Nation can tolerate perennial candidates advocating for the overthrow of the political system. Meanwhile, mainstream political candidates aren’t likely to make statements that provoke political violence—or so the Court thought.
Trump combines (1) extremely controversial statements with (2) tens of millions of followers on social media. His controversial statements don’t rise to the level of incitement under Brandenburg. Indeed, the chances that any of his statements will provoke violence or harassment among any one of his followers—the denominator—is low. But the number of followers—the numerator—is high enough that the aggregate risk of violence or harassment warrants concern. Put bluntly, I don’t think the drafters of Brandenburg envisioned a scenario where a presidential candidate with 80 million followers was threatening witnesses, and almost inciting violence under Brandenburg’s standard, on social media.
Common knowledge. Trump is the most famous person ever to be put on trial in the United States. Making factual findings on a closed record about Trump is a fool’s errand. No judge or juror can reasonably be expected to forget everything they know about Trump in adjudicating a dispute involving Trump. In making a judgment regarding the anticipated effects of Trump’s statements, it is inevitable that the judge will take notice of well-known social facts about Trump’s widespread popularity and influence. Still, Trump, like anyone else, is entitled to due process before becoming the subject of a prior restraint on speech.
Trump is so famous that it’s hard to understand what due process looks like in this context. Suppose the court wanted to conduct an evidentiary hearing on the question of “what will Trump’s followers do if Trump uses the phrase ‘Deranged Jack Smith’ over and over again?” What would this evidentiary hearing look like? Would you have prosecutors testify? Judges? Random Trump supporters? January 6 criminal defendants? Expert witnesses? Trump himself? Any hearing addressing the likelihood of Trump’s statements provoking unrest would be a circus.
I suspect that, at the end of the day, Trump will be subject to some form of pretrial gag order, even if it is not identical in all respects to the gag order imposed by the district court. We will see. However the gag order is ultimately formulated, it will be based on considerations that have never before been considered by any court. The gag order—like many other Trump-related proceedings in the coming months—will reflect the Law of Trump.
Happy Thanksgiving!
This is clearly a special case of the more general problem of stochastic terrorism. Brandenburg's standard of "imminent lawless action" is difficult to meet because of the intent requirement. But in the context of a gag order, where the defendant is already subject to sanctions for various behaviors before the court, that definition could be altered: sanctions in a situation where the defendant's speech DOES generate lawless actions. There are two tiers for this: the speaker (S), and the person who acts lawlessly based on the speech (LA==lawless actor). Clearly LA's can violate the law on threats, as the woman from Texas shows. Let's call the person thus threatened the "PV" (potential victim).
So the order would be set up with a clear schedule of sanctions (mostly monetary) if the speech (defined below) DOES result in a lawless action.
This would take away the "intent" or at least make the mens rea "recklessness." Lawless actions would include threats by outsiders against PVs in any way connected with the case or actual changes of testimony by witnesses as measured by what they have already told prosecutors (including sudden "I can't remembers" which may not arise to the level of perjury by the testifying witness but which can be shown by evidence that the inability to remember is pretty recent.)
Under this scenario, the act of the woman in Texas would indeed be a violation had it happened after the order was in place.
There would be a MUCH stronger set of sanctions, including incarceration, if the LA actually DID carry through on the threat--an actual attack on at PV as opposed to the simple threat of an attack.
As to what "speech" is covered: I'd say that it would be anything that an objective observer would call "inflammatory." Basically, name calling above and beyond mere criticism. Thus "Mike Pence betrayed the country by not acting to stop the election count" would be criticism, but "Mike Pence is a spineless traitor whose treason damaged the American People" would be inflammatory. Again, "political speech" would not be prevented if BOTH non-inflammatory nor resulting in any threat by a LA.
The sanctions would be set up by increasing the amount for each violation of the order (remember, violation is the occurrence of a result) and would in its higher ranges be tailored to the defendant's actual net worth. (This would create a real problem for trump, and would include the value of his "brand" as of the date the gag order was imposed. Would he admit to exaggerating his worth to keep the sanction amounts down? I doubt it).
This approach is in line with the general idea of the rule of law, that actions have consequences, something we've already seen in the slander laws. for speech acts. Trump could CHOOSE not to use inflammatory speech and avoid the sanctions entirely, or he could take a chance that his inflammatory speech would not actually result in his followers becoming LAs. For this purpose, the mental health of the LA would not be an issue. One has to assume that those who make actual criminal threats based on the speech of S have at least one screw loose.
There might have to be adjustments depending how many tier one threats occur after a particular Trutheet. But once at least ONE threat is made, the counter resets and the next inflammatory statement after it would result in tier two sanctions if it provoked another threat.
Funds from the sanctions would go to some cause like increasing the number of public defenders and the sanctions would a) not be dischargeable in bankruptcy and b) not be reversed even if the defendant were found either innocent or subject to a hung jury.
On the broader issue, beyond the court situation, stochastic terrorism could be dealt with in a similar way by giving a private right of action against S to anyone who is actually damaged by a LA who can be shown to be was motivated by the speech of S--burden on plaintiff to show this. Again, the inflammatory language standard would in place and damages would depend on the amount of harm--from worry to hiring a security guard to having to go into hiding. Further, the "numerator" of the number of people LIKELY to threaten based on S's utterances would need to be taken into account. The level of S would have to involve "fans" at the full state or national level, probably
I don't use "numerator" and "denominator" the way you do here. The probability that any of Trump's statements will incite a particular person to a lawless act is very low; but the number of people that that probability will be applied to is very large (the phrase for this is "stochastic terrorism"). Buying a lottery ticket is very low probability, but a lot of lottery tickets are sold; someone will eventually win.