dissenting:
This case is proof positive that hard cases make bad law, and that when the case is very hard — meaning that competing legal and moral imperatives pull with impressive strength in opposite directions — there is the distinct danger of making ve-ry bad law.
The majority opinion in this case suitably struggles with the difficult First Amendment issues before us concerning whether the posters and website at issue are or are not First Amendment protected speech. The legal standard the majority applies, however, is, in my view, insufficiently cognizant of underlying First Amendment values, for reasons that are largely explained in Judge Kozinski’s dissent, and for additional reasons that I develop below.
Moreover, the majority, in an offhand way, also decides two evidentiary issues that, I can say with some confidence, would not be decided so summarily, and would probably not be decided in the same way, were this a less wrenching case on its facts. Keeping one’s eyes on the broader picture is not always easy when people’s fives — in this case the fives of medical professionals — are being severely disrupted because they are performing constitutionally protected activities in a perfectly lawful manner at the behest of people who want their services and are entitled to have them. As judges, though, we need to recognize that we are not writing for this day and place only, and that rulings that appear peripheral in the present context will take on great significance as applied in another.
I
The First Amendment and True Threats
1. Clarifying the issue: The reason this is a hard First Amendment case becomes somewhat obscured in all the factual detail and quotation of precedent that we as judges engage in. The essential problem — one that, as far as I am aware, is unique in the plethora of “threat” cases and perhaps more generally in First Amendment jurisprudence — is that the speech for which the defendants are being held liable in damages and are enjoined from reiterating in the future is, on its face, clearly, indubitably, and quintessentially the kind of communication that is fully protected by the First Amendment.
The point is not simply that the two posters and the Nuremberg files contain no explicit threats that take them outside the free speech umbrella. We are not talking simply about ambiguous or implicit threats that depend on context for their meaning, such as the Ryder trucks in United States v. Hart, 212 F.3d 1067 (8th Cir.2000). Rather, the pivotal issue for me is that what the communications in this case do contain has all the attributes that numerous cases and commentators have identified as core factors underlying the special protection accorded communication under our Constitution.
*1102The posters and website are all public presentations on a matter of current moral and political importance; they provide information to the public on that matter and propose a — peaceful, legal — course of action; and they were presented with explicit reference to great moral and political controversies of the past. Cases that are a virtual First Amendment “greatest hits” establish that these kinds of expressions— those that provide information to the public (particularly when directed at publicly-available media), publish opinions on matters of public controversy, and urge others to action — are the kinds of speech central to our speech-protective regime, and remain so even when the message conveyed is, in substance, form, or both, anathema to some or all of the intended audience. See, e.g., Garrison v. Louisiana, 379 U.S. 64, 74-75, 85 S.Ct. 209, 13 L.Ed.2d 125 (1964) (“[Sjpeech concerning public affairs is more than self-expression; it is the essence of self-government.”); Roth v. United States, 354 U.S. 476, 484, 77 S.Ct. 1304, 1 L.Ed.2d 1498 (1957) (The First Amendment “was fashioned to assure unfettered interchange of ideas for the bringing about of political and social changes desired by the people.”); New York Times Co. v. Sullivan, 376 U.S. 254, 266, 84 S.Ct. 710, 11 L.Ed.2d 686 (1964) (The First Amendment “attempt[sj to secure the widest possible dissemination of information from diverse and antagonistic sources.”); Id. at 271, 84 S.Ct. 710 (“The constitutional protection does not turn upon the truth, popularity, or social utility of the ideas and beliefs which are offered.”); Thornhill v. Alabama, 310 U.S. 88, 102, 60 S.Ct. 736, 84 L.Ed. 1093 (1940) (“Freedom of discussion, if it would fulfill its historic function in this nation, must embrace all issues about which information is needed or appropriate to enable the members of society to cope with the exigencies of their period.”); Thomas v. Collins, 323 U.S. 516, 537, 65 S.Ct. 315, 89 L.Ed. 430 (1945) (“ ‘Free trade in ideas’ means free trade in the opportunity to persuade to action, not merely to describe facts.”); Terminiello v. City of Chicago, 337 U.S. 1, 4, 69 S.Ct. 894, 93 L.Ed. 1131 (1949) (Speech “may indeed best serve its high purpose when it induces a condition of unrest, creates dissatisfaction with conditions as they are, or even stirs people to anger. Speech is often provocative and challenging. It may strike at prejudices and preconceptions and have profound unsettling effects as it presses for acceptance of an idea.”).
Tested against these most basic premises, there can be no doubt that the documents upon which the damages judgment and injunction in this case were based were, on their face, “expression[sj of grievance and protest on one of the major public issues of our time,” and, as such, documents that “would seem clearly to qualify for ... constitutional protection.” New York Times, 376 U.S. at 271, 84 S.Ct. 710. The posters and website could not and would not have been proscribed, as “true threats” or otherwise, had there been no (1) history of similar — although not at all identical' — publications put out by other people that were followed by murders-by other people, not members of either of the two defendant organizations' — of health professionals who performed abortions; and (2) repeated advocacy by these defendants of the proposition that violence against abortion providers can be morally justified, advocacy that all concede was, standing alone, itself protected by the First Amendment. See Brandenburg v. Ohio, 395 U.S. 444, 447-48, 89 S.Ct. 1827, 23 L.Ed.2d 430 (1969) (“[Tjhe mere abstract teaching ... of the moral propriety or even moral necessity for a resort to force and violence, is not the same as preparing a group for violent action and steeling it to such action”) (quoting Noto v. United States, 367 U.S. 290, 297-98, 81 *1103S.Ct. 1517, 6 L.Ed.2d 836 (1961)).1 The precise question before us is therefore whether that context is sufficient to turn a set of communications that contain speech at the core of the First Amendment’s protections into speech that can be proscribed pursuant to an injunction and compensated for through damages.
2. An analogy: Stated in those terms, the issue bears a close resemblance to that faced by the courts with regard to First Amendment limitations on defamation actions, beginning with New York Times Co. v. Sullivan. Like “true threats,” false speech has long been understood as a category of communication that contains few of the attributes that trigger constitutional speech protection and so great a likelihood of harming others that we refer to the speech as being beyond the protection of the First Amendment. See R.A.V. v. City of St. Paul, 505 U.S. 377, 383, 112 S.Ct. 2538, 120 L.Ed.2d 305 (1992). Like “true threats,” false, defamatory speech can severely disrupt peoples’ lives, both by affecting them emotionally (as does apprehension of danger) and by impairing their social ties, their professional activities, and their ability to earn a living (as does the perceived need to protect oneself from physical harm).
The Supreme Court since the 1960s has developed a set of discrete principles designed not to provide false speech with constitutional protection, but to erect, on an ascending scale depending upon the perceived value of the particular kind of speech to the common dialogue that the First Amendment is designed to foster, doctrinal protections within defamation law that minimize self-censorship of truthful speech. Those protections are based upon realistic assessment of the vagaries of litigation and the fear of crippling damages liability.2
For example, New York Times observed that “[allowance of the defense of truth ... does not mean that only false speech will be deterred,” because “[u]nder such a rule, would be critics of official conduct may be deterred from voicing their criticism, even though it is believed to be true and even though it is in fact true, because of doubt whether it can be proved in court or fear of the expense of having to do so.” 376 U.S. at 279, 84 S.Ct. 710; see also Gertz v. Robert Welch, Inc., 418 U.S. 323, 342, 94 S.Ct. 2997, 41 L.Ed.2d 789 (1974) (“[T]o assure to the freedoms of speech and press that ‘breathing space’ essential to their fruitful exercise ... this Court has extended a measure of strategic protection to defamatory falsehood.”) (internal citation omitted). Without a federal constitutional requirement focusing on the speaker’s state of mind with regard to the truth of what he was saying (as well as careful scrutiny by the courts of any jury verdict based purely upon speech), the Court concluded, there would be a distinct danger that fear of defamation liability would “dampen[ ] the vigor and limit[ ] the variety of public debate,” to the detriment of First Amendment values. Id. The problem has been treated as one of balancing the very real injury caused by unwarrant*1104ed damage to reputation against the dangers to the system of free expression worked by rules of liability that are easy to misperceive or to misapply in particular instances. And the Court’s answer to this problem has, as noted, been far from unitary. Instead, the balance has been struck with regard to subcategories of defamation cases, according to the nature of the communication, the nature of the parties and, to some degree, the purpose of the speech.3
Our'problem here is similar. Any “true threats” within the three communications at Issue were encased in documents and public events that promoted — at least for those listeners not “in the know” — precisely the kind of “debate on public issues [that] should be uninhibited, robust, and wide-open, and that ... may well include vehement, caustic, and sometimes unpleasantly sharp attacks.... ” New York Times, 376 U.S. at 270, 84 S.Ct. 710. True, the targeted medical professionals and clinics were not public officials, but they were engaged in activities that the defendants, rightly or wrongly, regarded as both morally reprehensible and a matter for eventual governmental proscription through'the political process (presumably through a constitutional amendment). Moreover, as both the majority and Judge Kozinski recognize, the posters and website remained core First Amendment speech even though — quite aside from any coded threat of physical harm — they exposed the targeted plaintiffs to other, nonviolent but still extremely disturbing, interference with their daily lives (in the form of unwanted public exposure and inflammatory rhetoric directed at them, their families, and their customers, both at home and at work) and even if they induced fear in the plaintiffs that people unconnected with the defendants might harm them.4
Under these circumstances, the question for me becomes devising standards that, like the constitutional defamation standards that vary with the strength of the protection of the communication, rely not on an unitary “true threats” standard, as does the majority, but on considerations that lessen the danger of mistaken court verdicts and resulting self-censorship to a greater or lesser degree depending upon the nature of the speech in question and the role of speech of that nature in the scheme of the First Amendment.5
*11053. Some constitutional parameters: Judge Kozinski, in his dissent, makes one important suggestion toward this end with which, for all the reasons already canvassed, I fully agree: He suggests that “statements communicated directly to the target are much more likely to be true threats than those, as here, communicated as part of a public protest.” Kozinski dissent at 7162. As a first cut at separating out the kinds of allegedly threatening communications that are central to First Amendment values and therefore must be tested by particularly stringent criteria before they can be prohibited, these two criteria — the public nature of the presentation and content addressing a public issue (which can include matters of social or economic as well as political import for the individuals involved, see Bartnicki v. Vopper, 532 U.S. 514, 535, 121 S.Ct. 1753, 149 L.Ed.2d 787 (2001); Thornhill, 310 U.S. at 102-03, 60 S.Ct. 736) — are critical.
In a rare instance, a threat uttered in the course of a public political protest might conceivably exceed the bounds of protected speech. United States v. Kelner, 534 F.2d 1020 (2d Cir.1976), is illustrative. (I am not aware of any case in this circuit in which a defendant was, as in Kelner, punished or held liable for a threat uttered in the course of public protest activity — a gap in itself telling with regard to the importance and novelty of this case.) In Kelner, a member of the Jewish Defense League stated at a press conference held in New York just before Yassir Arafat was scheduled to be in the city that
“We have people who have been trained and who are out now and who intend to make sure that Arafat and his lieutenants do not leave this country alive.... We are planning to assassinate Mr. Arafat.... Everything is planned in detail .... It’s going to come off.” Id. at 1021. The press conference was broadcast on television that evening.6 Id. The Second Circuit upheld the defendant’s conviction for uttering the threat, over the objection that the speech was simply an extreme statement of opposition to Mr. Arafat, protected under the First Amendment as hyperbolic public discussion of a public issue. Id. at 1024-28.
In doing so, the Second Circuit recognized that where the asserted threat “is made in the midst of what may be other protected political expression,” courts must be vigilant to permit liability or conviction only in circumstances in which the danger to free expression is minimal; where that is the case, “the threat itself may affront such important social interests that it is punishable.” Id. at 1027. The criteria the Second Circuit suggested to police the dividing line were that “the threat on its face and in the circumstances in which it is made is so unequivocal, unconditional, immediate and specific as to the person threatened, as to convey a gravity of purpose and imminent prospect of execution.” Id. Measured against these criteria, Kelner held that, although politically motivated and designed to convey a public position of protest to Mr. Arafat’s *1106policies, the speech in question was not protected speech. Id. at 1028.
Kelner’s criteria for adjudging the protection accorded alleged threats uttered in the course of public communications on public issues seem appropriate to me— and, as I show below, consistent with NAACP v. Claiborne Hardware Co., 458 U.S. 886, 102 S.Ct. 3409, 73 L.Ed.2d 1215 (1982) — with one exception, an addition, and some explication:
First, the exception: I would not include the imminence or immediacy of the threatened action as a prerequisite to finding a true threat delivered as part of a public speech, if all of the other factors were present. The immediacy requirement calls to mind the standard the Supreme Court erected for proscription of inciting speech in Brandenburg. But as the majority can be read to recognize and as Judge Kozin-ski well explains, the separate constitutional category of unprotected speech for threats does not include statements that induce fear of violence by third parties.
Where there is no threat, explicit or implicit, that the speaker or someone under his or her control intends to harm someone, a statement inducing fear of physical harm must be either (1) a prediction or warning of injury, or (2) an inducement or encouragement of someone else to cause the injury. The former is, as Judge Kozinski suggests, clearly entitled to protection under the First Amendment as either informative or persuasive speech. The latter kind of statement may or may not be protected. Whether it is or not must be governed by the strict inducement standard of Brandenburg if the more than fifty years of contentious, development of the protection of advocacy of illegal action is not to be for naught. See Brandenburg, 395 U.S. at 447-48, 89 S.Ct. 1827 (overruling Whitney v. California, 274 U.S. 357, 47 S.Ct. 641, 71 L.Ed. 1095 (1927), and holding that “the constitutional guarantees of free speech and-free press do not permit a State to forbid or proscribe advocacy of the use of force or of law violation except where such advocacy is directed to inciting or producing imminent lawless action and is likely to incite or produce such action”); see also id. at 450-454, 89 S.Ct. 1827 (Douglas, J., concurring) (recounting the history of the “clear and present danger” doctrine for adjudging the constitutionality of restrictions upon advocacy of illegal action).
One can, however, justify a somewhat different standard for judging the constitutionality of a restriction upon threats than for a restriction upon inducement of violence or other illegal action. There is a difference for speech-protective purposes between a statement that one oneself intends to do something and a'statement encouraging or advocating that someone else do it. The latter will result in harmful action only if someone else is persuaded by the advocacy. If there is adequate time for that person to reflect, any harm will be due to another’s considered act. The speech itself, in that circumstance, does not create the injury, although it may make it more likely. The Supreme Court has essentially decided that free expression would be too greatly burdened by anticipatory squelching of advocacy which can work harm only indirectly if at all. See Kelner, 534 F.2d at 1027 n. 9 (“Short of [advocacy that is close, direct, effective and instantaneous in its impact] the community must satisfy itself with punishment of the one who committed the violation of law or attempted to do so, not punishment of the person who communicated with him about it.”) (quoting Thomas Emerson, The System of Freedom of Expression 404-05 (1970)); see also Ashcroft v. Free Speech Coalition, 535 U.S. -, 122 S.Ct. 1389, 1403, 152 L.Ed.2d 403 (2002) (because “the Court’s First Amendment cases draw vital distinctions between words and deeds,” the *1107government may not punish speech because it increases the chance that someone other than the speaker will commit an unlawful act).
A true threat, in contrast, implies a firmness of purpose by the person speaking, not mediated through anyone else’s rational or emotional reaction to the speech. Threatening speech thereby works directly the harms of apprehension and disruption, whether the apparent resolve proves bluster or not and whether the injury is threatened to be immediate or delayed. Further, the social costs of a threat can be heightened rather than dissipated if the threatened injury is promised for some fairly ascertainable time in the future — the “specific” prong — -for then the apprehension and disruption directly caused by the threat will continue for a longer rather than a shorter period. So, while I would police vigorously the line between inducement and threats — as the jury instructions in this case did,7 although the majority opinion is less clear on this point — I would, where true threats are alleged, not require a finding of immediacy of the threatened harm.
Second, the addition: Although this court’s cases on threats have not generally set any state of mind requirements, I would add to the Kelner requirements for proscribable threats in the public protest context the additional consideration whether the defendant subjectively intended the specific victims to understand the communication as an unequivocal threat that the speaker or his agents or coconspirators would physically harm them.8 Especially where the plaintiffs in such circumstances are relying only on surrounding context and are doing so to overcome the literal import of the words spoken, impairment of free public debate on public issues through self-censorship is a distinct possibility unless there is convincing proof that the literal meaning of the words was not what the defendants intended to convey.
The subjective intent requirement for alleged threats delivered in the course of public protest comports with Supreme Court precedent, both directly and by analogy. Although the Supreme Court has yet to outline fully the constitutional limitations applicable to proscription of threats, in its most- direct look at the subject the Court expressed “grave doubts” that a person could be liable for threatening expression solely on the basis of an objective standard. Watts v. United States, 394 U.S. 705, 707-08, 89 S.Ct. 1399, 22 L.Ed.2d 664 (1969). A few months later, in Brandenburg, the Court held that in an incitement case, the plaintiff or the government must not only prove that a statement “is likely to incite or produce” *1108imminent lawless action, but must also prove that the statement “is directed to inciting or producing” such action. 395 U.S. at 447, 89 S.Ct. 1827. This latter requirement is a subjective intent prerequisite, as it turns the speaker’s liability in an incitement case on how the speaker intends others to understands his words. See also Hess v. Indiana, 414 U.S. 105, 109, 94 S.Ct. 326, 38 L.Ed.2d 303 (1973) (speech cannot be punished when no evidence exists that “words were intended to produce” imminent disorder).
With regard to this subjective intent requirement, there is no meaningful distinction between incitement cases and threat cases such as this one — that is, cases involving public protest speech, especially where the alleged threat, on its face, consisted entirely of advocacy. The First Amendment protects advocacy statements that are likely to produce imminent violent action, so long as the statements are not directed at producing such action. To do otherwise would be to endanger the First Amendment protection accorded advocacy of - political change by holding speakers responsible for an impact they did not intend.
Similarly, a purely objective standard for judging the protection accorded such speech would chill speakers from engaging in facially protected public protest speech that some might think, in context, will be understood as a true threat although not intended as such. Unsure of whether their rough and tumble protected speech would be interpreted by a reasonable person as a threat, speakers will silence themselves rather than risk liability. Even though the Supreme Court has stated that protected political speech “is often vituperative, abusive, and inexact,” speakers wishing to take advantage of these protected rhetorical means may be fearful of doing so under the majority’s purely objective approach. Watts, 394 U.S. at 708, 89 S.Ct. 1399; see also Rogers v. United States, 422 U.S. 35, 47, 95 S.Ct. 2091, 45 L.Ed.2d 1 (1975) (Marshall, J., concurring).9
*1109When the district court issued the injunction against the defendants, the court, for reasons it does not explain, relied on a different definition of threat than the one it instructed the jury to use. In contrast to the definition relied upon by the jury, the definition used for purposes of the injunction correctly incorporated the subjective intent requirement mandated by the First Amendment. Planned Parenthood of the Columbia/Willamette, Inc. v. American Coalition of Life Activists, 41 F.Supp.2d 1130, 1155 n. 1 (D.Or.1999). In addition, the district court found that the defendants did intend to threaten. As a result, the injunction comes close to conforming on its face to the dictates of the First Amendment. The injunction still falls short, however, because the district court did not state that a threat must be unequivocal, nor did it find the posters to be unequivocal threats. As I explain below, any definition of threats that does not include the unequivocal requirement provides too little protection for public political speech.10
Third, the explication: “Unequivocal” cannot mean literal: Ryder trucks, in the United States in the 1990s, and burning crosses, in the United States in the twentieth and twenty-first centuries, have unambiguous meanings that the individuals targeted will be hurt (at least unless they do what the perpetrator of the threat wants them to do, whether it be stop performing abortions or move out of town). Instead, “unequivocal” means to me unambiguous, given the context. As such, the requirement is essentially a heightened burden of proof, requiring that a threatening meaning be clearly and convincingly apparent. And in determining whether that proof standard has been met, I would continue to apply the objective standard the majority embraces, based on our cases, in determining whether the speech in fact communicates an intent to harm specific individuals.
*1110This case, I repeat, is uniquely difficult because to perceive a threat, one must disregard the actual language used and rely on context to negate the ordinary meaning of the communication. Further, the actual language is, in its own right, core First Amendment speech, speech that to a naive reader communicates protected information and ideas. So the crux of the plaintiffs’ cause of action (once one accepts that only statements that evince an intention by the speaker or his or her agents to carry out the threat can be actionable) is really an assertion that the defendants were using Aesopian language or could be understood as doing so, and that the context in which the speech must be viewed provides the necessary evidence of the defendants’ true, albeit coded, meaning.11
The first set of contextual evidence involves the poster / — murder / poster / murder pattern the majority principally relies upon. Had the murders — or any murders, or any serious violence — been committed by the defendants and had the plaintiffs known that, the inference from the poster/murder pattern that the publication by them of posters similar to those previously followed by a murder might be a strong one.12 The inference would be stronger had the defendants also put out the earlier posters and had the plaintiffs known that. Neither is the case.
Plaintiffs’ main submission to fill this gap was extensive evidence concerning the defendants’ opinions condoning the use of violence against medical professionals who perform abortions, including general statements to that effect and particular statements concerning the people who murdered the doctors depicted on the previous posters, stating that their actions were justified and that they should be acquitted. Plaintiffs’ closing argument, for example, went on for pages and pages about defendants’ meetings and writings concerning the “justifiability of the use of force.”
This evidence is certainly of some pertinence as to what the defendants may have intended to do.13 It is more likely that someone who believes in violence would intentionally threaten to commit it. It is also pertinent to what persons in the plaintiffs’ position-that is, persons involved in the abortion controversy and alert to the division of opinion within it — would likely understand concerning defendants’ communication. Individuals who believe in violence are not only more likely to threaten to commit it but also actually to commit it, and so defendants’ views might well influence plaintiffs’ perception of their speech. And since the defendants would know that, defendants’ public statements approving the use of violence against doctors who perform abortion are relevant to whether reasonable speakers in defendants’ position would expect their communications to be understood as threats.
*1111At the same time, heavy reliance on evidence of this kind raises profound First Amendment issues of its own. One can not read plaintiffs’ closing argument in this case without fearing that the jury was being encouraged to hold the defendants liable for their abstract advocacy of violence rather than for the alleged coded threats in the posters and website, the instructions to the jury to the contrary notwithstanding. And while advocacy evidence may make both an intent to threaten and a perception that there was a threat more likely, that is not unequivocally so. People do not always practice what they preach, as the stringent incitement standard recognizes. If we are serious about protecting advocacy of positions such as defendants’ sanctioning of violence, as we are constrained to be, then permitting that protected speech to be the determinative “context” for holding other facially protected, public protest speech — the posters and website in this case — to be a “true threat” seems to me simply unacceptable under the First Amendment.
Finally, I note that the approach I’ve outlined here fully comports with Claiborne Hardware. Claiborne Hardware applied an “extreme care” standard in determining “liability on the basis of a public address-which predominantly contained highly charged political rhetoric.” 458 U.S. at 926-927, 102 S.Ct. 3409. It went on to note that “[i]n the passionate atmosphere in which the speeches were delivered, they might have been understood as inviting an unlawful form of discipline or, at least, as intending to create a fear of violence whether or not improper discipline was specifically intended.” Id. at 927, 102 S.Ct. 3409 (emphasis added). After reviewing the actual words used in context, however, the Court concluded that “Evers’ addresses did not exceed the bounds of protected speech.” Id. at 929, 102 S.Ct. 3409. As I read the opinion, it held, essentially, that the supposed threats were not on their face unequivocal and were not made unequivocal by any contextual factors. So here.
I would therefore hold that under the special rules I would apply to public protest speech such as that in this case, plaintiffs’ judgment cannot stand because, after a proper review of the record, we would have to conclude that there was no unequivocal, unconditional14 and specific threat.15
II
Federal Law Enforcement Officers’ Testimony Regarding Threats
I also disagree with the majority’s conclusion that permitting law enforcement officers to testify as to their opinions about the meaning and import of the posters at issue was within the district court’s discretion. The government may not seek to persuade a jury that certain speech contains characteristics that place it outside the realm of constitutionally protected speech by providing in testimony, as opposed to in a criminal indictment, its “nonjudicial determination” on the ultimate le*1112gal issue to be decided. Hill v. Rolleri, 615 F.2d 886, 890 (9th Cir.1980).
The district court permitted the officers to repeat in testimony the warnings that the officers gave the plaintiffs after the release of the posters, purportedly in order to show the plaintiffs’ state-of-mind in response to the posters. Under this rationale, the testimony had very little, if any, relevance to the issues before the jury, and, especially in light of the First Amendment concerns the testimony raises, the resulting prejudice greatly outweighed its minimal probative value. The district court, therefore, abused its discretion in failing to exclude this testimony under Fed.R.Evid. 403.16
1. The testimony: The district court permitted the plaintiffs to call to testify an FBI agent, with 25 years experience, and two United States Marshals, one with 26 years and the other with 14 years experience. The following testimony was given: “I told her that I was in receipt of threat information in the form of a flyer.”
“That I had received a copy of a list called the Deadly Dozen List, which listed 13 doctors, who perform abortions, and that it was threatening in nature
“I told her that I thought that her teachers, the teachers of her children, should know about this threat, as well, in order to maintain the security of the children.” “I told her that the children should be aware of — of the threat.”
“I told him that he was on a threat list
“I told her that if she received additional threats or wanted protection, these were the numbers to contact .... ”
“And we discussed the reasons we believed that the threat was serious .... We discussed the escalation in the incidents over the prior couple of years. We talked about the murder of Dr. Gunn in Florida. We talked about shootings involving Dr. Tiller in Kansas. We talked about shootings involving Dr. Christ. We talked about Michael Bray and his affiliation with the American Coalition of Life Advocates.”
“Well, because of the nature of the threats, and — I asked Dr. Hern to — he had a bulletproof vest. I thought it would be a good thing if he wore that.”
(emphases added). The testimony not only revealed the individual law enforcement officers’ opinions of the meaning of the posters, but also informed the jury about the opinion of “headquarters,” as follows:
“I told him ... that I had been given instructions to notify — to immediately notify him, so that he could take some personal precautions for his safety.”
“I was contacted by my headquarters in Washington, D.C. ... I advised her that the Marshal’s Service was offering her protection, because her name appeared on the list, and stated that if she wanted protection, I would forward the request to our headquarters, who would then forward it to the Department of Justice.”
“I was directed by my headquarters to immediately contact Dr. Warren Hern, because he was listed on the — the document. But, additionally, I was directed to contact all of the clinics in the district of Colorado. ” The officer further testified that he did contact all of the clinics in Colorado.
*1113 “[HJeadquarters was taking this threat very seriousily.”
“I explained to Dr. Hern that Michael Bray had been a conspirator in — or involved in a conspiracy to blow up several abortion clinics. And because of his affiliation, in addition to the other things we discussed, that toy headquarters believed that this ivas [a] serious threat, and something that — that we had to act on immediately.”
(emphases added).
My major concern here involves the First Amendment repercussions of allowing testimony by government employees as to the government’s opinion concerning whether speech is outside the First Amendment’s protections. In keeping with traditional Rule 403 analysis, however, I first explain why the testimony did not serve to elucidate any of the issues properly before the jury and then turn to the prejudicial effect the testimony had on the defendants’ First Amendment rights.'
2. Basic Rule J.03 analysis: The majority holds the law enforcement testimony probative because it has “some tendency to show the physicians’ state of mind when they found out they were named on ‘wanted’' — type posters....” Majority Op. at 1082. Under the definition of a true threat that the majority uses (and under the one I would adopt, see Part I, supra) the plaintiffs’ state of mind is relevant only to the extent that it tends to show “whether a reasonable person would foresee” that the plaintiffs would interpret the posters as threats. Majority Op. at 1074. The officers’ testimony concerning the warnings muddled rather than illuminated the inquiry into the question how a reasonable lay person would understand the posters, as that testimony revealed the officers’ reaction to the posters, not the plaintiffs’. The true threat standard focuses on how “those to whom the maker communicates the statement” would “interpret[ ]” it, not on the government’s determination of whether a threat was made. See id. So the officers’ reaction to the posters is largely irrelevant. Further, to the extent the testimony did tend to show the plaintiffs’ state of mind, it suggested what the plaintiffs’ reaction may have been to the officers’ warnings or to the combination of learning about the posters and receiving the warnings, not simply to the posters themselves.
During the testimony-of one of the officers, the district court instructed the jury to consider the testimony only for what it revealed about the state of mind of the recipient of the warnings and not to take the testimony as an “administrative decision” that the posters constituted true threats. To the extent that the officers’ testimony did bear on any pertinent issue — which, as I indicated above, is little if at all — the court’s limiting instruction did not do much to maintain the jury’s focus on this issue, as the court did not repeat the instruction when each of the law enforcement officers testified, nor did the court instruct the jury on this issue before deliberations, despite the defendants’ request that the court do so.
It is unlikely that a jury can put aside the opinions of an FBI agent and United States Marshals — and their headquarters — as to the nature of the speech and instead focus solely on how those opinions bore on the plaintiffs’ state of mind. See United States v. Gutierrez, 995 F.2d 169, 172 (9th Cir.1993) (“[T]he expert testimony of a law enforcement officer ... often carries an aura of special reliability and trustworthiness.”) (quoting United States v. Espinosa, 827 F.2d 604, 613 (9th Cir.1987)). On traditional evidentiary grounds alone, such testimony should not be admitted'in threats cases.
3. First Amendment-related prejudice: Turning now to the issue I find most trou*1114bling, the First Amendment ramifications of the law enforcement officers’ testimony:
Admitting testimony by law enforcement officers as to whether certain speech has the primary characteristic of an unprotected category (for instance, is a serious threat, or is obscene, or is false) allows the government not only to prohibit or burden that category of speech (true threats, obscenity, defamation), but also persuasively to shape the jury’s determination of what speech falls into the unprotected category. The obvious risk is that the government will use its “aura of special reliability and trustworthiness,” Gutierrez, 995 F.2d at 172, to describe as undeserving of constitutional protection speech that in fact is only unpopular with the government. In Watts, the Court looked to the reaction of those to whom the speech was directed to determine how the speech should be taken. 394 U.S. at 708, 89 S.Ct. 1399. Had the Secret Service run to the President to inform him of Watts’ speech and warn him of the “threat,” the Secret Service’s reaction, and the President’s resulting fear, presumably would not have been allowed to override the reaction of the actual audience to the speech.
Furthermore, the officers’ testimony here quite naturally tended to blur the lines between various categories of speech — true threats, incitement, and perhaps some form of “putting in harm’s way” — and therefore risked a jury finding of liability for speech that may not fall within the “threat” category as narrowly defined for First Amendment purposes. The officers testified that they told the plaintiffs the speech was a threat and one that should be taken seriously, but there is no indication that the officers distinguished between a “true” threat — a threat of violence by the speaker — and speech warning that a third party would harm the plaintiffs or speech containing a threatening quality because of its tendency to incite others or to put the plaintiffs in harm’s way. Nor did the district court instruct the jury that the officers might use the term “threat” in a way that differed from the type of “threat” that does not receive constitutional protection.
The majority also concludes that the district court properly admitted the officers’ testimony “to show the knowledge and intent of ACLA in distributing the posters regardless of the reaction they precipitated.” Majority Op. at 1082. Testimony as to the statements made by the officers to the plaintiffs has little relevance to the intent and knowledge of the defendants. And, more importantly, the same First Amendment concerns come into play here: Under this rationale, if federal law enforcement officials dislike certain speech, they can take a substantial step towards rendering it unprotected by expressing publicly the view that such speech is threatening, because if the speaker then repeats the speech he does so with knowledge of the reaction it precipitated.
To the extent that our law allows law enforcement officers otherwise to testify directly on ultimate factual and legal questions that the jury must decide, we should draw the line at permitting the use of this persuasive aura in testimony that certain speech is of such a nature that it is undeserving of constitutional protection. Permitting such testimony cannot be reconciled with the role of the First Amendment to protect freedom of speech from suppression by the government.
Ill
Deposition Summaries
The majority approves — quite in passing — the district court’s insistence that the parties submit as evidence summaries of deposition testimony, not the testimony itself. Majority Op. at 1083. As I read *1115Federal Rule of Civil Procedure 32, which governs the admission of deposition testimony, it does not permit the substitution of summaries for actual testimony. Nor is there anything in the Federal Rules of Evidence or this court’s case law to the contrary. Rather, it is a fundamental precept of our system for ascertaining facts that a jury is entitled to learn what a witness actually said, rather than an inexact rendition presented by counsel (and probably initially drafted by paralegals).
Language can be subtle, ambiguous and malleable. Paraphrases, as any judge reading lawyers’ briefs knows, are no substitute for quotation of the actual words spoken by a witness. As often as not, a check of the transcript will reveal that the language the witness actually spoke, in context, may well mean something other than what counsel has represented.
That does not mean that counsel is lying, but that shades of meaning can be critical. “[A]s the childhood game of ‘telephone’ well demonstrates, words change significantly in the course of their re-telling by third parties.” United States v. Pena-Espinoza, 47 F.3d 356, 364 (9th Cir.1995) (Reinhardt, J., dissenting). Indeed, the game of “telephone” requires only that a listener repeat the exact language that he or she heard; a summary, in contrast, necessarily requires the more subjective choice of different words to convey the general idea communicated by the original language. “There is simply no way to summarize the contents of a transcript without offering to some degree a subjective view of their meaning and import.” Id. Because that is so, summaries of witnesses’ testimony are likely to distort the import of the actual testimony given and so impede the jury’s search for truth.
Our legal system recognizes, in various contexts, that the same set of words may frequently lend itself to more than one reasonable interpretation. See, e.g., Chevron U.S.A., Inc. v. Natural Res. Def. Council, Inc., 467 U.S. 837, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984). There is no reason to believe that a lawyer will not adopt the interpretation most favorable to his or her client, so long as the interpretation is reasonable, even if not perhaps the most reasonable. See United States v. Leon-Reyes, 177 F.3d 816, 820 (9th Cir.1999) (“Summaries are normally prepared by an interested party and therefore may not be completely accurate or may be tainted with the preparing party’s bias.”). In our adversary system, it is the role of the trier of fact — not the advocate — ultimately to determine the meaning of witness testimony.
' Further, access to the actual language a witness used — even on a cold record — -is often essential to determining the witness’s credibility and hence the weight, if any, to be accorded the testimony. Equivocations, hesitations, and internal contradictions may all be smoothed over by summaries that purport to extract the content of a witness’s testimony. Requiring counsel to summarize testimony without allowing the trier of fact to have access to the testimony itself necessarily precludes the trier of fact from' properly exercising his or her truth-determining role.
The record here provides concrete examples of various ways in which summaries can distort the import of the actual testimony and thereby impair the truth-ascertaining process. For instance, the summary of Michael Dodds’ deposition condensed inaccurately the testimony he gave. The summary stated:
The other physician, on the Deadly Dozen List, from Dodds’ region, Dr. Douglas Karpen, is from the Houston, Texas, area. Dodds believes that defendant Donald Treshman provided that name.
What Dodds actually said in his deposition regarding the source of Karpen’s name for *1116the Deadly Dozen list was “1 don’t know.”17
The deposition of Roy McMillan provides an example of testimony that could reasonably be interpreted in either of two ways, but the summary provided the jury with only one interpretation. The summary stated:
As for the additional murder of Mr. Barrett [Dr. Britton’s escort], McMillan felt that if it was, quote, right for one person, it would be right for someone else, end quote.
A look at McMillan’s deposition transcript (which the defendants introduced in rebuttal) sheds a somewhat different light on the quotation included in the summary. In his deposition, McMillan was first asked about a petition in support of Michael Griffin, who killed Dr. Gunn, and then asked about “the second petition which was for Mr. Hill,” who killed Mr. Barrett and Dr. Britton. The testimony went as follows:
[Answer:] This is identical — pretty much identical to the one that was circulated about the first abortionist’s termination. And this was — this, the second one was regarding Paul Hill.
Question, and this one was put out by Michael Bray, is that right?
Answer, I am not sure who put it out, but I concurred that if it was right for one person, it would be right for someone else.
Thus, it appears that McMillan likely meant that if a petition in support of Griffin was right, so too was a petition in support of Hill. Either way, the interpretation should have been left entirely to the jury. (It is quite unlikely that this difference in meaning could have substantively affected the verdict, but that conclusion would require a separate inquiry.)
Finally, the record here also contains summary language that although technically accurate may nonetheless have conveyed a subtly different, but potentially important, sense of the speaker or of the events described from the testimony itself. The summary of Dawn Stover’s deposition began with the sentence:
Dawn Stover is the associate director of defendant Advocates for Life Ministries ....
Here is the excerpt from Stover’s deposition transcript:
Question, are you still the associate director of Advocates for Life Ministries? Answer, I would guess that, but I have been inactive for so long that — I am still affiliated. I still talk to a couple of people in Advocates, but I don’t do any directing and haven’t done any directing for years. So — and having a title has never been that big of an issue. Question, was that ever a paid position? Answer, no. Question, so you don’t know whether or not your status is currently associate director in terms of the eyes of the organization? Answer, I honestly don’t know how they would perceive me as. I don’t know, just because I have been inactive for so long, but they may still.
Certainly, the jury reasonably could have found from Stover’s testimony that she “is the associate director” of Advocates for Life Ministries, as the summary stated. At the same time, however, the jury might have considered Stover’s current role in the organization as quite different depending on which of the above versions of her testimony they heard. One set of words rarely conveys precisely the same meaning as a second, truncated version.
The majority today pays no heed to all these “dangers of witnesses summarizing *1117oral testimony.” United States v. Baker, 10 F.3d 1374, 1412 (9th Cir.1993), overruled on other grounds by United States v. Nordby, 225 F.3d 1053, 1059 (9th Cir.2000), and instead notes, without qualification, only that the presentation of deposition testimony “in the form of summaries was within the court’s discretion under Rule 611(a),” Majority Op. at 1083. The very first mandate of Rule 611(a), however, requires the trial court to “make the interrogation and presentation [of evidence] effective for the ascertainment of truth.” Fed.R.Evid. 611(a)(1).18 For all the reasons just discussed, substituting summaries of testimony for a word-for-word transcript itself can hardly serve as an “effective” mode “for the ascertainment of truth.” See id.
Moreover, it is Fed. R. Civ. Proc. 32, not Rule 611(a) of the Rules of Evidence, that directly, and with particularity, governs the presentation of deposition testimony. As I read it, Rule 32 decidedly does not permit courts to authorize the use of summaries in place of actual testimony.
Rule 32 begins with this general provision:
At the trial ... any part or all of a deposition, so far as admissible under the mies of evidence applied as though the witness were then present and testifying, may be used....
Rule 32(a) (emphasis added); see also Rule 32(b) (“[(Objection may be made at the trial or hearing to receiving in evidence any deposition or part thereof for any reason which would require the exclusion of the evidence if the witness were then present and testifying.”). A witness who is “present and testifying” is doing just thaWTestifying,”' not providing capsule versions of his or her testimony. And using “any part or all of a deposition” does not equate to using a paraphrased, condensed version of a deposition, any more than a course syllabus directing students to read “Hamlet” intends to subsume within that directive the Classic Comics version of “Hamlet.”
Rule 32 also specifically addresses the “Form of Presentation” of deposition testimony, giving no indication that a district court may admit a summary of deposition testimony in lieu of the testimony itself. The pertinent section states, in relevant part:
Except as otherwise directed by the court, a party offering deposition testimony pursuant to this rule may offer it in stenographic or nonstenographic form, but, if in nonstenographic form, the party shall also provide the court with a transcript of the portions so offered.
Rule 32(c) (emphasis added). The rule therefore clearly anticipates the admission of stenographic or nonstenographic forms of testimony, not summaries.19 Although *1118Rule 32(c) does apply “[ejxcept as otherwise directed by the court,” this caveat is most sensibly read to give the court discretion to direct either stenographic or non-stenographic presentation of deposition testimony, not to permit the presentation as “evidence” of summaries that approximate but do not reproduce the language the witness used in any form.
Bolstering this conclusion regarding Rule 32(c), Rule 28 provides for the taking of depositions in foreign countries “pursuant to a letter of request,” and expressly grants the district court the discretion to admit the response to such a letter even if it is not a “verbatim transcript” of the testimony or if it exhibits “any similar departure from the requirements for depositions taken within the United States under these rules.” See Rule 28(b). The assumption quite obviously underlying Rule 28 is that any report of testimony other than a “verbatim transcript” is a “departure from the requirements for depositions taken within the United States under these rules.”
More generally, Rule 32 demonstrates an overall preference for the presentation of testimony in the manner that, to the extent practical, best provides the jury with complete information concerning the witness’s demeanor. Rule 32(a), for example, clearly favors live testimony over deposition evidence by limiting the use of depositions to three situations: when an adverse party is the deponent; for impeachment purposes; or when the deponent is not available to testify at trial. Rule 32(a)(l)-(3). By so doing, the rule reflects the historical belief that live testimony better enables the jury to adjudge the credibility of a witness and therefore to determine the weight and import ascribed to the witness’s testimony. Deposition testimony is itself only second-best.
When the rules do allow the admission of deposition testimony in a jury trial, Rule 32(c) permits a party in some instances to insist upon the presentation of testimony in “non-stenographic form,” allowing the jury to hear and/or see the testimony as it was given. Rule 32(c) (“On request of any party in a case tried before a jury, deposition testimony offered other than for impeachment purposes shall be presented in nonstenographic form, if available, unless the court for good cause orders otherwise.”) (emphasis added). Rule 32(c), by favoring audio and video recordings over the reading of a cold transcript, therefore establishes a preference for testimony that is the most like live testimony. Under this scheme, the presentation of deposition testimony in stenographic form is third-best.
Such presentation, however, at least allows the jury to hear or read the actual words used by the witness. Deposition summaries, unless accompanied by a transcript of the testimony, deprive the jury of even this opportunity. With Rule 32’s clear preference for live testimony, or for testimony most resembling it, it makes little sense to think the rule tacitly allows for this new, fourth-best, form of evidence, so far removed from the in-person live testimony for which it is a substitute. I conclude that Fed. R. Civ. Proc. Rule 32 withholds from district courts the authority to require the substitution of summaries of deposition testimony for the testimony itself, where truth and falsity are at issue, and that the general language of Fed. *1119R.Evid. Rule 611 cannot override that determination.
There is nothing in our case law to the contrary. We have, while expressing great caution, allowed summaries of evidence in narrow circumstances, but never as a complete substitute for actual transcripts on material matters of historical fact.
For instance, in Leon-Reyes, 177 F.3d at 820, a perjury case, we held that the district court did not abuse its discretion in allowing the use of summaries of testimony from a prior trial in which the defendant had allegedly committed perjury, emphasizing that the district court instructed the jury to consider the summaries only for determining the materiality of the false statements and not for the truth of the witnesses’ underlying testimony. Pena-Espinoza, 47 F.3d at 360, permitted — although making clear that it did “not wish to condone such procedures” — admission of summaries prepared by the prosecution of telephone call transcripts. The court specifically noted that:
The transcripts themselves were in evidence and the jury had them to examine during deliberations; the ruling expressly permitted defense counsel to require a reading of the full transcript on cross-examination and to dispute the veracity of the readers’ summaries.
Id. Similarly, Baker, 10 F.3d at 1411-12, found district court discretion pursuant to Fed.R.Evid. 611(a) to permit a government witness to present summary testimony and a chart estimating, on the basis of testimony at trial, the value of narcotics transactions. Critically, the district court made clear to the jury that the testimony and chart did not constitute evidence:
The [district] court instructed the jury that the summary testimony and exhibits were not evidence, did not represent an opinion of the court or the prosecution on the credibility of witnesses, and were to be disregarded to the extent the jury found them conflicting with the testimony and evidence received at trial.
Id. at 1411. As in Pena-Espinoza, the Baker court emphasized that this court is “not blind to thé dangers of witnesses summarizing oral testimony” and that “such summaries should be admitted under Rule 611(a) only in exceptional cases.” Id. at 1412.
Thus, when this court has upheld the admission of summary evidence under the abuse of discretion standard, we have done so not as a substitute for transcript evidence on matters of historical fact, but either on issues other than the truth of the matter testified to or as an assistance to the jury, while also including the actual transcripts in the record for the use of the jury or reviewing courts. And none of our cases discuss the provisions of Fed. R. Civ. Proc. Rule 32, because none of them involved deposition summaries as opposed to summaries of other forms of evidence.20
This is not a case in which the parties reached any agreement as to the summaries presented, so I do not consider whether such an agreement would be. permissible. Nor did the defendants agree to the use of summaries at all; instead, they maintained a continuing objection to this procedure. And the district court did not review or *1120revise the summaries after receiving objections prepared by the defendants, as it had originally planned to do. Cf. Leon-Reyes, 177 F.3d at 820 (“Summaries ... must be scrutinized by the trial court to ensure that they are accurate, complete, not unduly prejudicial, limited to the relevant issues, and confined by appropriate jury instructions.”).
The district court did allow the defendants to present counter-summaries, colored with argument, and, in rebuttal, to introduce excerpts from the transcripts. Such an adversarial procedure, however, does not ensure that the jury will have before it the evidence necessary to informed decision-making. The party responsible for summarizing the testimony may have little reason to move for the admission of the underlying testimony, precisely because that party may prefer its summary to the testimony itself. Likewise, the adverse party will, hopefully, point out blatant inconsistencies between the summary and the testimony, but may choose otherwise to avoid providing the jury with testimony that largely supports a summary introduced by the other side.
Nevertheless, the fact remains that the jury must have the opportunity to review the actual evidence — -the transcripts of the testimony — when deliberating as to the meaning of testimony. It is nonsensical to expect the jury to determine the credibility of witnesses and testimony, the special province of the jury, without providing the jury with access to that testimony. Just as we, as judges, do not read attorneys’ paraphrases of statutes when we try to discover what the legislature meant, see Fed. R.App. Proc. Rule 28(f), jurors cannot sensibly evaluate the meaning and credibility of words without knowing what those words are.
One final note: The majority presumably finds that the district court has the discretion under Rule 611(a) to require deposition summaries in lieu of the testimony itself in order to “avoid needless consumption of time.” Rule 611(a)(2). Because the presentation of deposition summaries, without the agreement of the parties and the admission of the corresponding excerpts, is not “effective for the ascertainment of truth,” Rule 611(a)(1), the consumption of time caused by the presentation of actual testimony is not “needless.” Moreover, by providing an additional issue for the parties to dispute, the use of summaries is just as likely to increase as to decrease the time spent by counsel and by the court.
I recognize that district courts can and should reasonably limit the amount of time expended on the presentation of deposition testimony. This authority does not, however, give trial courts the discretion to replace such testimony entirely with a Reader’s Digest Condensed Books version.21
IV.
Conclusion
As waves of fervent protest movements have ebbed and flowed, the courts have been called upon to delineate and enforce the line between protected speech and communications that are both of little or no value as information, expression of opinion or persuasion of others, and are of considerable harm to others. This judicial task has never been an easy one, as it can require — as here — recognizing the right of *1121protesting groups to question deeply held societal notions of what is morally, politically, economically, or socially correct and what is not. The defendants here pose a special challenge, as they vehemently condone the view that murdering abortion providers — individuals who are providing medical services protected by the Constitution — is morally justified.
But the defendants have not murdered anyone, and for all the reasons I have discussed, neither their advocacy of doing so nor the posters and website they published crossed the line into unprotected speech. If we are not willing to provide stringent First Amendment protection and a fair trial to those with whom we as a society disagree as well as those with whom we agree — -as the Supreme Court did when it struck down the conviction of members of the Ku Klux Klan for their racist, violence — condoning speech in Brandenburg■ — the First Amendment will become a dead letter. Moreover, the next protest group — which may be a new civil rights movement or another group eventually vindicated by acceptance of their goals by society at large — will (unless we cease fulfilling our obligation as judges to be evenhanded) be censored according to the rules applied to the last. I do not believe that the defendants’ speech here, on this record and given two major erroneous evi-dentiary rulings, crossed the line into unprotected speech. I therefore dissent.
. In so stating — and elsewhere in this opinion — I do not address the constitutional viability of a cause of action for putting another in harm's way by publicizing information that makes it easier for known or suspected potential assailants to find an intended victim. There was no such cause of action in this case, as Judge Kozinski observes, and I express no view upon whether or under what circumstances such a cause of action could be stated under the law, including under the First Amendment.
. Similarly, the First Amendment's over-breadth doctrine extends some protection to speech that is without First Amendment value in order to limit self-censorship of speech that does possess this value. See Massachusetts v. Oakes, 491 U.S. 576, 581, 109 S.Ct. 2633, 105 L.Ed.2d 493 (1989).
.See New York Times, 376 U.S. at 279-80, 84 S.Ct. 710 (a public official may not recover damages "for a defamatory falsehood relating to his official conduct unless he proves that the statement was made with 'actual malice’ — that is, with knowledge that it was false or with reckless disregard of whether it was false or not”); Curtis Publ'g Co. v. Butts, 388 U.S. 130, 164, 87 S.Ct. 1975, 18 L.Ed.2d 1094 (1967) (opinion of Warren, C.J., concurring in the result) (New York Times standard applies to defamation cases brought by public figures); Gertz, 418 U.S. at 347-49, 94 S.Ct. 2997 (New York Times standard not required for cases brought by private figure plaintiffs; instead, the states may only not "impose liability without fault” for the defamation of a private figure plaintiff — although a different standard may apply if "the substance of the defamatory statement [does not] make[ ] substantial danger to reputation apparent” — but the states "may not permit recovery of presumed or punitive damages” without proof of actual malice); Dun & Bradstreet, Inc. v. Greenmoss Builders, Inc., 472 U.S. 749, 761, 105 S.Ct. 2939, 86 L.Ed.2d 593 (1985) (opinion of Powell, J.) (states may allow private figure plaintiffs to recover, without proof of actual malice, presumed or punitive damages for defamatory speech "involving no matters of public concern.”).
. I discuss below the constitutional importance of the latter requirement — that any proscribed threat communicate the intention of the speaker or his or her agents.
. I note that there is one way in which the speech here differs from defamation: False, defamatory speech, even on matters of public concern, does not have any significant First Amendment value. R.A.V., 505 U.S. at 383, 112 S.Ct. 2538. Although "true threats” also lack such value as a general matter, a "true *1105threat” that includes only facially-protected speech nonetheless does have First Amendment value, because it not only is threatening but also has another meaning — the literal, facially-protected meaning — which here falls within the heart of First Amendment speech. For this and other reasons, the categories of defamatory speech and the rules applicable to them cannot rigidly determine the analysis applicable in threats cases.
. During the press conference, Mr. Kelner, the defendant, "was seated in military fatigues behind a desk with a .38 caliber 'police special' in front of him,” next to “another man ... dressed in military fatigues.” Id. The gun and uniform seem to me simply a prop ánd costume designed to enhance the communication of seriousness of purpose, not proof that the defendant was involved contemporaneously in actual violence.
. The jury in this case was instructed that “the mere abstract teaching of the moral propriety or even moral necessity for a resort to force and violence is protected speech under the First Amendment,” and that "[y]ou are not to consider any evidence that the three statements allegedly 'incite' violence against plaintiffs.”
. In Kelner, the jury was instructed that it needed to find that "Kelner 'intended the words as a threat against Yassir Arafat and his lieutenants.’” 534 F.2d at 1025. The Kelner court rejected a different intent requirement, namely, an intent to carry out the threatened action. Id. at 1025-27. The majority erroneously concludes that the dissenters would require that to find a "true threat,” the speaker must have had the latter sort of intent — that is, the speaker must “actually intend to carry out the threat.” See Majority Op. at 1075. To the contrary, there has been no intimation in either dissent that the speaker need have the intention, or the ability, to do so. Rather, I propose the inclusion of a "specific intent” requirement with regard to the speaker’s intent to threaten-thal is, a requirement that the judge or jury determine whether the speaker intended to place the listener in fear of danger from-the speaker or his agents.
. The majority opinion does not appear to embrace any such subjective intent standard as a constitutional requirement, but does suggest that any such requirement was met here through the instruction on FACE’S statutory elements. The jury, however, was specifically instructed several times, quite emphatically, that there is no subjective intent requirement in adjudging whether or not a statement is a “true threat.” In closing argument, counsel for plaintiffs also informed the jury that subjective intent was not relevant. Further, in a separate instruction devoted to the case's various intent issues, the trial judge reminded the jury that it did have to find some form of intent when it considered the RICO charges against the defendants and when it considered punitive damages, but did not state that the jury had to find intent when considering whether the defendants violated FACE.
The very next instruction concerned the FACE cause of action and stated, ungrammatically and nearly incoherently, that the plaintiffs must prove that the defendant "made the threat of force to intimidate ... the plaintiff's ... ability to ... provide reproductive health services.” Although "intimidate” was defined, correctly, as "place a person in reasonable apprehension of bodily harm,” the FACE instruction left out the statute’s clear motive or purpose requirement — that to be liable the defendant must act "because that person is or has been, or in order to intimidate such person ... from ... providing reproductive health services,” 18 U.S.C. § 248(a)(1), substituting a confused and confusing locution. A jury specifically and repeatedly admonished not to take into account the defendants’ subjective intent would not likely understand the obscure FACE instruction as a requirement that it should do so.
There was also a RICO cause of action with separate elements. While the RICO instruction required,, in addition to a true threat, an "intent of depriving a plaintiff of his or her ... protected right to provide abortion services,” that is not equivalent to requiring an intent to communicate that the speaker or his or her confederates would physically injure the plaintiff.
. Part of the injunction fails to comply with the First Amendment for an additional reason. The injunction not only prevents the defendants from further distributing the posters, it also prevents them from possessing the posters or their equivalents. In effect, this latter part of the injunction regulates the type of written materials that the defendants may possess in the privacy of their homes, directly contradicting the Supreme Court’s holding in Stanley v. Georgia, 394 U.S. 557, 89 S.Ct. 1243, 22 L.Ed.2d 542 (1969):
If the First Amendment means anything, it means that a State has no business telling a man, sitting alone in his own house, what books he may read or what films he may watch. Our whole constitutional heritage rebels at the thought of giving government the power to control men’s minds.
Id. at 565; see also Free Speech Coalition, 122 S.Ct. at 1403 ("First Amendment freedoms are most in danger when the government seeks to control thought.... ”).
The majority nonetheless upholds the injunction against possession of the posters, distinguishing Stanley on the basis that "the posters in this case are quite different from a book; the 'wanted' — type posters themselves — not their ideological content — are the tool for threatening physicians. In this sense, the posters’ status is more like conduct than speech.” Majority Op. at 1087. But, whatever else they may be, the posters are speech-— they express ideas through the use of words and pictures.
The majority also asserts that the "First Amendment interest in retaining possession of the threatening posters is de minimis, while ACLA’s continued possession of them constitutes part of the threat.” Id. at 1087. This summation ignores the fact that the posters do have First Amendment value — they express ideas about one of the most contentious political and moral issues of our time. And confined to the defendants' homes, the posters do not place anyone in apprehension of danger or disrupt their lives; they only influence "the moral content of [the defendants’] thoughts.” Stanley, 394 U.S. at 565, 89 S.Ct. 1243. As such, the First Amendment value of the posters is not outweighed by any competing considerations.
. The term "Aesopian language” developed in Tsarist Russia to refer to language that, like Aesop’s fables, disguises the true meaning of speech by the use of metaphors, symbols and analogies, in order to avoid censorship.
. One defendant had been convicted of serious violence some years before the posters and website were published, so I except him from this part of the discussion.
. I note that on the instructions actually given to the jury, it is not easy to perceive the pertinence of much of this evidence. In particular, apart from extensive evidence of defendants' public statements concerning violence against abortion providers, there was also a great deal of evidence concerning their statements in meetings among anti-abortion activists. The jury was instructed that although speech can be a "true threat” no matter what defendants’ subjective intent, that intent is nonetheless pertinent context. I am not sure I see why, but since I would make subjective intent directly relevant, the point is not of great importance to this dissent.
. "Unconditional” refers to the degree of determination contained in the threat, not whether it is "conditioned” in the sense that the target could avoid the harm by bowing to the speaker's will.
. I note as well that the majority, while it articulated a de novo review standard with respect to the true threat standard it did apply, did not in fact review the record with an eye to First Amendment concerns such as those I have discussed, nor did it include the intent issue within its review. (Actual malice, a state of mind standard, is precisely the issue upon which the Supreme Court has closely scrutinized the record in defamation cases. See, e.g., Harte-Hanks Communications, Inc. v. Connaughton, 491 U.S. 657, 686, 109 S.Ct. 2678, 105 L.Ed.2d 562 (1989)).
. Rule 403 provides:
Although relevant, evidence may be excluded if its probative value is substantially outweighed by the danger of unfair prejudice, confusion of the issues, or misleading the jury, or by considerations of undue delay, waste of time, or needless presentation of cumulative evidence.
. The district court did not allow Treshman to respond in his closing argument to the plaintiffs’ argument based on the summaries because the actual transcripts were not in evidence.
. Rule 611(a) provides in its entirety:
The court shall exercise reasonable control over the mode and order of interrogating witnesses and presenting evidence so as to
(1) make the interrogation and presentation effective for the ascertainment of the truth,
(2) avoid needless consumption of time, and
(3) protect witnesses from harassment or undue embarrassment.
. The rules make clear that ''nonstenographic” refers to audio or visual recording. See, e.g., Rule 32 Advisory Comm. Note: ("This new subdivision [c] ... is included in view of the increased opportunities for video-recording and audio-recording of depositions under revised Rule 30(b).”); Fed. R. Civ. Proc. 30(b)(2) ("Any party may arrange for a transcription to be made from the recording of a deposition taken by non-stenographic means.”); Rule 30 Advisory Comm. Note ("The primary change in subdivision (b) is that parties will be authorized to record deposition testimony by nonstenographic means without first having to obtain permission of the court or agreement from other counsel.”); Rule 30 Advisory Comm. Note ("A party choosing to record a deposition only by video*1118tape or audiotape should understand that a transcript will be required by Rule 26(a)(3)(B) and Rule 32(c) if the deposition is later to be offered as evidence at trial ...."); Rule 26(a)(3) (''[A] party must provide ... the following information regarding the evidence that it may present at trial ... :(B) the designation of those witnesses whose testimony is expected to be presented by means of a deposition and, if not taken stenographically, a transcript of the pertinent portions of the deposition testimony.”).
. Although the Fourth, Fifth, and Seventh Circuits have generally approved the use of deposition summaries for the oral presentation of deposition evidence, I have found no case in which it is clear that the pertinent portions of the transcripts were not also admitted as evidence and available for jury review. See Oostendorp v. Khanna, 937 F.2d 1177, 1179-80 (7th Cir.1991); Walker v. Action Indus., Inc., 802 F.2d 703, 712 (4th Cir.1986); Kingsley v. Baker/Beech Nut Corp., 546 F.2d 1136, 1141 (5th Cir.1977) (deposition transcripts definitely admitted as evidence).
. I do not address whether the use of deposition summaries in this case was harmless error, see Cerrato v. San Francisco Cmty. Coll. Dist., 26 F.3d 968, 974 (9th Cir.1994) ("The harmless error standard in civil cases is whether the jury's verdict is more probably than not untainted by the error."), because the majority does not so hold.