United States v. Dylan Heatherly

PRECEDENTIAL UNITED STATES COURT OF APPEALS FOR THE THIRD CIRCUIT _______________ Nos. 19-2424 & 19-2932 _______________ UNITED STATES OF AMERICA v. DYLAN HEATHERLY, also known as Daniel Sotherland, also known as John Doe-9, Appellant in No. 19-2424 _______________ UNITED STATES OF AMERICA v. WILLIAM STAPLES, also known as Bill Simpson, also known as John Doe-7, Appellant in No. 19-2932 _______________ On Appeals from the United States District Court for the Middle District of Pennsylvania (D.C. Nos. 1:16-cr-00082-010 & 1:16-cr-00082-008) District Judge: Honorable Yvette Kane _______________ Argued: July 8, 2020 Before: McKEE, BIBAS, and FUENTES, Circuit Judges (Filed: January 14, 2021) _______________ Robert J. Daniels [ARGUED] Richard H. Katsifis Killian & Gephart 218 Pine Street P.O. Box 886 Harrisburg, PA 17101 Counsel for Appellant Dylan Heatherly M. Jason Asbell [ARGUED] Gibbel, Kraybill & Hess 2933 Lititz Pike P.O. Box 5349 Lancaster, PA 17606 Counsel for Appellant William Staples Stephen R. Cerutti, II Office of United States Attorney Middle District of Pennsylvania 228 Walnut Street P.O. Box 11754 220 Federal Building and Courthouse Harrisburg, PA 17108 2 Austin M. Berry [ARGUED] United States Department of Justice Criminal Division Child Exploitation and Obscenity Section 1400 New York Avenue, N.W. Washington, DC 20005 Counsel for Appellee _______________ OPINION OF THE COURT _______________ BIBAS, Circuit Judge. Dylan Heatherly and William Staples frequented an inter- net chat room where users regularly shared child pornography. One chat-room user repeatedly live-streamed himself raping and sexually abusing his six-year-old nephew. Heatherly and Staples encouraged him as he did so. And they repeatedly asked users for other child-pornography videos too. A jury convicted the men of receiving child pornography and conspir- ing to do the same. Though they challenge their convictions and sentences on many grounds, we find no error and will affirm across the board. In doing so, we hold that the District Court properly ad- mitted videos shown in the chat room of children suffering vi- olent sexual abuse. After reviewing that evidence for itself, the court properly found that the risk of unfair prejudice did not substantially outweigh its probative value. The evidence was highly probative of the conspiracy and the defendants’ aware- ness of what they were involved in. 3 I. BACKGROUND Child-pornography cases test our legal system’s commit- ment to fairness. That is doubly true of cases involving child rape and sexual abuse. Though the details to follow are unset- tling, to do justice we must describe the facts explicitly, with- out flinching. A. The investigation Long before it became a staple of working from home, Zoom was regularly used for sharing child pornography. Ra- ther than download images or videos, child pornographers and viewers can meet in a Zoom conference room while one user plays a video and shares his screen. The chat function lets them share messages either privately with a single user or publicly with everyone in the room. The child-pornography community has developed its own language and rules of conduct in these rooms. Typically, users require one another to turn their computers’ webcams on so that others can see that they are real users, not bots or police. Users have a shorthand to describe the kind of pornography they seek, like “K-9” for bestiality, “bby” for babies, or “no limit ped perv” for children of any age, even babies. App. 227– 28, 363–64. They use special lingo to approve and encourage sharing contraband, like saying, “hail [user]” after that user shares a video. App. 235. And their aliases themselves often embrace violence and pedophilia, like “Twisted Brutal R4pist,” “babyRaperSnuffer,” and “SEXeducation8-13.” App. 643, 667, 674. 4 This case stems from the undercover work of Detective Constable Janelle Blackadar of the Toronto Police. On the evening of July 22, 2015, she logged into a Zoom conference room that she knew was used to share child pornography. While in the room, she recorded the videos and images that were displayed and saved the public messages. Detective Constable Blackadar watched and preserved sev- eral prerecorded videos showing children being sexually abused. One video showed an eighteen-year-old man named William Augusta (Zoom alias “Guy Johnson”) grabbing his six-year-old nephew (Victim-1), hitting him, and forcing him to perform oral sex. Augusta then live-streamed from his room: as the Zoom users watched and egged him on, he abused his nephew again. After the livestream, users shared more pre- recorded videos, including one showing men raping babies. While Augusta was abusing his nephew live, his chat com- ments suggested that he was in Pennsylvania. So Blackadar reached out to her contacts in the U.S. Government to notify them about what she was witnessing. Federal agents then phoned Zoom’s CEO to help in the investigation. Zoom gave agents the IP address of the user “Guy Johnson.” Police iden- tified him as Augusta, used the Zoom images to identify his grandmother’s house in Carlisle, Pennsylvania, got a search warrant, and found his nephew there. Officers then found Au- gusta at his job and arrested him. Eventually, police managed to trace the IP addresses of many of the users who were in the Zoom room that evening. Federal prosecutors in the Middle District of Pennsylvania in- 5 dicted fifteen defendants. Twelve pleaded guilty, and a thir- teenth died before trial. This appeal focuses on the two defend- ants who made it to trial: Heatherly and Staples. B. The defendants 1. Dylan Heatherly. Heatherly, then 31, lived alone near San Diego. He used the Zoom alias “Daniel Sotherland” and logged into the room while users were sharing prerecorded child pornography. On July 22 at 6:59 p.m. EDT, he posted a public comment in the room: “so appreciate if someone showed vids—need to bust [that is, ejaculate] before work.” App. 678. Less than a minute later, Augusta began streaming his live abuse of his nephew. Within three minutes, Heatherly posted a second public comment: “so close here.” App. 679. The livestream continued for roughly twenty more minutes. Throughout it, other users expressed their approval. Right be- fore Heatherly’s second comment, “dirtypervy” told the other users that the live-rape scene “made me shoot haha.” App. 679. Other users publicly asked Augusta to abuse the six-year-old boy in particular ways: “nyc perv” told Augusta to “get him naked”; “babyRaperSnuffer” said, “smack him around a cou- ple of times”; and “Andy” asked him to “rape him.” App. 679– 81. IP logs linked “Daniel Sotherland” to Heatherly’s address. When federal agents got a search warrant and searched his home, he spoke to them for hours, admitting that “Daniel Sotherland” was his alias and that he watched child pornography. Forensic analysts found multiple child- 6 pornography images as well as a video on his electronic devices. 2. William Staples. Staples, then 56, lived alone in Ken- tucky. Using the alias “Bill Simpson,” he logged into the Zoom room on July 22 about an hour after Augusta’s livestream ended. During his session, Staples posted several comments asking other users to stream videos of babies: “any hot bby vids?”; “anyone have hot bby vids?”; and “any other….bby?” App. 687, 697, 699. After user “cigarffpumpboy” shared a video showing the rape of an infant, Staples replied: “hail cigarffpumpboy.” App. 692. As with Heatherly, federal agents used IP logs to trace “Bill Simpson” to Staples. They got a search warrant, searched his home, and seized electronic devices that had child pornography on them. When agents questioned him, he admitted using the alias “Bill Simpson” and watching child pornography. He also admitted that he knew that users shared child pornography in the rooms. But he maintained, as he would through trial and sentencing, that he was there not to watch child pornography, but only to watch other adult men masturbate. Forensic analysis of Augusta’s and Staples’s devices re- vealed that Staples had also been in the room on February 28, 2015, another time when Augusta had live-streamed his sexual abuse of his nephew. On that date, after Augusta started the stream, Staples sent a message asking him: “u live or is that a vid?” App. 772. Augusta replied: “you of all people asked; u know i’m live.” Id. After learning that he was watching live abuse, Staples encouraged Augusta: “can you walk around with your cock out?” App. 773. 7 3. The charges. Heatherly and Staples were each charged with four counts. One count charged each with receiving or distributing child pornography; a second charged each with conspiring to do the same. 18 U.S.C. § 2252(a)(2), (b)(1). A third pair of counts charged Heatherly and Staples respectively with knowingly publishing a notice or advertisement seeking to receive child pornography, and a fourth count charged each with conspiring to do the same. 18 U.S.C. § 2251(d), (e). C. Trial At trial, the Government argued that the users in the Zoom room, including Heatherly and Staples, met regularly to solicit and share child pornography. Detective Constable Blackadar caught only one of those meetings. But what she saw there sug- gested that there was an ongoing conspiracy. At trial, the Dis- trict Court let the Government introduce several exhibits over one or both defendants’ objections, including: • Exhibits 2–8 and 10, videos of child pornography played in the Zoom room on July 22, as captured by De- tective Constable Blackadar; • Exhibit 17, Zoom activity logs showing aliases and IP addresses of all users who were in the Zoom room on July 22; • Exhibit 21, internet subscriber information showing William Staples’s home IP address; • Exhibit 54, a compilation of Zoom public and private chat messages from July 22, combining information 8 captured by Blackadar that evening with information re- trieved from the defendants’ devices; • Exhibit 55, a spreadsheet listing Zoom public and pri- vate chat messages between January and July 2015, taken from Augusta’s devices; and • Exhibit 61, a spreadsheet listing all Zoom users whose information was found on either Augusta’s, Heatherly’s, or Staples’s devices. Heatherly and Staples called no witnesses. Rather than dis- pute facts, they tried to poke holes in the Government’s legal theories. In his closing argument, Heatherly’s lawyer focused on the conspiracy charges, arguing that the Government had showed only that Heatherly was a “spectator[,] . . . not a conspira- tor.” App. 383. And he challenged whether any of Heatherly’s chats amounted to an ad seeking child pornography. Staples’s lawyer challenged the Middle District of Pennsyl- vania as the wrong venue for prosecuting Staples for publish- ing a notice or ad. On the merits, he noted that on July 22, Sta- ples was not in the Zoom room until more than an hour after the live-streamed rape. And like Heatherly’s lawyer, he ques- tioned the legal theories supporting the conspiracy and adver- tising charges. The jury convicted both defendants of receiving or distrib- uting child pornography and conspiring to do the same. But it acquitted both defendants of publishing a notice or ad seeking child pornography. On the charge of conspiring to publish such 9 a notice or ad, the jury convicted Staples but deadlocked on Heatherly. D. After trial Because the crimes were so serious, the Sentencing Guide- lines recommended imprisoning Heatherly and Staples for forty and seventy years. But the District Court varied down- ward, sentencing Heatherly to twenty-five years and Staples to thirty. On appeal, Heatherly and Staples argue: • The District Court should have dismissed their con- victions because there was insufficient evidence of conspiracy and on venue; • It erred by admitting explicit child pornography into evidence; • It erred in admitting various other exhibits; • It should have severed Staples’s trial from Heatherly’s; • It gave an improper jury instruction on venue; and • It miscalculated their Sentencing Guidelines ranges. We reject all these arguments and will affirm. II. SUFFICIENCY OF THE EVIDENCE Both Heatherly and Staples challenge whether there was enough evidence of conspiracy to support their convictions. 10 We must take the evidence in the light most favorable to the prosecution and ask whether “any rational trier of fact could have found the essential elements of the crime beyond a rea- sonable doubt.” Jackson v. Virginia, 443 U.S. 307, 319 (1979) (emphasis omitted). To prove conspiracy, the Government had to show (1) that two or more people agreed to commit the substantive crime (of receiving child pornography, and in Staples’s case, posting an ad seeking to receive it); and (2) that the defendant joined the agreement knowing of its objective and intending to join with at least one other conspirator to achieve it. See United States v. Rigas, 605 F.3d 194, 206 n.9 (3d Cir. 2010) (en banc); United States v. Navarro, 145 F.3d 580, 593 (3d Cir. 1998); see also Whitfield v. United States, 543 U.S. 209, 213–16 (2005) (hold- ing that federal conspiracy statutes do not require an overt act unless their texts so provide, as in 18 U.S.C. § 371); 18 U.S.C. §§ 2251(e), 2252(b)(1) (not requiring an overt act). A conspiracy does not require an express agreement. A “tacit agreement” is enough. People can tacitly agree when they “engage[ ] as a group” to achieve “a common goal.” United States v. Smith, 294 F.3d 473, 478 (3d Cir. 2002). A group’s “many unusual acts” can be enough evidence to let a jury infer that the group was tacitly working together. Id. Here, there was ample evidence that Staples and Heatherly were part of a conspiracy. 11 A. Staples’s connection to the conspiracy Staples argues that the Government presented no evidence of any agreement to violate 18 U.S.C. § 2251(d). That subsec- tion criminalizes “knowingly mak[ing] . . . any notice or adver- tisement seeking . . . to receive . . . [a] visual depiction in- volv[ing] the use of a minor engaging in sexually explicit con- duct.” In fact, there was plenty of evidence that Staples and the other chatroom users conspired to violate § 2251(d). To start, while Staples was logged on, the chatroom users asked for child pornography countless times. On July 22, users asked the group for “Any more babys,” for “baby vids msg me,” and whether Victim-1 was “still there” to be abused. App. 695, 697. On February 25, users asked for “Any vids,” “any hot vids or pics?,” “any 0 to 3,” and “any vids guys?” App. 771–73. Staples argues that he never agreed with the other users to make those requests, but was just a bystander. There was plenty of evidence to rebut that argument. The chat log shows that the users swapped videos and egged one another on as part of a single, coherent group. E.g., App. 767 (“i shared last. who’s next?”); id. at 776 (“HAIL”). And Staples was an active participant in this group. He publicly celebrated when people posted videos. App. 692 (“hail cigarffpumpboy”). When there was a lull in the chatroom, he privately complained that “no one [was] shareing.” App. 771. And within moments of other people’s requests, he made nearly identical requests of his own. E.g., App. 687 (“any hot bby vids?”); id. at 699 (“any other….bby?”); id. at 771 (“did you have some hot ones to 12 share?”). A jury viewing this evidence could certainly con- clude that Staples was tacitly working with other members of the group to ask for and get child pornography. There was other circumstantial evidence of a conspiracy too. The users congregated in chat rooms that required a ten- digit room ID, so a user could not stumble into the room acci- dentally. And they protected their crimes by requiring every- one logged in to have his webcam on. These security measures suggested that the users were working together; at least, they had to tell each other what the password was and to turn on their cameras. It does not matter that Staples never explicitly agreed with another user to post a notice seeking child pornography. He was part of a group of like-minded people who got together repeatedly to do just that. The shocking videos, pictures, and comments discussed below further support the jury’s finding that by frequenting this room, Staples agreed to the overarch- ing conspiracy. Staples’s challenge to the evidence of venue for that con- spiracy likewise fails. Prosecutors must prosecute a crime in a district where the crime was committed. U.S. Const. art. III, § 2, cl. 3; id. amend. VI; Fed. R. Crim. P. 18. The Government bears the burden of proving that venue is proper for each count, but only by a preponderance of the evidence. United States v. Root, 585 F.3d 145, 155 (3d Cir. 2009). Venue is proper in any district where a defendant did any of the “crucial elements” of the crime. Id. at 156; 18 U.S.C. § 3237(a). For conspiracy, venue lies “wherever a co-conspirator has committed an act in furtherance of the conspiracy.” United States v. Perez, 280 13 F.3d 318, 329 (3d Cir. 2002) (citing Hyde v. United States, 225 U.S. 347, 363–64 (1912)). There was enough evidence linking Staples to the Middle District of Pennsylvania. Staples argues that he was not logged in during Augusta’s July 22 sexual abuse. But that does not matter. “Although [Staples] himself did not act in the [Middle] District of Pennsylvania or direct any of his actions there, [a] co-conspirator[ ] ,” Augusta, live-streamed the video from there. United States v. Renteria, 903 F.3d 326, 331 (3d Cir. 2018). Augusta’s “overt act[ ] in furtherance of the conspiracy [is] certainly sufficient to establish venue.” Id at 331–32. B. Heatherly’s connection to the conspiracy Heatherly argues that he was just a viewer, not a member of the conspiracy. We take that argument as a concession on the substantive charge of receiving child pornography. Be- cause he never sought a judgment of acquittal on the conspir- acy charge, we review for plain error. There was none. Heatherly’s claim fails for the same reasons that Staples’s does. Indeed, unlike Staples, Heatherly was in the room and commenting while Augusta live-streamed his rape of his nephew on July 22. We need not canvas all the other evidence here; we explore much of it below. III. ADMISSION OF THE CHILD PORNOGRAPHY At trial, the Government introduced and showed the jury condensed versions of videos shown in the Zoom room on July 22. These videos, Heatherly and Staples object, were inadmis- sible under Federal Rules of Evidence 402 and 403. Rule 402 14 requires courts to exclude irrelevant evidence; Rule 403 lets them exclude some relevant but unfairly prejudicial evidence. The jury also learned of child-pornography images and videos found on their devices. Heatherly and Staples claim that the District Court should have excluded that evidence under Rules 403 and 404(b) (which requires courts to exclude some evi- dence of bad acts). A word before we go on: The descriptions of the Zoom vid- eos that follow are horrifying. Though often we can resolve legal issues “without subjecting the reader to the graphic and disturbing details of the [child] pornography,” those details are unavoidable when we confront a fact- and context-specific challenge under Rule 403. United States v. Cunningham, 694 F.3d 372, 377 n.8 (3d Cir. 2012). “To say that the . . . descrip- tions of the video excerpts are loathsome is an understate- ment.” Id. at 381 n.10. Even so, we cannot shy away from scru- tinizing them. A sanitized description would obscure whether the videos were unfairly prejudicial. Thus, we must confront the troubling details head-on. A. The Zoom videos At the start of trial, Detective Constable Blackadar de- scribed the eight challenged videos (ages and times are approximate): • Exhibit 2 runs for two minutes and shows a man vag- inally penetrating a four-year-old girl. The jury watched a five-second clip of it at normal speed. 15 • Exhibit 3 is a five-and-a-half-minute-long compila- tion of “hurtcore,” a subgenre involving the “torture or pain and the sexual assault of children.” App. 236. Many of the clips show a man vaginally or anally raping a three- or four-year-old girl. One clip shows a man urinating on that toddler’s face. Yet another shows a man ejaculating on the vaginal area of a four-to-six-year-old girl. The jury watched this video on fast forward for thirteen seconds. • Exhibit 4 runs for a minute and a half and shows a man anally raping a four-to-six-year-old boy. The Government did not show this video to the jury. • Exhibit 5 runs for almost eight minutes and shows an eight-year-old girl performing oral sex on a man. Near the end, the man forces his penis into her mouth and then ejaculates on her face. The jury watched part of this video on fast forward for twelve seconds. • Exhibit 6 runs for three and a half minutes, showing Augusta grabbing, hitting, and forcing his nephew to perform oral sex on him. The jury watched a ten- second clip at normal speed. • Exhibit 7 is a nine-minute recording of Augusta’s live abuse of his nephew. Blackadar captured only part of the twenty-two-minute livestream. During that part, six-year-old Victim-1 performs oral sex on Augusta. Augusta then puts his nephew in his lap, masturbates him, penetrates the boy’s anus with his 16 finger, and then puts that finger into the boy’s mouth. The jury watched the first two minutes of that video on fast forward over thirteen seconds. • Exhibit 8 is a three-and-a-half-minute-long compi- lation of adults sexually abusing prepubescent chil- dren. The Government did not show this video to the jury. • Exhibit 10 is a thirteen-minute-long compilation of adults sexually abusing babies. Parts show men forc- ing their penises into their victims’ mouths. One part shows a tied-up baby being anally penetrated. The jury watched the first five seconds at normal speed, then a seven-minute clip on fast forward over the course of ten seconds. No audio was played with any of the videos. Heatherly and Staples make three arguments why the Dis- trict Court should have excluded these exhibits. First, the Gov- ernment could not prove that Heatherly watched any of the vid- eos on July 22; indeed, he was not even logged into the room when most of them were shown. Staples likewise was not logged in on July 22 while any of the videos played. Thus, they argue, the videos were irrelevant. Second, in deciding whether the videos were admissible un- der Rule 403, the District Court never explained on the record how it balanced the videos’ probative value against their po- tential for unfair prejudice. Heatherly and Staples argue that this failure was a per se reversible error. And third, they claim 17 the District Court abused its discretion by letting the jury see these “shockingly prejudicial” videos. Heatherly Br. 22. We review the District Court’s admission of this and all ev- idence for abuse of discretion. United States v. Tyson, 947 F.3d 139, 142 (3d Cir. 2020). On this record, we find no abuse of discretion. 1. The videos were relevant. The Government’s theory of the conspiracy was that over several years, Heatherly, Staples, and many other coconspirators would meet online to share and watch child pornography. To prove an illegal conspiracy, the Government had to show that the users had a “tacit agreement.” Appellee’s Br. 26. And the more taboo the events in the Zoom room, the more likely that such a tacit agreement existed. That is especially true when one considers the alternative: that these men, all interested in illegal child pornography, happened to wander into the same private Zoom room at the same time. The challenged exhibits show how taboo the room was and what happened there, whether or not Heatherly and Staples saw these exact videos on that particular evening. They thus make it more likely that those present shared a tacit agreement. Fed. R. Evid. 401. So they are relevant. 2. Failure to articulate Rule 403 balancing was not re- versible error. District courts may exclude evidence “if its pro- bative value is substantially outweighed by a danger of . . . un- fair prejudice.” Fed. R. Evid. 403. When a district court faces a Rule 403 objection, it must balance these two competing fac- tors. See United States v. Finley, 726 F.3d 483, 491 (3d Cir. 2013). It should explain its reasoning on the record; when it does so, “we will rarely disturb its ruling.” Id. (quoting United 18 States v. Sampson, 980 F.2d 883, 889 (3d Cir. 1992)). But we do not prescribe exactly what it must say, so long as it makes clear that it balanced the Rule 403 factors. We have affirmed a district court’s overruling of a Rule 403 objection even when it just said that the probative value outweighed the risk of un- fair prejudice. Id. We prefer that the district court show its work. That is the wiser and sounder course. But we will affirm so long as it makes clear that it did the weighing itself. Compare, e.g., id. (affirming because it was clear that the district court did the balancing), with Sampson, 980 F.3d at 889 (reversing because it was not clear that the court did the analysis at all), and Gov’t of the V.I. v. Pinney, 967 F.2d 912, 918 (3d Cir. 1992) (same). Here, the District Court said just enough to confirm that it did the Rule 403 balancing. Before trial, the lawyers argued back and forth about Rule 403. At trial, when the Government first tried to introduce the videos, both Heatherly and Staples objected that the videos would be “highly prejudicial.” App. 244. The District Court acknowledged the Rule 403 objection and heard further argument from defense counsel and the Gov- ernment. It then held that the Government had shown that the videos would not unduly prejudice Heatherly, but deferred its ruling pending further foundation as to Staples: “[B]efore we even get to the [Rule] 403 analysis, I want to know when it is you’re planning to show [the videos’] connection” to Staples. App. 244. Near the end of trial, after laying that foundation, the Government again moved to admit Exhibits 2–8 and 10. The court noted the previous objection, said it was preserved, and admitted the videos. True, it did not articulate its Rule 403 19 analysis or mention probative value and prejudice. But it did refer back to the Rule 403 analysis that it had promised to do once the Government laid more of a foundation. In context, we see that the court did the analysis and con- cluded that Rule 403 did not bar the evidence. As we said in Finley, the more a district court says, the better. See 726 F.3d at 491. But as we also said in Finley, a district court need say only enough to show that it weighed the Rule 403 factors. It said just enough here. 3. The District Court did not abuse its discretion by admit- ting the exhibits. District courts deserve broad deference in ap- plying Rule 403. The rule is written in discretionary terms: “The court may exclude . . . .” Fed. R. Evid. 403 (emphasis added). And the caselaw “stress[es] the extraordinary breadth of discretion that this Rule invites.” Christopher B. Mueller & Laird C. Kirkpatrick, 1 Federal Evidence § 4:12 (4th ed. 2020). So we normally reject bright-line rules about what evidence must be excluded under Rule 403. See Cunningham, 694 F.3d at 391 (declining to adopt a bright-line rule excluding poten- tially inflammatory child pornography). Rather, we must judge each case on its own facts and record. Another point bears note. Rule 403 bars not all prejudice, but only unfair prejudice. United States v. Bergrin, 682 F.3d 261, 279 (3d Cir. 2012). It does not protect defendants from devastating evidence in general. The exhibits were disturbing because the alleged crimes themselves were extraordinarily disturbing. Rule 403 is not a shield to keep juries from learning details of horrific crimes. United States v. DeMuro, 677 F.3d 550, 559 (3d Cir. 2012). 20 These videos were highly probative. Proving a conspiracy required showing more than that someone played child pornog- raphy on July 22. The Government had to prove that the room was a haven for child pornography, a place where pedophiles would keep gathering to share in their vile pastime. Yet be- cause the undercover agent was there only one time, the Gov- ernment could not prove directly that anyone played child por- nography at other times. For those other times, all the Govern- ment had were Zoom usernames and comment logs. So it needed to give the jury a snapshot so that the jury could match the commentary to the visual events. That snapshot was the evening of July 22. Having heard descriptions and seen clips of the child pornography streamed that evening, the jury could better grasp the scope of the alleged conspiracy and the tacit agreement driving it. To do that, it had to see those videos firsthand. And jurors had to see the videos themselves, not just hear descriptions or stipulations, to appreciate that anyone who fre- quented that room must have approved of the images shared there. Both defendants claimed at some point that they were not really interested in child pornography but visited the room because they wanted to watch other men masturbate. If the jury had believed those defenses, it might have doubted whether the defendants really agreed to request child pornography. But the horrific videos demolished those defenses. They proved that Heatherly and Staples would not have frequented the room un- less they wanted to watch images of children being sexually abused. No one would have gone back to the room innocently, thinking that the images the first time around were borderline or a fluke. 21 The probative value here was thus greater than in run-of- the-mill cases of simple possession of child pornography. In those cases, defendants often concede that anyone who watched the videos would know that they were child pornog- raphy. Instead, the only factual dispute is whether the defend- ant knew the files were on the computer. See, e.g., United States v. Welshans, 892 F.3d 566, 571 (3d Cir. 2018); Cunning- ham, 694 F.3d at 379–80. The videos here did much more than that. True, some of our recent Rule 403 analyses of child por- nography have come out differently. In Cunningham and Welshans, we held that the district courts erred in admitting the evidence. 694 F.3d at 383; 892 F.3d at 576. But Rule 403 anal- ysis is always fact- and context-specific. And each case is dis- tinguishable in several important ways. Start with Cunningham. That case involved no conspiracy charges. The district court had let the jury watch excerpts of two graphic child-pornography videos. 694 F.3d at 379–82. The court overruled a Rule 403 objection, relying only on the Government’s descriptions of the videos and without watching the videos itself before it ruled. Id. at 379. We held that it had abused its discretion, agreeing with several other circuits that a trial court must see the challenged exhibits for itself. See id. at 383–87. And because the court had not watched the videos, we did not afford it the usual Rule 403 deference. See id. at 388. That procedural error was intertwined with the substantive er- ror and crucial to Cunningham’s outcome. Id. at 392; see also Welshans, 892 F.3d at 575. 22 On our independent review, we held that the court should have excluded the challenged videos. Cunningham, 694 F.3d at 388. The exhibits were probative, as they were the actual child pornography that the defendant had possessed, and their lurid content tended to show that he knew what they depicted. See id. at 389. But because neither issue was disputed at trial, that probative value was minimal. See id. at 377–80. So their admission already tested the limits of the Old Chief maxim that the Government may put on evidence to prove its case and need not accept an offer to stipulate. See Old Chief v. United States, 519 U.S. 172, 189 (1997). And given the limited purpose of introducing the videos, we saw no reason why the Government needed to show so many clips or why the excerpts needed to be so shocking and violent. See Cunningham, 694 F.3d at 389– 90. Here, by contrast, the District Court did review the chal- lenged evidence before admitting it. So we give its ruling sub- stantial deference. Plus, the nature, content, and volume of vid- eos prove Heatherly’s and Staples’s involvement in a broader conspiracy in the Zoom room. That was not an issue in Cun- ningham. And the risk of unfair prejudice was much lower here than in Cunningham. The jury had to understand the videos’ graphic details to decide whether there was a conspiracy, so at some point it had to hear brutal descriptions of abuse. That was already going to inflame the jury’s emotions. The marginal prejudice from seeing short, and in many cases fast-forwarded, video clips, after hearing detailed descriptions, was thus lower. Welshans, which relied heavily on Cunningham, is likewise not on point. See Welshans, 892 F.3d at 575–76. That case also 23 involved child-pornography videos played for the jury. But the crux of the appeal was repeated, detailed descriptions of other child pornography that the defendant allegedly possessed. See id. at 571–73, 575. Though we reiterated that there are no per se rules on this issue, we held that admitting the descriptions violated Rule 403. Id. at 575–76 (citing Cunningham, 694 F.3d at 391). A key to the outcome in Welshans, as in Cunningham, was that “the Government had extensive evidence that did not in- volve violent or sadistic content, and Welshans stipulated that the files recovered were child pornography.” Welshans, 892 F.3d at 576. The descriptions were relevant only to show that the contraband was child pornography, an issue no one dis- puted. Here, however, the videos were highly relevant to prove the conspiracy, which was hotly disputed. And here, the Gov- ernment did not choose the most disturbing descriptions and videos from a mountain of evidence. On the contrary, the only images the Government had from the Zoom room were the handful of videos that Detective Constable Blackadar recorded on July 22. 4. Any error was harmless. Even if admitting the videos was an error, substantively or procedurally, there was no prej- udice. Heatherly worries that “[a]nger is a wind that can extin- guish the flame of reason.” Heatherly Br. 2. But here, the winds of anger did not blow out that candle or dim its flame. As Heatherly admits, the jurors “were able, to some extent, to di- vorce their natural emotion from reason.” Id. at 3. The jury ev- idently considered the evidence separately on each count and for each defendant. United States v. Franz, 772 F.3d 134, 153– 24 54 (3d Cir. 2014). It convicted on some counts, acquitted on others, and on one count reached different conclusions about each defendant. This is not the mark of a jury blinded by rage. See id. (explaining that in context, the split jury verdict “strongly suggest[ed] that the jury was . . . not swayed by the prejudicial character of the” child pornography); United States v. Pelullo, 14 F.3d 881, 899 (3d Cir. 1994) (citing with ap- proval Judge Friendly’s treatment of the jury’s “discriminating acquittal on one of the counts” as “evidence that the jury was able to overcome any prejudic[ial taint]” in United States v. Ivic, 700 F.2d 51, 65 (2d Cir. 1983)). And lots of other evi- dence, including the defendants’ graphic messages and the de- tective’s descriptions of the videos, supported the convictions too. See United States v. Vosburgh, 602 F.3d 512, 540 (3d Cir. 2010). In closing, we again urge trial courts to articulate their Rule 403 analyses on the record. That is especially important for in- flammatory evidence. If they do so, we will rarely disturb their rulings. Still, we can affirm as long as the trial court clearly did the Rule 403 analysis. Here, the District Court did. B. Child pornography from the defendants’ computers Heatherly and Staples each challenge the admission of child pornography found on their electronic devices. We reject those challenges. 1. Images from Heatherly’s devices. The District Court ad- mitted Exhibits 33A–F, child pornography found on his de- vices. Heatherly argues that doing so amounted to using other 25 acts as character evidence, in violation of Federal Rule of Evi- dence 404(b)(1). He asserts that the evidence had little rele- vance to him because (unlike Staples) he never claimed at trial that he was in the room to watch other men. He also says that there was “scant” evidence that he viewed or even knew about these images. Heatherly Br. 30. And he argues that the court erred by not weighing the Rule 403 factors on the record. Heatherly’s Rule 404(b) argument is misguided. Though Rule 404(b) does exclude evidence of prior bad acts like un- charged crimes, Rule 414 carves out an exception for child mo- lestation prosecutions (including child pornography). All the crimes charged here fall within that exception. See Fed. R. Evid. 414(d)(2)(B), (F). The evidence was relevant to show his interest in child por- nography. That supported the Government’s theory that he (and Staples) sought to receive child pornography by posting comments in Zoom rooms, an element of the notice-or-adver- tisement count. It also rebutted Heatherly’s pretrial statement that he went to the room to watch adult men masturbate, not to watch child pornography. So while the evidence was more pro- bative for Staples because he denied any sexual interest in chil- dren, it was probative for Heatherly too. As for whether Heatherly ever watched or even knew about the images on his devices, that goes to the weight of the evi- dence, not its admissibility. He was free to deny his awareness of them but chose not to. Finally, the District Court did a Rule 403 analysis. Before trial, it deferred ruling on admissibility until the court could 26 analyze at trial “the relative weights of [their] probative value and prejudicial effect under Rule 403.” D.C. Dkt. No. 836, at 5. True, when it later admitted the evidence, it did not explain its logic or mention the Rule 403 factors. But nothing suggests that it forgot its promise to balance them. For the reasons given earlier, we see just enough confirmation that the District Court balanced those factors. In any event, as explained, any error was harmless. 2. Images from Staples’s devices. Staples makes similar ar- guments against admitting the images that were found on his computer (Exhibits 62A–I). The District Court did not abuse its discretion in admitting them. For Staples, the Rule 403 analysis is even simpler: the evi- dence rebutted his defense. Staples claimed he was a gay man who went to Zoom rooms just to watch men masturbate and had no interest in child pornography. He opened the door to this evidence and cannot complain that the Government walked through it. Staples also argues that the court should have excluded the images because he offered to stipulate that they were child por- nography. But an offer of stipulation does not make evidence inadmissible. “As the Supreme Court remarked in Old Chief, ‘[a] syllogism is not a story, and a naked proposition in a court- room may be no match for the robust evidence that would be used to prove it.’ The government is thus entitled to put forth the relevant evidence that it chooses.” Finley, 726 F.3d at 492 (alteration in original) (citation omitted) (quoting Old Chief, 27 519 U.S. at 189). And Staples does not explain how this evi- dence unfairly prejudiced him. Indeed, at this trial, the jury saw far worse. IV. OTHER EVIDENTIARY RULINGS Besides the videos and child pornography, the Government also introduced other evidence of the Zoom activity of Heatherly, Staples, and their coconspirators. The District Court properly admitted this evidence. A. Zoom chat messages and user-avatar data Heatherly and Staples object to three forensic compilations of Zoom data: Exhibits 54, 55, and 61. But they have forfeited these objections. Heatherly’s opening brief spends just one page discussing them and does not cite a single legal authority for why it was error to admit them. Staples has even less to say, simply joining Heatherly’s arguments. By not developing these arguments properly, Heatherly and Staples have forfeited them. See Sikirica v. Wettach (In re Wettach), 811 F.3d 99, 115 (3d Cir. 2016). B. Zoom user-activity logs The Government introduced a list of user logs subpoenaed from Zoom as Exhibit 17. The list showed which users were in the room during the relevant times on July 22, 2015. The logs are hearsay. But they are classic business records (a hearsay exception): they were made contemporaneously based on in- formation from someone with knowledge; they were kept in the course of Zoom’s regular business activity; and keeping those records was Zoom’s regular practice. Fed. R. Evid. 28 803(6)(A)–(C). Plus, the Government introduced Exhibit 17A, a certification by the custodian of records that the records meet those requirements. Id. R. 803(6)(D), 902(11). So there is no merit to Heatherly and Staples’s objection that the agent on the stand had no personal knowledge of Zoom’s recordkeeping practices. Heatherly also misreads Exhibit 17A as saying that Exhibit 17 was merely based on business records and is not itself a business record. That is not what it says. The custodian certi- fied not that the records (Exhibit 17) are based on a review of business records, but rather that the certification itself (Exhibit 17A) is based on a review of Zoom records. He added that Ex- hibit 17 reflects “business records which are created contem- poraneously with the activity,” “are kept in the ordinary course of Zoom’s business and pertain to the regular business of Zoom.” Exhibit 17A (App. 658). That satisfies Rule 803(6) and 902(11). C. IP information Staples, but not Heatherly, challenges the admission of Ex- hibits 21, 21A, 41, and 41A. In Exhibit 21, Staples’s internet provider ViaSat links Staples with the Zoom alias “Bill Simp- son.” Exhibit 21A certifies that Exhibit 21 is a business record. Exhibit 41 is a log of IP addresses and usernames from a Zoom room in December 2015; Exhibit 41A is Zoom’s certification of it as a business record. Staples argues that Exhibit 21 is signed not by a ViaSat em- ployee but rather one who worked for a firm that manages other 29 companies’ business records. But nothing in Rule 902(11) re- quires that the records custodian be an employee of the same company. Nor, contrary to Staples, do the rules require the cus- todian to explain in detail who compiled the records, how, or how she knows they are accurate. As for Exhibits 41 and 41A, Staples contests not their sub- stance, but only that the Government updated its certification too close to trial. But he did not object at trial, so we review only for plain error. And though he asserts that the update prej- udiced him, he never explains how it did or what he would have done differently if he had gotten it earlier. He has not shown that any error affected his substantial rights. See United States v. Greenspan, 923 F.3d 138, 147 (3d Cir. 2019). V. SEVERANCE Before trial, Staples moved to sever his trial from Heatherly’s. The District Court reasonably denied that motion. We review the District Court’s denial of a severance for abuse of discretion. United States v. Eufrasio, 935 F.2d 553, 568 (3d Cir. 1991). We presume that courts will try codefendants jointly, especially when both are charged in a single conspiracy. See id. To overcome that presumption, “[a]n appellant’s burden is heavy: he must demonstrate ‘clear and substantial prejudice resulting in a manifestly unfair trial.’ ” Id. (quoting United States v. Reicherter, 647 F.2d 397, 400 (3d Cir. 1981)) (emphases omitted). It is not enough to show that a severance would have increased the chance of acquittal, or that some evidence applied to some defendants more than others or was more damaging to some defendants. Id. 30 Staples’s severance argument centers on Exhibit 7, which showed Augusta’s live-streamed abuse of his nephew on July 22, 2015. He stresses that this video, “the centerpiece of the trial,” related only to Heatherly, since Staples was not in the room when Augusta streamed it. Staples Br. 21. That objection fails. The Government introduced the video not to show what each defendant was doing that night, but to prove the pe- dophilic culture that the conspirators shared. Staples likewise objects to several images and chats seized from Heatherly’s computer. But the court gave a proper limit- ing instruction, telling the jury to consider the evidence against each defendant and on each count separately. And the jury would have had no difficulty keeping the evidence from each defendant’s computer separate. In short, the court did not abuse its discretion by denying the requested severance and trying the two coconspirators together. VI. THE VENUE INSTRUCTION Neither Heatherly nor Staples is from the Middle District of Pennsylvania or was there during the conspiracy. Rather, they were convicted based solely on Augusta’s live broadcasts of abuse from there. At trial, Staples argued that venue was thus improper in the Middle District. The District Court instructed the jury that for venue to be proper, “the government must convince you that some act in furtherance of the crimes charged” happened in the district. App. 396. Staples asked the court to insert the words “on a par- ticular count” before that phrase, but the court denied that re- 31 quest. App. 1160–61. On appeal, he renews his request and ob- serves that the Third Circuit’s model jury instruction uses the singular “offense” and “crime” rather than the plural. Third Circuit Manual of Model Criminal Jury Instructions § 3.09 (2018). Using the singular, he argues, would have clarified that the jury must find venue separately on each count. We agree that the use of the plural could have misled a jury into thinking that venue was proper on all charges so long as a crucial element of one charge happened there. Using the sin- gular would have prevented any possible confusion. Still, the instruction made no difference. See United States v. Guada- lupe, 402 F.3d 409, 415 (3d Cir. 2005) (requiring proof that an erroneous jury instruction so influenced the jury’s delibera- tions that it “produce[d] a miscarriage of justice”). In his clos- ing argument, Staples’s counsel contested venue only on the count of knowingly publishing an ad seeking child pornogra- phy. No one mentioned venue on any of the other charges. And the jury acquitted Staples on that one count, even though it con- victed him on all the others. Evidently, the use of the plural did not produce a miscarriage of justice. So this argument fails. VII. SENTENCING The District Court sentenced Heatherly to twenty-five years’ imprisonment and Staples to thirty. Both now challenge their sentences. Each asserts that the court should not have ap- plied the guideline for involvement in producing child pornog- raphy. Heatherly also argues that his sentence was far more se- vere than that of a codefendant, whom he claims was compa- rable. In making that claim, he insists that the sentencing court 32 mistakenly considered conduct of which he had been acquitted. These arguments fail as well. A. The district court applied the right guideline The applicable guideline for receiving, possessing, adver- tising or soliciting child pornography is U.S.S.G. § 2G2.2. Or- dinarily, that guideline sets a base offense level of 22. § 2G2.2(a)(2). But a cross-reference provides that if the crime “involved causing, . . . or seeking by notice or advertisement, a minor to engage in sexually explicit conduct . . . for the purpose of transmitting a live visual depiction of such conduct,” the guideline for producing child pornography applies. § 2G2.2(c). That guideline sets a much higher base offense level of 32, add- ing decades to a defendant’s recommended sentence. § 2G2.1(a). The more serious guideline applied here. Heatherly and Staples’s sentences properly depended not only on their own conduct, but also on that of their coconspirators. Under the rel- evant-conduct guideline, the court had to consider the acts of coconspirators that were “within the scope of the jointly under- taken criminal activity,” “in furtherance of” it, and “reasonably foreseeable.” U.S.S.G. § 1B1.3(a)(1)(B). Coconspirator Au- gusta produced a “live visual depiction” of child pornography at least twice, on February 28 (when Staples was in the room) and July 22 (when Heatherly was). § 2G2.2(c)(1). So the only question is whether Heatherly and Staples agreed with Augusta to have him live-stream child pornography. The District Court reasonably found that they did. 33 Heatherly argues that he sought images only of past abuse. By asking for “someone [to] show[ ] vids,” he argues, he was not seeking live abuse, because “vid[eo]s” supposedly implies recordings. App. 678. Not so. Videos can be live. A video is just a set of “images for display on a television screen or other electronic device.” Video (def. 1a), Oxford English Dictionary (3d ed. 2016). Indeed, we held argument in this very case by video: that is, a live videoconference. And when Augusta started to stream a video, Heatherly encouraged him to con- tinue because he was “so close here.” App. 864. The evidence for Staples was even clearer. Staples specifi- cally asked Augusta whether he was live. When Augusta said he was, Staples egged him on to “walk around with [his] cock out.” App. 773. Staples objects that the District Court made no factual find- ings specifying how much he agreed with, was involved in, and could have reasonably foreseen the conspiracy’s crimes. But at sentencing, the District Court specifically rejected Staples’s portrayal of his own conduct, finding that he “encouraged [Au- gusta’s] behavior and was an active participant in it.” App. 479. It adopted the position advanced by the Government and ac- cepted by the presentence report that Staples knew the object of the conspiracy and encouraged Augusta’s live abuse. The factual findings in the report suffice to support the court’s reli- ance on the conspiracy. See United States v. Collado, 975 F.2d 985, 990 (3d Cir. 1992). 34 B. The sentencing court properly sentenced Heatherly more harshly than a less-culpable coconspirator Finally, Heatherly raises a sentencing disparity. He objects that his sentence was almost four times as long as that of his coconspirator Ed Westbury, who pleaded guilty to conspiring to distribute child pornography and received only six-and-a- half years. But the District Court saw that discrepancy and grappled with it thoughtfully. It compared Heatherly’s conduct to that of twelve other defendants who had already been or would be sentenced. Most had pleaded guilty and been sen- tenced to decades in prison. And it evidently saw Westbury as an “outlier” not only because he had pleaded guilty, but also because he had sought “teen vids” rather than those of prepu- bescent children. App. 453, 455, 459. The sentencing court treated Heatherly in line with most of the other defendants. At the same time, it saw that he was nei- ther “the least [culpable] among them” nor “the most culpa- ble.” App. 460. He was less culpable than several other defend- ants who had been sentenced to 30 to 40 years. In particular, the court noted that another man had “telephoned the little boy[,] terrorizing him.” App. 456. That man had been sen- tenced to 30 years, so the court reasoned that Heatherly’s sen- tence should be somewhat less. The sentencing court put enormous care and thought into distinguishing the defendants based on their conduct. As it ex- plained: “It’s very difficult to draw these distinctions, but we have to do it. And, in fact, based on what we know about [Heatherly’s] conduct, he is less culpable than some others who were in that room, and he needs to be treated as such.” 35 App. 460. The court did treat him as such, varying downwards from the forty-year sentence recommended by the Guidelines to twenty-five years. The punishment, it judged, fit the severe crimes. We respect that considered judgment. In trying to prove a sentencing disparity, Heatherly com- plains that the District Court must have considered conduct of which he was acquitted. But he cites nothing in the record sug- gesting that. In any event, the Supreme Court has held that sen- tencing courts may consider acquitted conduct at sentencing. United States v. Watts, 519 U.S. 148, 149, 155–57 (1997) (per curiam) (explaining that an acquittal shows only reasonable doubt, which is consistent with the preponderance-of-the-evi- dence threshold at sentencing). The sentence was proper. * * * * * The child-pornography videos were essential to prove the tacit conspiracy, the culture of the Zoom room, and Heatherly and Staples’s awareness of what they were seeking and encour- aging. The District Court properly admitted those videos and other evidence, denied a severance, found the evidence suffi- cient, and sentenced Heatherly and Staples based on their roles in instigating sexual abuse. Though the venue instruction was not ideal, in context the jury grasped it and was careful enough to acquit on some counts while convicting on others. We will affirm. 36 McKee J., concurring. The importance of subjecting inflammatory evidence to the Fed. R. Evid. 403 balancing test cannot be overemphasized. This is particularly true when the evidence in question is as potentially inflammatory as it was here. My colleagues do not overstate the nature of these video clips in citing to our statement in United States v. Cunningham.1 As the Majority explains, we there stated: “[T]o say that the . . . descriptions of the video excerpts are loathsome is an understatement.”2 Accordingly, my colleagues quite correctly, and without exaggeration, warn that the descriptions which they present are “horrifying.”3 Regrettably, those descriptions are nonetheless necessary for the reasons my colleagues so carefully explain. This tension between the potential for unfair prejudice on the one hand, and substantial probative value on the other, results in a Rule 403 balance that is as difficult as it is delicate in these kinds of cases. Although the District Court was aware this balance must be struck under Rule 403 before admitting such evidence, I am not convinced that the court actually undertook the precarious task of weighing the probative value of this evidence against its obvious potential to generate animus and undue prejudice. The court did note the 403 objection and was therefore clearly aware of that issue as my colleagues explain. However, being aware of the Rule 403 issue does not necessarily establish that the required balancing was actually undertaken by the District Court. I nevertheless concur in the Majority Opinion for two reasons. First, the fact that the jury did not convict defendants of all charges means that jurors’ emotions where not overwhelmed, and they were nevertheless able to thoughtfully evaluate the evidence that was introduced. Thus, any error in admitting the videos was harmless. Second, we have held that when the trial court fails to conduct this analysis, we may do so on review.4 As I shall explain, given the charges here, I believe the probative value of the evidence 1 694 F.3d 372, 377 n.8 (3d Cir. 2012). 2 Maj. Op. 15-16. 3 Id. at 16. 4 See United States v. Eufrasio, 935 F.2d 553, 572 (3d Cir. 1991). 1 did outweigh its prejudicial potential. Thus, the trial court did not err in allowing it. I. Harmless Error The District Court did state “before we even get to the 403 analysis, I want to know when it is, you’re planning to show [the connection to Staples].” However, the court later admitted the exhibits without conducting any further analysis on the record.5 The government argues that the District Court conducted the analysis because the court “stated that it had reviewed the exhibits and recognized the need for it to conduct a Rule 403” analysis at some unspecified time in the future.6 Although the District Court thus clearly “recognized it needed to perform” the analysis, there is no indication it later did so.7 Nevertheless, I conclude any error would have been harmless even though I reject the government’s suggestion that the impact of the videos was mitigated by partially fast- forwarding through them.8 It by no means follows that fast- forwarding through disgusting video images makes them any less disgusting or shocking. There is absolutely nothing in the record to support such conjecture, and the government has not offered anything to us to support it. Indeed, one could just as forcibly argue that fast-forwarding these images would make them even more repulsive. Moreover, the twenty seconds of regular speed footage included approximately ten seconds of “hurtcore” porn involving a five or six-year-old boy being hit and grabbed by the neck.9 Heatherly argues with some force that the kind of anger that will necessarily be cultivated in the minds of jurors viewing the kind of unspeakable cruelty and depravity in these 5 Appx 244; see also Appx 365 (admitting exhibits without mentioning 403, probative value, prejudice, etc.); Heatherly Br. at 19-20 (the only comment regarding 403, noted above, was on day two and the exhibits were admitted without further discussion on day five.). 6 USA Br. at 28. 7 USA Br. at 29. 8 USA Br. at 30-31. 9 Heatherly Br. at 23-24. 2 video clips is of such force that it is akin to “a wind that can extinguish the flame of reason.”10 However, as my colleagues explain, this jury did not convict either defendant on all counts in the indictment.11 Thus, the flame of reason continued to flicker, though it may well have been stressed. The Majority Opinion succinctly focuses on the significance of that fact and its relevance to our Rule 403 inquiry.12 The jury’s ability to reason and individualize the evidence was not overwhelmed by the emotion, anger, or animosity that the video clips almost certainly engendered against these defendants. The jury refused to convict the defendants of publishing a notice or advertisement to receive child pornography.13 Absent that consideration, the risk that the jurors’ reason was overwhelmed would have been too great to sustain the decision to admit this evidence. Nevertheless, given the verdicts here, and assuming we can perform the Rule 403 balance on appeal, I conclude that the trial court did not abuse its discretion in admitting this evidence over the 403 objections.14 10 Id. (citing to Heatherly Br. at 2). 11 Heatherly and Staples were charged with the following criminal Counts: (11) Conspiracy to Receive or Distribute Child Pornography” in violation of 18 U.S.C. § 2252(a)(2) and § 2252(b)(1); (12) Receipt or Distribution of Child Pornography in violation of 18 U.S.C. § 2252(a)(2), § 2252(b)(1), and § 2; (13) Conspiracy to Publish a Notice or Advertisement of Child Pornography 18 U.S.C. § 2251(d); (16) and (17) Publishing a Notice or Advertisement to Receive Child Pornography 18 U.S.C. § 2251(d). Appx 5-12. Dylan Heatherly was found guilty of Counts 11 and 12 and not guilty of Counts 13 and 17. Appx 13. Staples was found guilty of Counts 11, 12, 13 and not Count 16. Appx 20-21. 12 Maj. Op. at 25. 13 Appx 8; 12. 14 I use conditional language because it is not clear to me that our practice of conducting an independent review under Rule 403 when the District Court fails to do so is consistent with Supreme Court precedent. Eufrasio was decided in 1991. Thereafter, in 2008, the Supreme Court decided Sprint/United Mgmt. Co. v. Mendelsohn, 552 U.S. 379, 384. There, in referring to the propriety of affording a District Court’s 3 II. The Evidence Was Admissible Under Rule 403 While district courts should conduct the Rule 403 balancing test on the record, this is one of the narrow situations where we have held that we may be permitted to conduct the Rule 403 weighing analysis ourselves.15 In undertaking that balance, I conclude that, when the risk of unfair prejudice is balanced against the probative value of these videos under Rule 403, the balance tips toward allowing the jurors to see these video clips.16 The charges against these defendants included conspiring to receive or distribute child pornography. Thus, the government had to prove the people in the Zoom chat had entered into such an agreement. Merely establishing that several individuals independently watched the same livestream evidentiary rulings deference, the Supreme Court said: “This is particularly true with respect to Rule 403 since it requires an on-the-spot balancing of probative value and prejudice, potentially to exclude as unduly prejudicial some evidence that already has been found to be factually relevant.’’ (internal quotation marks omitted, emphasis added). We cited that language in Finley. See 726 F.3d at 491. However, there, the issue of our conducting the 403 balance did not arise because we concluded that the District Court did conduct the required balancing and we agreed that the evidence was properly admitted. Nevertheless, given my colleagues’ conclusion that the District Court did satisfy (though barely) the balancing requirement, and since I would find any failure to do so harmless error, it is neither necessary nor appropriate to attempt to resolve any tension between our jurisprudence and Supreme Court precedent here. I mention it only because it is relevant to the structure of my discussion. 15 Eufrasio 935 F.2d at 572. I emphasize my colleagues’ admonition that: “[w]e prefer that the district court show its work. That is the wiser and sounder course.” Maj. Op. at 19. Indeed, it is. For reasons that should be apparent to everyone, the atmosphere in a courtroom must be factored into the Rule 403 balance. There is obviously no way for an appellate court to have a feeling for the level of tension or emotion during a trial. 16 Maj. Op. at 25. 4 would not necessarily imply an agreement to receive or distribute child pornography. Such an agreement is, of course, at the heart of a conspiracy charge. “[T]he essence of a conspiracy is a unity of purpose or common design.”17 The video clips are highly relevant to proving the existence of such unity of purpose and consequently, to showing implicit agreement to receive and/or distribute child pornography. The video clips establish the culture of the Zoom chat and thus support an inference that there was an implicit agreement and unity of purpose amongst the people who watched the livestream that it would be distributed to them and received by them. Indeed, the evidence here is so horrendous that it arguably undermines any suggestion to the contrary. Staples claimed he only sought to engage with adult men in the chatroom.18 The videos belie that assertion. They were critical to establishing that both defendants encouraged the distribution and receipt of the livestreams. Although Staples did not specifically ask for more, Heatherly specifically did ask for “vids” so he could “bust before work.”19 Given the nature of what they were watching, the defendants’ continued participation in the Zoom chat after seeing what was occurring there is as much a request for more of the same content as a verbal request would have been. 17 United States v. Kates, 508 F.2d 308, 310-11 (3d Cir.1975) (internal quotation marks omitted) (“It is well established that the gist of a conspiracy is an agreement. However slight or circumstantial the evidence may be, it must, in order to be sufficient to warrant affirmance, tend to prove that the appellant entered into some form of agreement, formal or informal, with his alleged co-conspirators. Similarly, we have stated that the essence of a conspiracy is a unity of purpose or common design.”). 18 Staples Br. at 12. (“Mr. Staples admitted that he visited the Zoom chat room under the alias “Bill Simpson” but contended that his purpose was interacting with adult men, not viewing child pornography. (Appx 286 (Transcr. p. 403- 405))”). 19 Heatherly Br. at 8; Appx 1069 (Heatherly stating “so appreciate if someone showed vids-need to bust before work;” and, approximately three-minutes later, “so close”). 5 In United States v. Bailey, we explained that the probative value of such evidence partially depends on the availability of an alternative, less prejudicial means of establishing the same facts.20 Here, even though a detective testified and described the contents of all of the videos in great detail, it is fair to assume that there is no way that the experience of any of the jurors would have enabled them to fully grasp the horrors of the livestream the defendants were watching or the significance of them remaining in the chat. The adage: “seeing is believing” encapsulates the need for these jurors to actually see the recordings from that Zoom chat. The culture of that chat room simply defies the ability of the most skilled witness to articulate the horror of what the defendants were watching. This culture is highly relevant to determining if there was an agreement amongst those who knowingly participated in the Zoom chats. It is the very fact that the videos cause such revulsion that suggests that one could not willingly view them without tacitly agreeing to the receipt of the livestream.21 This does not, of course, mean that there is a different threshold under Rule 403 for all child pornography cases or an implied exception to the required Rule 403 balance in cases involving child pornography or acts of extreme depravity. Nor does it mean that anyone who knowingly views such conduct necessarily conspires in its production, distribution or receipt. Rather, Rule 403 requires a very delicate, and exceedingly difficult balance based upon the unique circumstances of a given case before the probative value will outweigh its very real and substantial potential to prejudice. That balance is better left to the trial judge who can best assess the atmosphere in the courtroom and the likely impact of such graphic videos. It is an incredibly difficult inquiry that will depend on the unique circumstances of a given trial. Indeed, my colleagues thoughtfully explain why this case is distinguishable from other child pornography cases where we concluded that the potential for harm was simply too grave to admit certain graphic evidence despite its highly probative value.22 20 United States v. Bailey, 840 F.3d 99, 122 (3d Cir. 2016). 21 Heatherly Br. at 22. 22 See Maj. Op. 22-23 6 Because the kind of graphic evidence that was admissible under Rule 403 here is so highly prejudicial, it is important that nothing I have said be misconstrued. I reiterate, there is no child pornography exception to the very difficult balance that Rule 403 requires. Rather, regardless of what is charged or the extent to which disputed evidence tends to offend, each case must be evaluated on its unique facts, the evidence produced, and the trial atmosphere. As I have explained, the other important consideration here is that the defendants were charged with (and convicted of) conspiracy to receive or distribute child pornography in addition to receipt or distribution of child pornography.23 For the reasons I have noted, the video clips helped to establish the culture that permeated the Zoom chats. That was an important part of proving that the participants were involved in such a unity of purpose and common undertaking that they had necessarily entered into an agreement that this type of material be received or distributed. Nevertheless, merely including a conspiracy charge does not conclusively determine the outcome of the 403 balance nor necessarily increase the chances that the probative value of such evidence will outweigh its potential to inflame. The balance in such cases, as illustrated here, is particularly precarious because it is the extreme nature of the charged criminal conduct that yields both its probative value and its potential to unfairly prejudice a defendant. This is not like the video of a violent shooting that we held was improperly admitted in violation of Rule 403 in Bailey. A regrettable consequence of the violence that has become so very pervasive and accessible in today’s mass media is that it is well within the ability of lay jurors to imagine someone being violently gunned down by a shot to the head. Thus, gratuitous depictions of such violence do not often overcome their prejudicial impact and they are rarely necessary to convey the essence of an event to jurors. What we have here is different. The government’s attempt to verbalize what the defendants were watching may well have been inadequate to communicate the nature of the 23 Appx 5, 6, 9, 10. 7 Zoom chats or whether the unity of purpose between these defendants was such that it suggested an implicit agreement to participate in these livestreams, as opposed to “merely” separately observing them. Nevertheless, absent the fact that the jury still did not convict on all charges, I doubt that the centrality of the culture of the Zoom chat would have been sufficient to override the prejudice of subjecting jurors to these videos under Rule 403. Thus, I am not concerned that the government may begin reflexively including a conspiracy charge whenever two or more defendants are involved in viewing child pornography in an attempt to put more weight on the probative value side of the 403 scale. That would risk all defendants being convicted of all charges. In such a case, it would be exceptionally difficult to thereafter establish that the probative value of the evidence outweighed its extreme potential to inflame and unduly prejudice jurors. The record would contain little if anything to show that the jury’s capacity to reason was not overwhelmed or that the court did not abuse its discretion in letting jurors view such evidence. Moreover, when the Rule 403 balance is struck, the trial court has no way of knowing what the jury’s verdict will be. In conclusion, as is evident from all that my colleagues and I have said, these cases present extraordinarily difficult challenges to trial courts under Rule 403 because it will always be extremely difficult for any juror to refrain from being overwhelmed by emotion or animus when confronted with the kind of shocking evidence that these jurors were permitted to view. Nevertheless, this record establishes that these jurors maintained their ability to reason despite whatever emotions they may have harbored. I therefore agree that defendants have not established that the court abused its discretion in admitting these video clips over their Rule 403 objection. 8