United States v. Rundo

FOR PUBLICATION UNITED STATES COURT OF APPEALS FOR THE NINTH CIRCUIT UNITED STATES OF AMERICA, No. 24-932 D.C. Nos. Plaintiff - Appellant, 2:18-cr-00759- CJC-1 v. 2:18-cr-00759- CJC-1 ROBERT RUNDO; ROBERT 2:18-cr-00759- BOMAN, CJC-2 Defendants - Appellees. OPINION Appeal from the United States District Court for the Central District of California Cormac J. Carney, Senior District Judge, Presiding Argued and Submitted June 18, 2024 Pasadena, California Filed July 18, 2024 Before: RICHARD A. PAEZ and MILAN D. SMITH, JR., Circuit Judges, and JON S. TIGAR, District Judge. * Opinion by Judge Milan D. Smith, Jr. * The Honorable Jon S. Tigar, United States District Judge for the Northern District of California, sitting by designation. 2 USA V. RUNDO SUMMARY ** Criminal Law The panel reversed the district court’s judgment dismissing, based on a claim of selective prosecution, an indictment charging Robert Paul Rundo and Robert Boman with conspiracy to violate the Anti-Riot Act as well as with substantively violating the Act; reinstated the indictment; and remanded the case for trial. The indictment alleges that Rundo is a founding member of the “Rise Above Movement” or “RAM,” an organization that represents itself “as a combat-ready, militant group of a new nationalist white supremacy and identity movement.” It also states that Rundo and his colleagues, including Boman, attended a number of peaceful protests, when they chased down and violently attacked counter-protesters. The district court concluded that the government prosecuted RAM members such as Defendants while ignoring the violence of members of Antifa and related far- left groups because RAM engaged in what the government and many believe is more offensive speech. On a selective prosecution claim, the defendant bears the burden to demonstrate that (1) other similarly situated individuals have not been prosecuted and (2) his prosecution was based on an impermissible motive. Noting that this court has employed both a de novo standard and a clearly erroneous standard when reviewing a ** This summary constitutes no part of the opinion of the court. It has been prepared by court staff for the convenience of the reader.  USA V. RUNDO 3 selective prosecution claim, the panel did not resolve any purported difference because it held that Rundo has no viable selective prosecution claim under any standard of review. As to the first prong, the panel held that Rundo did not meet his burden to establish that similarly situated individuals were not prosecuted, that the district court erred by comparing collective conduct to individual conduct when it referred broadly to “Antifa and far-left groups” and comparing them to “Defendants,” and that the district court erred in holding that the individual Antifa members it did discuss were similarly situated to Rundo. As to the second prong, the panel held that Rundo failed to meet his burden to demonstrate that his prosecution was based on an impermissible motive. The panel explained that timing is a permissible reason to selectively prosecute; that Defendants’ statistics are unimpressive; and that Defendants’ other alleged sources of improper motive—a press release by the U.S. Attorney’s Office published after Rundo’s indictment was issued, alleged changes in charging decisions after Charlottesville, and the government’s own invocation of Defendants’ speech in its papers—were insufficient. 4 USA V. RUNDO COUNSEL Alexander P. Robbins (argued), Assistant United States Attorney, Deputy Chief Criminal Appeals Section; David R. Friedman, Solomon D. Kim, Kathrynne Seiden, Annamartine Salick, and Anna Boylan, Assistant United States Attorneys, Criminal Appeals Section; Bram M. Alden, Assistant United States Attorney, Criminal Appeals Section Chief; Cameron L. Schroeder, Assistant United States Attorney, National Security Division Chief; Mack E. Jenkins, Assistant United States Attorney, Criminal Division Chief; Elena S. Artson, Assistant United States Attorney, Appeals Section Chief; E. Martin Estrada, United States Attorney; United States Department of Justice, Office of the United States Attorney, Los Angeles, California; for Plaintiff-Appellant. Caroline S. Platt (argued), Assistant Federal Public Defender; Julia Deixler, Deputy Federal Public Defender; Cuauhtemoc Ortega, Margaret A. Farrand, and Erin Murphy, Federal Public Defenders, Central District of California; Federal Public Defender’s Office, Los Angeles, California; Peter Swarth, Law Offices of Peter Swarth, West Hills, California; for Defendants-Appellees.  USA V. RUNDO 5 OPINION M. SMITH, Circuit Judge: Defendants Robert Paul Rundo and Robert Boman were charged with conspiracy to violate the Anti-Riot Act, 18 U.S.C. §§ 2101–02, as well as with substantively violating the Act. In a prior appeal, the district court held that the Act was unconstitutional due to facial overbreadth under the First Amendment. United States v. Rundo, 990 F.3d 709, 712 (9th Cir. 2021) (Rundo I). We reversed and remanded because we determined that the Act was not facially overbroad except for certain severable portions. Id. On remand, the district court again dismissed the indictment, this time based on a claim of selective prosecution. For the reasons below, we reverse, reinstate the indictment, and remand this case for trial. FACTUAL AND PROCEDURAL BACKGROUND The indictment in this case charges Rundo and Boman 1 with (1) conspiring and agreeing to riot; and (2) aiding and abetting one another in using facilities of interstate commerce (e.g., the internet, a telephone, and a credit card) with intent to riot. Rundo, 990 F.3d at 713. It alleges that Rundo is a founding member of the “Rise Above Movement” or “RAM,” an organization that represents itself “as a combat-ready, militant group of a new nationalist white supremacy and identity movement.” Id. at 712–13. It also states that Rundo and his colleagues attended a number of 1 The original indictment charged Rundo, Boman, and two other individuals. The superseding indictment removed one defendant, and another subsequently pled guilty, leaving only Rundo and Boman at issue in this appeal. 6 USA V. RUNDO peaceful political protests, where they chased down and violently attacked counter-protestors. The affidavit in support of the criminal complaint recounts the fruits of an investigation conducted by FBI Agent Scott Bierwirth, who specializes in investigating domestic terrorist groups. Bierwirth, who had previously obtained a warrant to search one RAM member’s cell phone, recounts RAM’s involvement in four political rallies: one on March 25, 2017, in Huntington Beach; one on April 15, 2017, in Berkeley; one on June 10, 2017, in San Bernardino; and one on August 11, 2017, in Charlottesville, Virginia. We describe each rally below, as well as Rundo and Boman’s participation in the rallies. A. March 25, 2017 Huntington Beach rally The indictment begins by alleging that Rundo prepared RAM for the “Make America Great Again” (“MAGA”) rally at Huntington Beach by organizing training for “group fighting.” About a week before the rally, a RAM member sent messages through a social media platform to inform other members of future combat training, “listing defendant Rundo as the point of contact, to prepare RAM members and others to engage in violence at [the] upcoming political events.” On March 25, 2017, “at least several hundred people” attended the MAGA rally in Huntington Beach. Videos show that a small group of “counter-protestors [] turned away from the group of rally attendees and walked north along the beach while a small group, led by Rundo . . . pursued them.” The counter-protestors “continued walking north away from the RAM members, as the RAM members  USA V. RUNDO 7 continued pursuing them.” The affidavit explains how Rundo and Boman began attacking counter-protestors: Seconds later, videos show Rundo approached another of the counter-protesters from behind and punched him in the back of the head. Boman ran up behind Rundo toward the counter-protestor, but turned back after a second counter-protester released pepper spray. Rundo then turned to the second counter-protestor, punched him in the back of the head, grabbed the back of his neck, and threw him to the ground, landing on top of him. Rundo then held the counter- protester down with his left hand and threw several punches at the counter-protester’s head while other RAM members looked on, cheered, and prevented others from intervening . . . After several seconds, video shows the counter-protestor released pepper spray, leading Rundo to back away from him. The crowd eventually dispersed. A few days later, Boman posted images of the fight on social media, celebrating the attack. Bierwirth found an image which appears to have been posted by RAM’s Instagram account which “depicts Rundo punching a counter-protester, with the words ‘Physical Removal’ 2 superimposed across the top.” The image, as well as other screenshots from videos of fights, are 2 According to Bierwirth, “physical removal” is “a term commonly used by white supremacy extremists to refer to the goal of separating or ‘physically removing’ those with viewpoints or lifestyles that are viewed as undesirable.” 8 USA V. RUNDO included in the affidavit. Members of RAM later prepared to attend another rally in Berkeley, including by renting a passenger van on a credit card. Rundo and Boman also used a credit card to book hotel rooms for themselves in the area. B. April 15, 2017 rally in Berkeley, California A few weeks later, Rundo and Boman attended another rally in Berkeley. There, the authorities had set up fencing between rally attendees and counter-protestors. “After several minutes, several RAM members crossed the orange fencing and assaulted counter-protesters.” The affidavit describes some of Rundo’s role in the violence: According to Berkeley Police Department (“BPD”) officers, a BPD officer saw Rundo punching [an] apparently defenseless person in the head, and ordered Rundo to stop, but Rundo did not respond. The BPD officer knocked Rundo to the ground to stop the ongoing assault, and Rundo punched the officer twice in the head before BPD officers subdued and arrested him. Following the rally, RAM members boasted about their violent attack. Rundo, who was offered to be interviewed as a RAM leader, suggested to the interviewer that he “could mention [] how we were the first guys to jump over the barrier and engage,” i.e., how they were the ones to initiate the violence. Rundo and Boman also celebrated online by posting videos and images of themselves assaulting counter- protestors. For example, “on September 14, 2017, the RAM Twitter account posted a picture of Rundo and another RAM member assaulting counter-protesters at the Berkeley rally.”  USA V. RUNDO 9 C. Other rallies Videos also show that Rundo and others attended an “Anti-Islamic Law” rally in San Bernardino, California. Images on social media show that Rundo was at the rally and engaged in violence. Rundo messaged a RAM member on a social media site and offered to send him a video showing himself “smashing the antifa car and chasing [counter- protestors]” at the rally. RAM members also planned to attend another rally in Charlottesville. On July 25, 2017, Rundo texted another RAM member telling them that he might have “a place for you guys to stay in Charlottesville.” He also sent a message to another member saying he “hope[d] [h]e can get y’all some more commie beatdown vids soon.” Despite Rundo’s desire to send “vids” of him beating people at the rally, neither he nor Boman ultimately attended the Charlottesville event. The affidavit shows that Rundo and Boman self- generated much of the evidence supporting the indictment. For example, Rundo posted a message “giving a shoutout to the only alt right crew that actually beats Antifa 3 senseless and wins rallies.” Bierwirth also reviewed text messages between Rundo and another RAM member which suggested 3 Throughout the record, the parties and the district court refer to “Antifa” as if it were a formal organization, using language such as “Antifa member.” However, Antifa, which stands for “anti-fascist,” “is not a well-structured organization, but rather a loosely organized, secretive movement of like-minded far-left activists. There are no leaders, no hierarchy and no formal membership.” Aram Roston, “American Antifa,” Reuters (Aug. 25, 2021), https://www.reuters.com/investigates/special-report/usa-antifa-profile. Nonetheless, for the sake of consistency, we adopt the district court’s terminology in this opinion. 10 USA V. RUNDO that Rundo “orchestrated the creation of [a] video” showing “Rundo . . . and other RAM members assaulting counter- protestors,” and “RAM members training in hand-to-hand combat.” The affidavit concludes that “Rundo . . . along with other RAM members, have used the Internet to prepare to incite and participate in violence at various political events, have committed violent assaults while at those events, and have applauded each other for it and publicly documented their assaults in order to recruit more members to engage in further assaults.” JURISDICTION AND STANDARD OF REVIEW We have jurisdiction pursuant to 18 U.S.C. § 3731. Our court “has employed both a de novo standard and a clearly erroneous standard when reviewing a selective prosecution claim.” United States v. Culliton, 328 F.3d 1074, 1080 (9th Cir. 2003) (per curiam). Because we hold that Rundo “has no viable selective prosecution claim under any standard of review, we need not resolve any purported difference.” Id. at 1080. ANALYSIS “In our criminal justice system, the Government retains broad discretion as to whom to prosecute.” Wayte v. United States, 470 U.S. 598, 607 (1985) (quotations omitted); United States v. Armstrong, 517 U.S. 456, 464 (1996) (“The Attorney General and United States Attorneys retain ‘broad discretion’ to enforce the Nation’s criminal laws.” (cleaned up)). However, prosecutorial discretion is not “unfettered.” Wayte, 470 U.S. at 608. In particular, the decision to prosecute may not be “deliberately based upon an unjustifiable standard such as race, religion, or other arbitrary classification, including the exercise of protected  USA V. RUNDO 11 statutory and constitutional rights.” Id. (internal citation and quotations omitted). To prove a selective prosecution claim, a defendant must show “discriminatory effect and discriminatory intent.” Armstrong, 517 U.S. at 468. 4 Specifically, the defendant bears the burden to “demonstrate that (1) other similarly situated individuals have not been prosecuted and (2) his prosecution was based on an impermissible motive.” United States v. Sutcliffe, 505 F.3d 944, 954 (9th Cir. 2007) (citing Culliton, 328 F.3d at 1081). “The standard for proving [a selective prosecution] claim ‘is particularly demanding, requiring a criminal defendant to introduce ‘clear evidence’ displacing the presumption that a prosecutor has acted lawfully.” Id. (quoting Reno v. American-Arab Anti-Discrimination Comm., 525 U.S. 471, 489 (1999)); Armstrong, 517 U.S. at 464 (“The presumption of regularity supports [the government’s] prosecutorial decisions and, in the absence of clear evidence to the contrary, courts presume that [prosecutors] have properly discharged their official duties.” (internal quotations omitted)); see also Sellers, 906 F.3d at 852 (explaining that the Supreme Court adopted a “rigorous standard” to test selective prosecution claims). As such, we “view the evidence in the light most favorable to the government” 4 Although Armstrong technically involved the standard for obtaining discovery on a selective prosecution claim, 517 U.S. at 468, we have since explained that “[t]he standard for obtaining discovery on a selective prosecution claim is ‘nearly as rigorous as that for proving the claim itself.’” United States v. Sellers, 906 F.3d 906 F.3d 848, 852 (9th Cir. 2018). 12 USA V. RUNDO when reviewing such claims. Culliton, 328 F.3d at 1080– 81. 5 I. Rundo did not meet his burden to establish that similarly situated individuals were not prosecuted. A. The district court erred in its points of comparison. To establish discriminatory effect, “the claimant must show that similarly situated individuals . . . were not prosecuted.” Armstrong, 517 U.S. at 465. The district court analyzed six rallies where it found that members of “far-left groups” such as Antifa engaged in “the same, if not worse,” conduct than that of Defendants. Three of the events are separate from and pre-date any mentioned in the indictment. The three other events are the rallies specifically discussed in the indictment. Based on these incidents, the district court concluded that the government “prosecute[d] RAM members such as Defendants while ignoring the violence of members of Antifa and related far-left groups because RAM engaged in what the government and many believe is more offensive speech.” As a threshold matter, the government argues that the district court erred by choosing to compare the collective conduct of groups to Defendants rather than comparing the conduct of individuals to Defendants. See Armstrong, 517 U.S. at 465 (“[T]he claimant must show that similarly situated individuals . . . were not prosecuted.” (emphasis 5 In addition to the problems identified below, we note that the district court rested its conclusions on evidence of doubtful quality. The district court seemed overly focused on expressing its concerns about Antifa instead of reviewing evidence in the record, a problem it compounded by relying on other “evidence” it made no effort to authenticate— including, among other things, an online article by “Anonymous Contributor.”  USA V. RUNDO 13 added)). For example, in describing the February 2016 KKK rally, the district court noted that “[s]everal dozen Antifa extremists initiated an altercation with the Klansmen that led to multiple injuries and three stabbings.” Similarly, in discussing an “event at UC Berkeley,” the district court noted that “100 to 150 agitators” who were part of a group called “By Any Means Necessary,” or “BAMN,” collectively “smashed a half a dozen windows.” The district court drew on these examples to conclude that “Antifa and related far-left groups attended the same Trump rallies as Defendants with the expressly stated intent of shutting down, through violence if necessary, protected political speech.” By referring broadly to “Antifa and far-left groups” and comparing them to “Defendants,” the district court compared apples to oranges. The point of the “similarly situated” analysis is to “isolate the factor allegedly subject to impermissible discrimination.” United States v. Aguilar, 883 F.2d 662, 706 (9th Cir. 1989), superseded on other grounds by statute, Immigration Reform and Control Act of 1986, Pub. L. No. 99-603, 100 Stat. 3359, as recognized in United States v. Gonzalez-Torres, 309 F.3d 594 (9th Cir. 2002). That is impossible to do when comparing collective conduct to individual conduct. A good illustration of the “isolation” principle can be found in United States v. Steele, 461 F.2d 1148 (9th Cir. 1972). In that case, a panel of our court vacated the convictions of four individuals who were part of a census resistance movement and were prosecuted for refusing to fill out the census forms. Id. at 1150–52. Defendant Steele had “held a press conference, led a protest march, and distributed pamphlets” demonstrating opposition to the census. Id. at 1151. In making his selective prosecution claim, “Steele . . . located six other persons who had 14 USA V. RUNDO completely refused on principle to complete the census forms,” but were not prosecuted. Id. By comparing individuals (i.e., Steele and each of these six individuals) to their individual actions (i.e., refusing to fill out the census), the panel was able to “isolate” the impermissible factor: “tak[ing] a public stand against the census.” Id. That kind of exercise is impossible to do here, where we do not know, for example, which of the “dozens” of individual Antifa members assaulted people at the KKK rally. Some Antifa members at the rally may have injured multiple people; some may have injured none. See Aguilar, 883 F.2d at 706 (“Absent a similarly situated control group, the government’s prosecution of a defendant exercising his constitutional rights proves nothing.”). As it stands, only one of the six events chosen by the district court involve the specific conduct of named individuals: the March 2017 Huntington Beach Rally, where the individuals J.M.A., J.F., and J.A. were present. All other examples only refer to collective conduct. 6 Defendants devote their brief almost entirely to the conduct of J.A., J.F., and J.M.A at the Huntington Beach Rally. And they take a significantly narrower position than that of the district court: in Defendants’ view, “the only appropriate comparison [to Defendants] [are] the violent, far-left-wing individuals who went to the same political 6 The district court also discussed the conduct of individuals at the April 2017 Berkeley rally. However, the identities of the individuals at this rally remain unknown. See Sellers, 906 F.3d at 853 (“In a meritorious selective prosecution claim, a criminal defendant would be able to name others arrested for the same offense who were not prosecuted by the arresting law enforcement agency”) (quoting Chavez v. Ill. State Police, 251 F.3d 612, 640 (7th Cir. 2001)).  USA V. RUNDO 15 rallies at issue in this indictment, and who themselves engaged in violence.” Nonetheless, sprinkled throughout Defendants’ brief are references to people who are not “individuals who went to the same political rallies at issue in this indictment.” For example, Defendants cite the conduct of an individual named Yvette Felarca six times in their brief. 7 But there is no evidence that Felarca went to the same rallies at issue in the indictment or was even known to federal officials. By Defendants’ own definition, then, she is not a similarly situated comparator. We therefore only consider the conduct of J.A., J.F., and J.M.A. in analyzing Defendants’ selective prosecution claim. B. The district court erred in holding that the individual Antifa members it did discuss were “similarly situated” to Rundo. To establish discriminatory effect, a defendant must show that the comparator is “the same as the defendant in all relevant respects, except that defendant was, for instance, exercising his first amendment rights.” Aguilar, 883 F.2d at 706. As discussed above, the district court only cited one 7 Felarca was never mentioned by name in the district court’s order, but she appears to be the “leading BAMN member” who was interviewed in an article cited by the district court. This portion, and many other portions of the district court’s order, “recounted hearsay and reported personal conclusions based on anecdotal evidence.” Armstrong, 517 U.S. at 470 (noting that defendants’ evidence in support of their claim, including “affidavits [] which . . . recounted hearsay and reported personal conclusions based on anecdotal evidence,” did not remedy the fact that they were unable to find similarly situated individuals not prosecuted); see also United States v. Turner, 104 F.3d 1180, 1185 (9th Cir. 1997) (discounting evidence in the form of “newspaper anecdotes and hearsay”). 16 USA V. RUNDO example where it discussed the conduct of those whom it believed were similarly situated individuals: the March 25, 2017 MAGA rally at Huntington Beach. According to the district court’s review of police reports from this event, a far-left activist named J.A. pepper sprayed a Trump supporter without provocation. Another individual, J.M.A., kicked and punched those around him. And another, J.F., was identified by a Trump supporter as having pepper sprayed her in the face. As the district court noted, all three individuals were arrested, “[b]ut none were charged federally for violating the Anti-Riot Act.” On this basis, the district court concluded that Rundo met the first prong of his selective prosecution claim. On appeal, the government argues that the left-wing activists cited above are not similarly situated to Defendants because their conduct at these events was not as organized or repetitive as that of Defendants. The government also argues, more generally, that the comparators are not similarly situated with respect to factors bearing on the decision of whether to federally prosecute the alleged Antifa members. i. Nature of alleged conduct J.A., J.M.A, and J.F. are not similarly situated to Defendants when the nature of their conduct is compared to that of Defendants. First, Defendants repeated their conduct at other rallies, while these individuals did not (or, at minimum, Rundo failed to provide “clear evidence” that they did). The indictment lists numerous rallies at which Rundo and his co-defendants repeatedly engaged in the same conduct—they communicated with one another regarding rallies, attended the rallies, attacked people, and gloated  USA V. RUNDO 17 about those attacks on the internet. There is no similar evidence in the record that J.A., J.M.A., or J.F. did the same. Second, Defendants behaved like leaders of an organized crime group. They coordinated combat training sessions; created materials to recruit others; and planned cross- country travel to commit their acts. Nothing in the record indicates that J.A., J.F., or J.M.A. were similarly organized. That is, there is no evidence that these individuals trained together, coordinated their attendance at other rallies, recruited others, or otherwise did anything other than commit violence at the Huntington Beach rally. Defendants’ response to this point is two-fold. First, they take issue with the conclusion as a factual matter. Defendants note that “Felarca . . . attended at least three [other rallies],” and that “J.F. also admitted [to] attending multiple rallies.” But for the reasons discussed above, Felarca is not a similarly situated individual Defendants can rely on in this analysis. And as to J.F., all the cited police report states is that J.F. “was an activist and had been to numerous protests.” Nothing in the report indicates that J.F. planned to commit, and had in fact committed, violence at these other protests. 8 Defendants also point to the district court’s “finding” that J.A., J.F., and J.M.A. were not “less organized than RAM.” But that “finding” suffers from some flaws. The district court again compared individuals (J.A., J.F., and J.M.A.) to groups (RAM). And the district court reached this conclusion because in its view, the police reports for 8 To the contrary, J.F. told the arresting officer that a video existed of him showing that he was punched by another person—not that he hit another person himself. 18 USA V. RUNDO J.A., J.F., and J.M.A. “unequivocally state[d]” that “the far- left activists present at the Huntington Beach rally intended to commit violence.” 9 But intending to commit violence does not speak to whether the comparators were as organized as Defendants were. The comparators are thus not similarly situated. 10 Defendants’ second argument is that the distinctions drawn by the government are simply irrelevant. But we have previously held that the types of distinctions drawn by the government here (i.e., repetitive and organized nature of 9 Only J.A.’s police report contains an admission on her part suggesting that she intended to commit violence. As discussed below, the police report for J.F. shows that he had claimed self-defense, and that he brought his pepper spray “for his personal protection.” The police report for J.M.A. does not “unequivocally state” that he intended to commit violence. 10 Defendants suggest that the conduct of J.A. and J.F. may have been worse than that of Rundo’s or Boman’s because while J.A. and J.F. “deliberately brought pepper spray” to rallies, Defendants were “unarmed.” In the context of this case, any such suggestion is absurd. There is no question that despite not being “armed” with pepper spray, Rundo was able to seriously beat individuals with his fists (which he had taped beforehand), often only stopping once pepper spray was deployed. Defendants also suggest that the conduct of J.F., J.A., and J.M.A. was more serious than that of their own because “Rundo was present at each of these rallies as a participant, not a counter-protestor.” Again, that characterization is both factually incorrect and legally immaterial. Rundo cannot fairly be characterized as a “participant” at these rallies— in fact, the right-wing organizer of the San Bernardino rally called the police to report Rundo, whom she expressly described as “not part of the larger [rally] group.” Moreover, framing both sets of individuals as protestors or counter-protestors overlooks the fact that (1) people have the right to be both, and (2) both sets of individuals here crossed the line when they stopped picketing and started throwing punches or deploying pepper spray at others.  USA V. RUNDO 19 conduct) can set a defendant and a purported comparator apart. For example, in Sutcliffe, we held that an individual was not similarly situated to a defendant who was charged with transmitting threats, because the defendant had repeated his conduct far more times than the comparator. 505 F.3d at 954. The comparator in Sutcliffe “sent a single . . . email to Defendant in response to illegal and provocative communication previously posted online by Defendant,” while the defendant had, “over the course of several months” made “multiple threats of violence” and posted the personal information of “over a thousand” individuals online. Id. at 951, 954. Similarly, we have previously credited the government’s representations that organized crime is more pernicious than the crimes of individual offenders, such that it is acceptable for the government to focus on prosecuting the former. See, e.g., Aguilar, 883 F.2d at 707 (acknowledging as a permissible, facially-neutral consideration “the [prosecutor’s] focus . . . []on organized smuggling rings”); Turner, 104 F.3d at 1183–85 (activities of gang members prosecuted for sale of cocaine after having “been subjected to two or more stings by law enforcement” which resulted in “large quantities of crack” being discovered was not comparable to “the single sale of cocaine by individuals”). Therefore, Defendants’ repeated conduct at these rallies and their operation as an organized group make them dissimilar to the comparators. ii. Factors bearing on the decision to prosecute The comparators cited by the district court are also not similarly situated with respect to “factors bearing on the decision [of] whether to federally prosecute them.” In 20 USA V. RUNDO Wayte v. United States, the Supreme Court explained that these factors include the “strength of the case, the prosecution’s general deterrence value, the Government’s enforcement priorities, and the case’s relationship to the Government’s overall enforcement plan.” 470 U.S. at 607. The Court cautioned against judicial review of these factors, as they are generally left to prosecutors and “are not readily susceptible to the kind of analysis the courts are competent to undertake.” Id. With that framing in mind, we address the arguments regarding prosecutorial decision-making in this case below. See Culliton, 328 F.3d at 1081 (examining “[defendant’s] selective prosecution claim with the caveat that the decision to prosecute is particularly ill-suited to judicial review.” (internal quotations omitted)). The government first notes that the individuals cited by the district court do not share the characteristics of Defendants that may have incentivized the government to prosecute the latter. Specifically, there is no “clear evidence” (and possibly no evidence at all) that J.A., J.F., or J.M.A. had previously been convicted of violent crimes. In contrast, the parties do not dispute that Rundo was previously convicted of stabbing a person, and that Boman previously was convicted of various violent crimes. The closest possible comparison is a statement in a police report that J.F. was convicted of “remaining at the scene of a riot,” which does not indicate that he engaged in violence at the scene. Defendants do not address this point in their answering brief, and indeed, our case law supports the point that differences in criminal histories could warrant prosecution of some individuals over others. See Turner, 104 F.3d at 1183 (describing as “neutral” criteria such as “prior felony convictions”); cf. Armstrong, 517 U.S. at 460 (holding that  USA V. RUNDO 21 there was no evidence of selective prosecution where the prosecutor “explained in an affidavit that . . . several of the defendants had criminal histories including narcotics and firearms violations”). Permitting the consideration of criminal history makes sense, in light of the government’s desire to deter repeat offenders. See Wayte, 470 U.S. at 607 (prosecutorial decisions include factors such as “the prosecution’s general deterrence value”). Similarly, unlike the “left-wing” individuals cited by the district court, Rundo clearly had a leadership role in RAM. The importance of Rundo to RAM, especially when compared to the lack of evidence demonstrating that J.A., J.F., and J.M.A. were at all important to Antifa’s leadership, provides a facially neutral explanation that the judiciary is not well-equipped to second-guess. Cf. United States v. Alanis, 265 F.3d 576, 585 (7th Cir. 2001) (concluding that the defendant was not similarly situated with his two co- conspirators because he had a larger role in the conspiracy and was not cooperating with the government as were his co- conspirators). But perhaps the strongest point in the government’s favor is the strength of the evidence against Defendants versus that against J.A., J.F., and J.M.A. See Wayte, 470 U.S. at 607 (recognizing that prosecutorial decisions include factors such as “strength of the case”). For example, Rundo and his co-defendants conveniently bragged about their exploits online, providing the government with ample evidence of their alleged unlawful conduct. In response, Defendants assert that “members of the far-left” bragged about their exploits “as well.” In support of their claim, Defendants point to the actions of Felarca and an individual named S.H., neither of whom are similarly situated comparators by Defendants’ own admission. Defendants 22 USA V. RUNDO attempt to take characteristics of random individuals and piece them together with those of J.A., J.F., and J.M.A. in an attempt to create an ideal comparator, but this comparator does not actually exist (or, rather, Rundo has not shown it to exist). Defendants further suggest that the government’s focus on Defendants’ bragging evinces a discriminatory motive because “publicly bragging” about these exploits “is protected speech.” But the relevance of the “bragging” evidence is not that it may be speech. Rather, the bragging evidence is relevant because it goes to the strength of the evidence, i.e., how easily a prosecutor can prove that an individual committed a crime. That is a permissible consideration in making prosecutorial selections. Cf. Armstrong, 517 U.S. at 460 (finding no selective prosecution where the prosecutor had “explained in an affidavit that . . . the overall evidence in the case was extremely strong, including audio and videotapes of defendants”). To illustrate this point, compare the conduct of Boman with J.F., as the district court did. The police report for J.F. states that he told the officer that he “came to protest fascism, not to fight personally,” and that he used pepper spray in “self-defense” when witnessing J.A. “get punched in the face.” In contrast, the day after attending the same rally J.F. did, Boman openly posted and commented on an article and a photo of himself indicating that he had “physically removed” Antifa members. And he did this with nearly all of the events he attended, according to the affidavit. It was only years after the filing of his indictment, at a change of plea hearing, that Boman—a self-proclaimed white supremacist—asserted he acted violently to protect “this  USA V. RUNDO 23 Black kid” at a rally. 11 That some of Defendants’ statements, divorced from their context, can be framed as “protected speech” after the fact does not preclude them from also being used as evidence in support of a decision to prosecute Boman and Rundo. In sum, Rundo has not met his burden to show that J.A., J.F., or J.M.A. are similarly situated comparators. We thus hold that Rundo failed to meet the first prong of the selective prosecution test. II. Rundo did not meet his burden to establish impermissible motive. Rundo has also failed to meet his burden to demonstrate that his prosecution was “based on an impermissible motive.” Culliton, 328 F.3d at 1081. “Discriminatory purpose implies more than intent as awareness of consequences. It implies that the decisionmaker selected or reaffirmed a particular course of action at least in part ‘because of,’ not merely ‘in spite of,’ its adverse effects upon an identifiable group.” Wayte, 470 U.S. at 610 (citing Pers. Adm’r of Massachusetts v. Feeney, 442 U.S. 256, 279 (1979)) (cleaned up). “Mere selectivity in prosecution creates no constitutional problem.” Steele, 461 F.2d at 1151. The district court viewed this prong of the test as presenting a “closer question.” Nonetheless, it stated that “the only conclusion” it could reach was that the government prosecuted Defendants “based solely on [their] . . . speech 11 The district court conveniently omitted information regarding the timing and circumstances of Boman’s self-defense claim when it “found” that he was similarly situated to J.F. To the extent a clear error standard applies, we are left with a “definite and firm conviction that a mistake has been committed” when it drew an equivalence between these two individuals. Easley v. Cromartie, 532 U.S. 234, 242 (2001). 24 USA V. RUNDO and beliefs,” because (1) the government began prosecuting Defendants after the death of a left-wing counter-protestor at the Charlottesville alt-right rally, (i.e., “timing”); and (2) the government “did not prosecute far-left activists who were also responsible for violence at political rallies in the period and places at issue.” The district court’s reasoning is flawed. As to the first ground, we have never held that the timing of prosecutions is suggestive of improper motive. Moreover, “timing” can merely be the sign of the government’s change in enforcement priorities—which, as explained below, is a permissible reason to selectively prosecute. See Wayte, 470 U.S. at 607. As to the second ground cited by the district court, it is the same as the first prong of the selective prosecution test. See Sutcliffe, 505 F.3d at 954 (“To succeed on this claim, Defendant must demonstrate that (1) other similarly situated individuals have not been prosecuted . . . ”). The government is thus correct when it asserts that the district court collapsed the two prongs of the selective prosecution analysis. Nonetheless, Defendants expound on the district court’s (incorrect) statement that “the government has only prosecuted RAM members and not prosecuted any members of Antifa,” 12 in an attempt to show impermissible motive. 12 The government has in fact been prosecuting members of left-wing groups, albeit not under the Anti-Riot Act. See e.g., United States v. Wilson, C.A. No. 23-50016; United States v. Harold, 23-CR-49, Docket No. 39 at 9-11 (D. Or. Apr. 12, 2024) (seeking discovery to support the assertion that federal law enforcement targeted “individuals who were protesting in support of left-wing causes”).  USA V. RUNDO 25 Citing the Supreme Court’s decision in Arlington Heights, 13 they argue that evidence of impermissible motive may not only come in the form of direct evidence, but circumstantial evidence as well, such as statistics. And they argue that impermissible motive is evident from the statistics alone because “[z]ero percent of the violent left-wing counter- protestors were charged federally, and 100 percent of the [Anti-Riot Act] charges in this district are against alleged white supremacists.” Defendants’ “statistics” are unimpressive, because these numbers “do[] not insure that all distinctions extraneous to the first amendment expression are removed.” Aguilar, 883 F.2d at 707. In other words, the “statistics” do not reflect the fact that there could be other, permissible reasons why Antifa and/or BAMN members have not been prosecuted that have nothing to do with Defendants’ ideology. 14 See 13 Selective prosecution claims “draw on ordinary equal protection standards.” Armstrong, 517 U.S. at 465 (internal quotation marks omitted). However, the standards governing selective prosecution claims and equal protection claims are not one and the same; i.e., the Arlington Heights analysis does not somehow lessen the “rigorous” standard applied to the selective prosecution claims. 14 Take, for example, the case of Felarca, who Defendants cite repeatedly in their brief. Based on the record before us, it appears that Felarca is a controversial leader of BAMN who has, just like Defendants, gone to rallies and injured people. But what Defendants fail to mention is that Felarca did “face[] a felony assault charge over the clash” at one of the rallies she attended. The charge was not brought by the federal authorities, but by the Sacramento District Attorneys’ office. Why was Felarca not charged federally? It could be because, as Defendants argue, there was an “impermissible” change in practice by US Attorneys’ offices after the deadly Charlottesville rally to begin federally prosecuting rioters as well. On the other hand, Felarca might 26 USA V. RUNDO United States v. Bass, 536 U.S. 862, 864 (2002) (“[R]aw statistics regarding overall charges say nothing about charges brought against similarly situated defendants.”). More fundamentally, these figures do not even constitute a valid statistical comparison. Defendants compare the percentage of “violent left-wing counter-protestors” prosecuted to the number of “alleged white supremacists” who were charged. That is not an apples-to-apples comparison. Assuming for the sake of argument that such a comparison would be relevant, the appropriate basis of comparison would be the percentage, not the number, of alleged white supremacists who were prosecuted. But Defendants do not provide that information. not have been federally prosecuted because she was already being prosecuted by the state—which, as the government notes, is a “valid reason to decline federal charges.” See Petite v. United States, 361 U.S. 529, 530 (1960) (noting Department of Justice policy generally to avoid “duplicating federal-state prosecutions” out of “fairness to defendants and of efficient and orderly law enforcement”); United States v. Venable, 666 F.3d 893, 901 (4th Cir. 2012) (federal charging decisions may consider whether the state is prosecuting). Given that prosecutorial decisions such as these are often privileged, we may never know the answer to this question. Sellers, 906 F.3d at 853. But the bigger takeaway from this example is that—even assuming statistical evidence were ever enough to show impermissible motive— Defendants have not made that showing here, given the number of variables at play. See Aguilar, 883 F.2d 662, 706 (9th Cir. 1989) (rejecting similar statistic that “the indictments in this case represent 100% of the ‘non-economic/non-exploitative’ substantive and conspiracy indictments under section 1324 . . . [while] growers and ranchers employing illegal aliens had not been prosecuted in the previous ten years”); see also Turner, 104 F.3d at 1184 (reversing district court where it had concluded that statistics of defendants charged with offenses, “standing alone” raised an inference of discrimination).  USA V. RUNDO 27 Since statistics alone cannot prove discriminatory motive, Defendants must provide other evidence of motive to meet the second prong of Armstrong. They raise: (1) a press release by the U.S. Attorney’s Office published after Rundo’s indictment was issued; (2) alleged changes in charging decisions after Charlottesville; and (3) the government’s own invocation of Defendants’ speech in its papers. We address each source of alleged improper motive in turn. First, Defendants rely on the U.S. Attorney’s Office press release cited by the district court to assert improper motive. That press release, published after Rundo was indicted, states that the indictment was issued to address “an orchestrated effort to squelch free speech as members of [Rundo’s] conspiracy travelled to multiple locations to attack those who hold different views.” Defendants go further to argue that the press release “itself demonstrates the government’s discriminatory motive for this prosecution, highlighting that it was targeting the alleged ‘white supremacists’ for prosecution.” But a simple review of the press release shows nothing of the sort. Nothing in the press release says that RAM was “targeted” because they were “white supremacists”; instead, the words “white supremacists” are only used to describe a statement of fact: that RAM consists of them. See Press Release, Four Local Members of White Supremacy Group Face Federal Charges in Attacks at Political Rallies across California, U.S. Att’y’s Off., Cent. Dist. of Cal. (Oct. 24, 2018) (“Press Release”) (describing defendants as a group of “[f]our Southern California men who allegedly are members of a white supremacy extremist group”); Turner, 104 F.3d at 1184 (“The kind of intent to be proved is that the government undertook a particular course of action at least in part 28 USA V. RUNDO ‘because of,’ not merely ‘in spite of’ its adverse effects upon an identifiable group.” (internal quotation marks omitted)). And as to the assertion that the case was initiated to address violent attempts to “squelch free speech”? That statement in and of itself is not suspect, and read in the context of the press release, it puts the non-discriminatory motive of the prosecutors’ office on display. It reads: “Every American has a right to peacefully organize, march and protest in support of their beliefs – but no one has the right to violently assault their political opponents,” said United States Attorney Nick Hanna. “The allegations describe an orchestrated effort to squelch free speech as members of the conspiracy travelled to multiple locations to attack those who hold different views. This case demonstrates our commitment to preserve and protect the freedoms guaranteed by the Constitution.” Press Release at 2. Defendants’ reliance on the press release is thus unpersuasive. Next, Defendants argue that federal prosecutors “[d]epart[ed] from the normal procedural sequence” in prosecuting RAM members, and that such a departure raises an inference of impermissible discriminatory motive. Arlington Heights, 429 U.S. at 267. Specifically, they assert that prior to the death of Heather Heyer in Charlottesville, federal prosecutors permitted local prosecutors to pursue criminal charges against rioters. After Charlottesville, “when it came to alleged white supremacists, the federal government no longer deferred to local authorities’ charging  USA V. RUNDO 29 decisions regarding wholly intrastate ‘riot’ conduct. But it did not decide the same as to violent left-wing rally-goers.” The problem with Defendants’ contention is that the federal government (as opposed to this court) is permitted to set its own “enforcement priorities,” and create an “overall enforcement plan,” if it has a legitimate reason to do so. Wayte, 470 U.S. at 607. Presumably, such a reason would include the death of an American exercising her constitutional rights at a political rally where RAM’s co- founder—an individual named Ben Daley—was present. Cf. Turner, 104 F.3d at 1185 (rejecting policy argument that a certain group “must be denied the protection of law enforcement by the federal government because the likely suspects are overwhelmingly apt to be members” of the same). Defendants respond to this point about enforcement priorities by arguing that Rundo and Boman had not even attended that Charlottesville rally; that Daley’s comments post-Charlottesville regarding his involvement is protected speech; and otherwise implying that the events at Charlottesville are “unrelated” to this case. But these points are not persuasive. Charlottesville was historic in that a counter-protestor was killed at the hands of someone who disagreed with her. It therefore comes as no surprise that a counter-protestor’s death would trigger (1) investigations into the individuals present at the event, (2) their associates, and (3) prosecutions of those associates. An investigation spurred by a high- profile event is not unconstitutional. That Rundo and Boman were not at Charlottesville or that Daley’s comments might be protected speech is beside the point, because the former are being prosecuted based on 30 USA V. RUNDO evidence generated from a post-Charlottesville investigation into the latter—not on a layperson’s observations of the rally itself. 15 Given that we “view the evidence in the light most favorable to the government” in these kinds of cases, Culliton, 328 F.3d at 1081, we hesitate to conclude that a “departure from the normal procedural sequence” here was based on anything other than the executive’s focus to protect members of the public after this notable event. Finally, Defendants argue that the government’s invocation of Rundo’s protected speech and ideology in its papers is proof that “[t]he government does not like Rundo and (now-defunct) RAM’s far-right ideology.” But if referring to the facts of a case were enough to raise an inference of discriminatory motive, then the government would never be able to prosecute an individual who has committed a crime, and whose ideology was one of the motives (or the principal motive) behind the crime. As the Court in Wayte warned: Judicial supervision in this area . . . entails systemic costs of particular concern. Examining the basis of a prosecution delays the criminal proceeding, threatens to chill law enforcement by subjecting the prosecutor’s motives and decisionmaking to outside inquiry, and may undermine prosecutorial effectiveness by revealing the Government’s enforcement policy. All these are substantial concerns that make the courts properly 15 After all, Defendants are not bringing a selective enforcement claim— which alleges discriminatory practices in initiating investigations—but only a selective prosecution claim, which alleges that the subsequent prosecution using the fruits of an investigation is improper.  USA V. RUNDO 31 hesitant to examine the decision whether to prosecute. Wayte, 470 U.S. at 607–08. The district court’s strained attempt to find impermissible motive in this case highlights the importance of the “similarly situated” exercise in prong one. Without, say, an individual left-wing comparator who is similar to Rundo “in all relevant respects,” a district court cannot actually isolate the reason for the defendant’s prosecution. That is by design; the presumption of regularity, which “supports prosecutorial decisions . . . gives a measure of protection (and confidentiality) to prosecutors’ deliberative processes, which are covered by strong privileges.” Sellers, 906 F.3d at 853 (cleaned up). Based on the record before it, the district court could only guess that Rundo was prosecuted because of his speech. Such a guess is insufficient to rebut the presumption of regularity. CONCLUSION We REVERSE the district court’s dismissal order, REINSTATE the indictment, and REMAND this case for trial.