NGUYEN, Circuit Judge

Gail Ivens (argued), Gail Ivens Attorney at Law, King City, California; Tony F. Farmani (argued), Rancho Santa Fe, California; and David J. Kaloyanides (argued), Chino, California; for Defendants-Appellants.

Victoria A. Degtyareva (argued), Carol A. Chen, Kathy Yu, Lindsay M. Bailey, and Chelsea Norell, Assistant United States Attorneys; Bram M. Alden, Assistant United States Attorney, Acting Chief, Criminal Appeals Section; Tracy L. Wilkison, Acting United States Attorney; United States Attorney's Office, Los Angeles, California; for Plaintiff-Appellee.

Before: Marsha S. Berzon, Carlos T. Bea, and Jacqueline H. Nguyen, Circuit Judges.

Opinion by Judge Nguyen ;

Partial Concurrence and Partial Dissent by Judge Berzon

NGUYEN, Circuit Judge:

Enrique Holguin, Emanuel Higuera, and Donald Goulet appeal their convictions on charges arising from criminal activity on behalf of Canta Ranas, a multi-generational street gang based in Whittier, California, with ties to the Mexican Mafia criminal organization. Appellants were charged in a sweeping indictment along with dozens of other individuals associated with Canta Ranas.

Following a jury trial, appellants were convicted of conspiracy in violation of the Racketeer Influenced and Corrupt Organizations Act ("RICO"), 18 U.S.C. § 1962(d). The jury also found Goulet and Higuera guilty of drug trafficking conspiracy, 21 U.S.C. § 846, but acquitted Holguin of the same charge. Additionally, Holguin was convicted of assault under the Violent Crimes in Aid of Racketeering Activity ("VICAR") statute, 18 U.S.C. § 1959(a)(6), Goulet was convicted of money laundering conspiracy, 18 U.S.C. § 1956(h), and Higuera was convicted of possession with intent to distribute at least five grams of methamphetamine, 21 U.S.C. § 841(a)(1), (b)(1)(B)(viii). On appeal, appellants raise numerous challenges to the government's use of three expert witnesses, who collectively played a central role in the government's case. Appellants challenge the district court's admission of expert testimony and its handling of one expert's dual-role testimony. We have jurisdiction under 28U.S.C. § 1291, and we affirm.

Because the district court enjoys "broad latitude" with regard to "how to determine reliability," Kumho Tire Co. v. Carmichael , 526 U.S. 137, 142, 119 S.Ct. 1167, 143 L.Ed.2d 238 (1999) (emphasis removed), we cannot say that its failure to hold a hearing in this case was an abuse of discretion. Yet it would have been prudent to hold such a hearing, or employ other procedures such as focused voir dire, because district courts must make explicit findings that the government's expert testimony was reliable. See, e.g., United States v. Valencia-Lopez , 971 F.3d 891, 899 (9th Cir. 2020) ; United States v. Ruvalcaba-Garcia , 923 F.3d 1183, 1190 (9th Cir. 2019). Here, the district court permitted the government's experts to testify without making any findings; indeed, prior to trial, the record was not sufficient to support a reliability finding. We therefore hold that the district court abused its discretion by failing to make any findings that the experts' testimony was reliable. The lack of such findings, however, does not warrant a reversal of appellants' convictions. Because the trial record shows that the government's expert witnesses had sufficient relevant experience and gave adequate explanations for their interpretations of letters and phone calls, the district court's error was harmless.

We have previously addressed the government's common use of law enforcement professionals as dual-role witnesses, and we explained the risks of such testimony. See United States v. Torralba-Mendia , 784 F.3d 652, 658 (9th Cir. 2015) (collecting cases). Dual-role testimony can give an officer "unmerited credibility," blunt cross-examination, confuse jurors, and encourage testimony without proper foundation. Id. While the district court implemented several safeguards to mitigate those risks, they were amplified by confusing instructions and admission of lay opinion resting on limited foundation. However, these issues largely were not objected to at trial, and they do not amount to plain error.

I. Background

Prior to trial, appellants filed several motions in limine concerning the government's proposed expert testimony and requested a Daubert hearing. The district court denied the request for a Daubert hearing, tentatively denied the motions in limine, and allowed the expert testimony without explanation. At the pre-trial conference, when ruling on motions in limine, the district court did not mention reliability and did not clearly address these aspects of the motions.

At trial, the government called three expert witnesses. Officer Robert Rodriguez, lead case agent for the Whittier Police Department, provided background information about Canta Ranas and explained how the gang operates. Officer Rodriguez testified that Canta Ranas had about 140 members active in the Santa Fe Springs and Whittier areas. He explained that the gang generates revenue through criminal activity – primarily drug trafficking. Younger gang members sell drugs "fronted" to them by gang leaders. Officer Rodriguez also explained that Canta Ranas collects taxes both from their own membership and from the surrounding community. Revenue generally goes to Canta Ranas's leader, an incarcerated Mexican Mafia member named David Gavaldon.

Officer Rodriguez then transitioned into lay testimony, describing investigative activities he conducted, explaining the meaning of gang language and monikers, and characterizing the roles that individuals played in the organization. During this phase, Officer Rodriguez identified David Gaitan as a senior gang member who distributed narcotics to foot soldiers.

The government's second expert witness was Rene Enriquez, a former member of the Mexican Mafia who has analyzed Mexican Mafia communications for law enforcement. Enriquez testified about Mexican Mafia practices. Enriquez explained that Mexican Mafia members such as Gavaldon are generally incarcerated, but they exert power in prison through drug trafficking and extortion schemes and outside prison through their ties to local street gangs. Enriquez also interpreted correspondence that Holguin sent to Gavaldon.

The government's third expert witness was DEA Special Agent Steven Paris, who was not involved in this investigation. Based on his experience working on thousands of drug investigations, Agent Paris interpreted intercepted calls involving Goulet, Higuera, and Gaitan, and testified about indicia that seized drugs were for distribution rather than personal use.

II. Expert Testimony

"We review the district court's decision to admit expert testimony for an abuse of discretion. This includes not only the court's ultimate admissibility determination under Daubert and Rule 702, but also its decisions regarding the type of proceedings required to conduct the gatekeeping inquiry in a particular case." United States v. Calderon-Segura , 512 F.3d 1104, 1109 (9th Cir. 2008) (citations omitted).

A. Failure to Hold a Daubert Hearing

Appellants first argue that the district court erred in denying their request for a Daubert hearing.

Rule 702 gives district courts "broad latitude" to structure proceedings concerning expert testimony. Kumho Tire , 526 U.S. at 142, 119 S.Ct. 1167. The Rule 702 inquiry is "flexible" and "must be tied to the facts of a particular case." United States v. Alatorre , 222 F.3d 1098, 1102 (9th Cir. 2000) (quoting Kumho Tire , 526 U.S. at 150, 119 S.Ct. 1167 ). The district court's gatekeeping can be performed through numerous procedures – such as motion in limine briefing and oral argument, voir dire, and cross-examination at trial. See, e.g., Calderon-Segura , 512 F.3d at 1108-10 ; United States v. Hankey , 203 F.3d 1160, 1169 (9th Cir. 2000).

We have consistently held that Daubert hearings are "not required." Est. of Barabin v. AstenJohnson, Inc. , 740 F.3d 457, 463 (9th Cir. 2014) (en banc), overruled on other grounds by United States v. Bacon , 979 F.3d 766 (9th Cir. 2020) (en banc); Alatorre , 222 F.3d at 1100 ("[W]e conclude that trial courts are not compelled to conduct pretrial hearings in order to discharge the gatekeeping function."). Accordingly, it was within the district court's "broad latitude" to deny a Daubert hearing in this case. Kumho Tire , 526 U.S. at 142, 119 S.Ct. 1167. We caution, however, that even if not required, it will often be beneficial for district courts to conduct some proceeding, focused on the reliability of expert testimony, such as a Daubert hearing or voir dire of proffered expert testimony. See Valencia-Lopez , 971 F.3d at 899 n.5 ("[V]oir dire is a recommended method for the district court to conduct a reliability determination and discharge its gatekeeping obligations."). Without such proceedings, it may be difficult in many cases for the district court to clearly discern an expert's methodology and to evaluate how that methodology connects to the expert's opinions. When an expert's methodology is directly presented and probed, the district court will be well positioned to make the reliability findings that our cases require, as discussed below. Moreover, such proceedings prevent the jury from hearing potentially prejudicial foundation testimony if the expert's opinions are ultimately excluded. See United States v. Hermanek , 289 F.3d 1076, 1095 n.7 (9th Cir. 2002) (suggesting that an "[expert's] potentially prejudicial qualifying testimony ... should have been presented outside the presence of the jury"); Alatorre , 222 F.3d at 1105 ("[A]t least ensuring an opportunity for voir dire outside the presence of the jury may be appropriate in certain cases.").

B. Failure to Make Explicit Reliability Findings

Appellants next argue that the district court abdicated its gatekeeping role by failing to make the explicit reliability findings that our cases require.

While a district court's inquiry is "flexible," Alatorre , 222 F.3d at 1102, "the flexibility afforded to the gatekeeper goes to how to determine reliability, not whether to determine reliability." Valencia-Lopez , 971 F.3d at 898 (emphasis in original). A district court "abdicates its gatekeeping role, and necessarily abuses its discretion, when it makes no reliability findings." Id. ; see also Barabin , 740 F.3d at 464.

Reliability findings must be made "explicit" on the record – an "implicit" finding does not suffice. Ruvalcaba-Garcia , 923 F.3d at 1190 ; see also United States v. Irons , 31 F.4th 702, 716 (9th Cir. 2022) (same). This requirement ensures that district courts engage in the reliability inquiry and create a record of that inquiry to facilitate appellate review.

In some places, appellants frame their argument in terms of the qualifications of Officer Rodriguez and Agent Paris to testify as expert witnesses. That is a distinct issue from the question whether the district court made adequate gatekeeping findings.

An expert may be qualified to give expert testimony by "knowledge, skill, experience, training, or education," Fed. R. Evid. 702, which need only exceed "the common knowledge of the average layman," United States v. Finley , 301 F.3d 1000, 1007 (9th Cir. 2002). This standard is "liberal." 4 Jack B. Weinstein & Margaret A. Berger, Weinstein's Federal Evidence § 702.04 (Mark S. Brodin, ed., 2d ed. 2021). Law enforcement professionals are routinely qualified to offer expert testimony based on their training and experience. See, e.g., United States v. Figueroa-Lopez , 125 F.3d 1241, 1246-47 (9th Cir. 1997) (testimony about training and experience "clearly" established qualifications to offer expert testimony about the modus operandi of drug dealers).

Here, Officer Rodriguez and Agent Paris both testified to their extensive training and experience with gang and drug investigations. Their testimony about gang members and drug dealers involved "specialized knowledge" beyond the common knowledge of the jury. Id. at 1244-46. Even if, as appellants contend, Officer Rodriguez and Agent Paris have knowledge that is ordinary among their law enforcement colleagues, Rule 702 does not demand distinction in a particular field. Many people who have received the same training and experience will have the same specialized knowledge beyond the jury's common knowledge of a topic. Accordingly, we conclude that Officer Rodriguez and Agent Paris have the specialized knowledge to testify as experts.

Critically, reliability is an entirely separate question. A district court must distinguish an expert's qualifications from the reliability of the expert's principles and methods. See Daubert v. Merrell Dow Pharms., Inc. , 43 F.3d 1311, 1315-16 (9th Cir. 1995) ("[S]omething doesn't become ‘scientific knowledge’ just because it's uttered by a scientist."). It is "an abuse of discretion to confuse Daubert 's reliability and qualification requirements." Weinstein & Burger, supra , § 702.04. "While ‘there is inevitably some overlap ... they remain distinct concepts and the courts must take care not to conflate them.’ " Id. (quoting Moore v. Intuitive Surgical, Inc. , 995 F.3d 839, 851 (11th Cir. 2021) ).

To carry out its gatekeeping role, a district court must find that an expert's testimony is reliable – an inquiry that focuses not on "what the experts say," or their qualifications, "but what basis they have for saying it." Daubert , 43 F.3d at 1316. A district court cannot be silent about reliability when challenged. Cf. Valencia-Lopez , 971 F.3d at 899 (district court abdicated gatekeeping role where "[n]othing was said about reliability"); Barabin , 740 F.3d at 463 ("Absent from the [district court's] explanation is any indication that the district court assessed, or made findings regarding, the scientific validity or methodology of Mr. Cohen's proposed testimony."). And the district court must make more than a "conclusory statement." Roach , 582 F.3d at 1207. But the inquiry need not be exhaustive. See United States v. Johnson , 916 F.3d 579, 587-88 (7th Cir. 2019) (district court fulfilled gatekeeping obligation by considering gang expert's "significant qualifications and experience" and explaining how they shed light on topics of testimony).

For some experts, "the relevant reliability concerns may focus upon personal knowledge or experience." Kumho Tire , 526 U.S. at 150, 119 S.Ct. 1167. In such cases, the inquiry may cover whether the expert's experience supports the expert's conclusions, see, e.g., Valencia-Lopez , 971 F.3d at 900-01 ; whether the expert's reasoning is circular, speculative, or otherwise flawed, see, e.g., United States v. Vera , 770 F.3d 1232, 1247-48 (9th Cir. 2014) ; or whether the expert's reasoning is adequately explained, see, e.g., Hermanek , 289 F.3d at 1094-95.

The government argues that reliability findings are not required, but our cases clearly require explicit findings. See Valencia-Lopez , 971 F.3d at 899 (district court abused its discretion by admitting expert testimony "without explicitly finding his proposed testimony reliable"); Ruvalcaba-Garcia , 923 F.3d at 1190 ("To satisfy its gatekeeping duty under Daubert , the court must make an explicit reliability finding." (cleaned up)). The government invokes a statement from Kumho Tire that reliability may sometimes be "taken for granted," but the relevant portion of Kumho Tire addresses when "reliability proceedings " – not findings – are "unnecessary." 526 U.S. at 152, 119 S.Ct. 1167 (emphasis added).

The government next argues that the district court "diligently honored" its gatekeeping obligation, offering a string citation in support that spans three pages. In fact, the district court addressed appellants' motions in limine without addressing the reliability issues that they raised. The district court did say that one argument "would go towards weight, not admissibility," but that is not a reliability finding. See Valencia-Lopez , 971 F.3d at 899 ("Dismissing an argument as going to the weight, not admissibility, of the expert's testimony is not a reliability finding.") (cleaned up). Beyond motions in limine, the government cites minor skirmishes over testimony. At most, the government suggests an implicit reliability finding. But "an implicit finding of reliability is not sufficient." Ruvalcaba-Garcia , 923 F.3d at 1190 (citation and quotation marks omitted).

Because appellants challenged the reliability of the government's expert testimony, and the district court did not make explicit reliability findings, the district court abused its discretion. See Valencia-Lopez , 971 F.3d at 899.

C. Reliability of Expert Testimony

We next determine whether the district court's error was harmless. The government can establish harmlessness in this context if the record supports the reliability of its expert testimony. See Ruvalcaba-Garcia , 923 F.3d at 1190.

The government's experts generally testified based on their experience rather than based on any systematic methodology. That by itself does not undermine the reliability of their testimony. Rule 702 "works well for this type of data gathered from years of experience and special knowledge." Hankey , 203 F.3d at 1169. The Rules Advisory Committee has explicitly recognized that "the application of extensive experience" is a "method" that can reliably support expert testimony. Fed. R. Evid. 702 advisory committee's note to 2000 amendment. Experience alone is a reliable basis for the expert testimony regarding gang structure and activities as well as the meaning of familiar expressions. We evaluate that testimony first, and then testimony interpreting gang communication, which requires a more robust foundation.

i. Gang Structure and Activities/Familiar Expressions

A law enforcement expert can reliably testify about the structure and activities of criminal organizations based solely on experience. See, e.g., United States v. Rodriguez , 971 F.3d 1005, 1018 (9th Cir. 2020) (officers' training and experience reliably supported testimony about gang "structure and operation"); Hankey , 203 F.3d at 1169 (officer could testify about gang "tenets," including "code of silence," based on "street intelligence" from investigations). Similarly, "[o]fficers may testify about their interpretations of ‘commonly used drug [or gang] jargon’ based solely on their training and experience." Vera , 770 F.3d at 1241 (citations omitted); see, e.g., Rodriguez , 971 F.3d at 1018 (officers' training and experience reliably supported testimony about "the meanings of terms with fixed meanings").

Officer Rodriguez's testimony about Canta Ranas was amply supported by and within the scope of his extensive investigative experience. In his nearly ten years with the Whittier Police Department, Officer Rodriguez investigated hundreds of gang crimes, many involving Canta Ranas, and he received information about the gang from contacts with members and associates, confidential informants, crime victims, and other investigators. Officer Rodriguez could reliably use his own experience as an investigator, including what he learned from these sources, to form conclusions about how Canta Ranas operates. The same experience allowed Officer Rodriguez to testify reliably about gang terminology familiar from investigations.

Appellants argue that Officer Rodriguez's expert testimony relied on hearsay and, in doing so, violated the Confrontation Clause. However, under Rule 703, an expert may rely on hearsay "[i]f experts in the particular field would reasonably rely on those kinds of facts or data in forming an opinion on the subject." Fed. R. Evid. 703. Appellants do not attempt to argue that a gang investigator would not rely on the kind of information Officer Rodriguez described. Where, as here, a gang expert "applied his training and experience to the sources before him and reached an independent judgment," without "directly repeating what someone else told him," his testimony about gang operations does not offend the Confrontation Clause. See Vera , 770 F.3d at 1239-40 (cleaned up).

Rene Enriquez's testimony about the Mexican Mafia was likewise supported by his seventeen years of personal participation in Mexican Mafia affairs and his seventeen years of assistance to law enforcement in hundreds of investigations. For example, as a former member, Enriquez communicated with his crew, and he reviewed many gang letters for law enforcement. He could therefore reliably testify based on his experience with such communications that only those who are "trusted" can communicate directly with Mexican Mafia members.

Given the continuing flow of information he reviewed since his defection, Enriquez's knowledge was not so stale as to render his expert testimony excludable as unreliable. Although he defected in 2002, since then, Enriquez had interpreted thousands of calls and writings involving Mexican Mafia members in connection with hundreds of investigations. Enriquez is also housed in protective custody, where recent gang dropouts are transferred, and they provide Enriquez with fresh information about Mexican Mafia affairs "on a regular basis." Defense counsel was able to highlight on cross-examination that Enriquez had not been an active Mexican Mafia member for a considerable time, that the Mexican Mafia had changed its operations since Enriquez defected, and that Enriquez was relying on limited information to stay current about those changes. There was a valid connection here between Enriquez's testimony and his experience, but it was for the jury to determine under these circumstances how seriously the passage of time since Enriquez's defection undermined the strength of that connection.

Finally, Agent Paris's opinions about drug traffickers and their familiar expressions were reliably supported by his experience. Agent Paris had twenty-seven years of experience conducting thousands of drug investigations, many involving Hispanic street gangs. His investigative work included thousands of searches, interviews, and controlled buys. That work gave him applicable experience about how drug traffickers keep their belongings. On that basis he could reliably identify indicia that seized drugs were for distribution as opposed to personal use. Similarly, Agent Paris reviewed recorded calls as part of his investigative practice, and he could therefore reliably testify to the meaning of expressions familiar from that experience.

Appellants suggest that all three experts gave disguised percipient testimony based on personal observations rather than objective principles. However, "[a]n expert may base an opinion on facts or data in the case that the expert has ... personally observed." Fed. R. Evid. 703. Moreover, a gang expert offers proper expert testimony by "distill[ing] and synthesiz[ing]" what is personally observed, making it "an original product." Vera , 770 F.3d at 1239 (citation omitted). Even if some of this testimony could have been formulated as lay opinion, "the line between lay and expert opinion depends on the basis of the opinion, not its subject matter," United States v. Barragan , 871 F.3d 689, 704 (9th Cir. 2017), and the experts here testified based on specialized knowledge, cf. Fed. R. Evid. 701(c).

Therefore, the extensive and applicable experience of the government's experts reliably supported their testimony about gang and drug trafficking structure and activities and the meaning of expressions familiar from those contexts. Of course, our conclusion is not based on any categorical determination. The gatekeeping inquiry is always case-specific. See Kumho Tire , 526 U.S. at 150, 119 S.Ct. 1167.

ii. Interpreting Gang Communications

While experience alone supported the experts' testimony in the two categories discussed above, Enriquez and Agent Paris's interpretations of gang communications required a more robust foundation.

We have distinguished testimony about familiar expressions with fixed meanings from testimony interpreting unfamiliar expressions. See Rodriguez , 971 F.3d at 1018-19 (recognizing this distinction). "To interpret the meaning of coded language encountered for the first time in the specific investigation at issue, ... an officer's qualifications, including his experience with narcotics investigations and intercepted communications, are relevant but not alone sufficient...." Vera , 770 F.3d at 1241. Such interpretations require an explanation of the expert's method. See Hermanek , 289 F.3d at 1093-94 (interpreting unfamiliar expressions ... "requir[es] the government to explain [the expert's] method").

That does not, however, mean that an expert must articulate a systematic methodology to interpret unfamiliar expressions. "[R]eliability" in the context of gang or drug experts "depends heavily on the knowledge and experience of the expert, rather than the methodology or theory behind it." Hankey , 203 F.3d at 1169. The Advisory Committee contemplated that expert testimony of this nature – interpreting coded language in drug transactions – could be reliably given on the basis of experience:

[W]hen a law enforcement agent testifies regarding the use of code words in a drug transaction, the principle used by the agent is that participants in such transactions regularly use code words to conceal the nature of their activities. The method used by the agent is the application of extensive experience to analyze the meaning of the conversations. So long as the principles and methods are reliable and applied reliably to the facts of the case, this type of testimony should be admitted. Nothing in this amendment is intended to suggest that experience alone—or experience in conjunction with other knowledge, skill, training or education—may not provide a sufficient foundation for expert testimony. To the contrary, the text of Rule 702 expressly contemplates that an expert may be qualified on the basis of experience. In certain fields, experience is the predominant, if not sole, basis for a great deal of reliable expert testimony.

Fed. R. Evid. 702 advisory committee's note to 2000 amendment (emphasis added).

We have previously held that a law enforcement expert can interpret communications based on the expert's understanding of the surrounding context. For example, in United States v. Reed , we concluded that a law enforcement expert could reliably interpret terms referring to PCP manufacturing based on his experience with PCP investigations. 575 F.3d 900, 923 & n.17 (9th Cir. 2009). We have also allowed a law enforcement expert to interpret disputed words based on training and experience as well as knowledge of related communications where the government "focused" on explaining "the expert's methodology." United States v. Decoud , 456 F.3d 996, 1013-14 (9th Cir. 2006).

Enriquez and Agent Paris offered testimony interpreting communications based on their understanding of the surrounding context. A significant example is Enriquez's interpretation of a 2007 letter that Holguin sent to Gavaldon. The relevant portion of the letter reads:

I was staying in Chino for the last 6 months – it's all messed up in that city, sabes? My roommate wanted to get at you so if it's all good por favor let me know his name is Oso Varrio Trece San Gabriel Valley. Me, him and a few other camaradas were on the east Chino and there was no one there to take care of the ranch, so we as a group got together and looked out for the ranch so it runs smooth. But the camarada who was in "palm" springs wanted [word obscured] to go check in with him in "palm" springs, but I had already moved back to the hood so now my roommate is the only one left behind with the headache, so if you can help him out por favor. The vato wants to know who gave us permission to take care of the ranch. Oso just got at me the other day and says he keeps getting mail from that vato and he wants us all to go see him at "palm" springs. So I'm going to get at Oso and tell him to write me and I'll send it to you.

Enriquez gave the following interpretation at trial:

[Holguin] was in Chino State Prison and it's all messed up there, the mesa is all messed up, the ad hoc commission for Sureños, that he's had a guy there by the name of Oso from Varrio Trece, and him and another couple other guys established a commission for this Mexican Mafia member, that they're in East, Chino East, and they got together to run the prison on behalf of this Mexican Mafia member....

"Mesa" is – it's a Spanish word for "table." It's like an ad hoc commission comprised of soldiers for the organization, Southsiders, Sureños, who will control that prison and generate revenue for the Mexican Mafia member they represent....

[Holguin] is paroled, but that Oso still remains incarcerated in Chino, and a shot caller in Palm Hall, which he calls Palm Springs, wants him to turn himself in to ad seg and come explain why he's doing this and who gave him the authority to do it. So he wants a response so that this guy can be provided with a response denoting authorization to run the prison on [Gavaldon's] behalf.

Although Enriquez did not testify to prior familiarity with the language in the ranch letter, he did testify to his extensive experience with the surrounding context – the Mexican Mafia's operations in California prisons and their use of coded gang communications. Specifically, based on his participation as a member and his recent assistance to law enforcement, Enriquez testified that members generate money in prison through "extortion collection scheme[s]." They set up informal commissions called "mesas" that collect a third of the proceeds of drug dealing and other illegal activities in prison.

Similarly, based on his communications as a former member and his more recent review of communications for law enforcement, Enriquez explained how coded gang communications work. About "90 percent" of the content covers "mundane topics." "[O]ne or two sentences" will contain "illicit information," designed to go "beneath the radar of the staff that are responsible for monitoring communications." Members can identify illicit information because they are "predisposed to understand the underlying content." To identify illicit topics, Enriquez looks for a break in the "normal fluidity of conversation." When interpreting code, Enriquez considers each expression in "the total context of the entire message."

Enriquez detailed how he applied this experience to the ranch letter. Beginning with individual expressions, Enriquez identified "roommate" as a reference to a cellmate, as the letter designates the roommate's gang moniker and gang, "Oso from Varrio Trece." Enriquez identified "Chino" as a reference to Chino State Prison given the references to east, a section of the prison, and to "the ranch," which Enriquez identifies as a mesa.

Enriquez applied the same method to individual expressions in the remaining portion of the letter. "Palm Springs" was a reference to "Palm Hall," a wing of Chino State Prison that was the Mexican Mafia's "home front." That Holguin had "moved back to the hood" but his roommate is "left behind" meant that Holguin had been paroled but Oso was incarcerated. When the letter said that someone from Palm Hall was asking "who gave us permission to take care of the ranch," it made sense to Enriquez that Oso was being asked why and on whose behalf he was running the mesa. Visits to Palm Hall, as the "vato" had requested, happen when an inmate deliberately "turn[s] himself in to ad seg," to be voluntarily sent to Palm Hall.

Connecting these pieces, Enriquez explained that Holguin wanted to inform Gavaldon that he and Oso set up a mesa at Chino State Prison, and the purpose of the message was twofold: first, to get recognition for generating revenue for Gavaldon; and second, to get a letter of authorization from Gavaldon to allow Oso to continue to run the Chino mesa without interference.

Although more intuitive than systematic, Enriquez's experience gave him a background context through which Holguin's coded, cryptic, or ambiguous messages could be reliably understood. See Reed , 575 F.3d at 923 & n.17. Enriquez explained how he applied his experience, discerning the likely meaning of individual expressions in the context of Mexican Mafia operations, and producing a meaning that makes sense as a whole. While his explanation regarding some expressions is more thorough than others, that does not by itself establish that he cannot reliably draw intuitive inferences about the meaning of expressions from relevant experience that he explicitly identifies. Contrary to the partial dissent's contention, Enriquez's interpretation of the ranch as a mesa was not supported merely by his "general qualifications" but by his specific experience with the Mexican Mafia's methods of communication and the affairs they discussed. See Partial Dissent at 871–72. Cf. Hermanek , 289 F.3d at 1094 (agent's interpretations relied only on general qualifications where he "failed to explain in any detail the knowledge, investigatory facts and evidence he was drawing from").

A more organized explanation, cleanly separated from the interpretation itself, would surely aid a district court in making reliability findings. Prosecutors should accordingly endeavor to separate testimony about methodology from substantive opinions. However, Enriquez's explanation here was adequate under the circumstances. See Decoud , 456 F.3d at 1013-14. Enriquez's other interpretations are similar, and reliable under the same reasoning.

Agent Paris's interpretations are similarly reliable. Agent Paris interpreted several intercepted phone calls that Goulet and Higuera had with David Gaitan, the right-hand man to shot caller Jose Loza. Like Enriquez, Agent Paris applied his investigative experience and adequately explained his reasoning.

In one call, Goulet told Gaitan, "I have a hundred and forty right now I still have 5 grams, you know I mean? [T]o make up more than the three hundred, you know what I mean?" Through his experience on thousands of drug investigations, Agent Paris knew that drugs are often "fronted," meaning "given out ... to dealers who sell and then bring the money back at a later time." With that knowledge of fronting, Agent Paris interpreted Goulet's statement that he has "140 right now," "still ha[s] 5 grams ... to make up for more than the 300," to mean that Goulet has 5 grams left to make up the balance of a $300 debt from fronted drugs.

Relying on his knowledge of typical drug prices to interpret a call between Higuera and Gaitan, Agent Paris testified that Higuera's request for "a half ... for two-fifty," refers to "half an ounce" of methamphetamine. See York , 572 F.3d at 429 ("[K]nowledge of common quantities and prices gave [a law enforcement expert] a reliable basis to interpret the otherwise undefined terms ‘six’ and ‘nine’ as $6,000 and nine ounces of cocaine."); Hermanek , 289 F.3d at 1097 ("We find no error in ... [admitting agent's] testimony translating coded numbers into quantities and prices of cocaine.").

Agent Paris also interpreted a statement in a letter Holguin sent to his nephew from jail. Holguin asked to have his "hookup" sent to a woman named Patricia Ramirez and to put "from Boxer CR," Holguin's gang moniker. Agent Paris was asked to assume that the letters related to the context of drug trafficking. With that assumption, Agent Paris interpreted Holguin's statement to mean, "tell her who my supplier was" and convey that she should tell the supplier she was sent by Boxer CR so that the supplier trusts her. This interpretation reliably explained why a drug trafficker would forward his source based on Agent Paris's experience with how drug traffickers operate.

We note that these considerations of reliability do not exhaust every potential challenge under Rule 702. Appellants do not argue that Agent Paris interpreted language that was not properly connected to drug trafficking. See United States v. Cruz , 363 F.3d 187, 196 (2d Cir. 2004) ("An expert witness called on to testify about the meaning of narcotics codes strays from the scope of his expertise when he interprets ambiguous words or phrases and there is no evidence that these terms were drug codes."). Nor do appellants argue that any of the expert testimony was unhelpful because it interpreted clear language. Our task is only to determine whether the district court's failure to make reliability findings was harmless, and on this record we conclude that it was.

III. Officer Rodriguez's Dual-Role Testimony

We have described several dangers with a witness testifying in both lay and expert ("dual-role") capacities. "[A]n agent's status as an expert could lend him unmerited credibility when testifying as a percipient witness, cross-examination might be inhibited, jurors could be confused and the agent might be more likely to stray from reliable methodology and rely on hearsay." Vera , 770 F.3d at 1242.

Despite these dangers, dual-role testimony is not "categorically prohibited." United States v. Freeman , 498 F.3d 893, 904 (9th Cir. 2007). The district court must "engage[ ] in vigilant gatekeeping" to ensure that "jurors are aware of the witness's dual roles." Id. District courts must instruct jurors about how to evaluate a dual-role testimony. See Vera , 770 F.3d at 1243 (omitting instruction is plain error). The dangers of dual-role testimony can also be mitigated by separating testimony into lay and expert phases, requiring specific foundation testimony, and preventing witnesses from engaging in speculation, conveying hearsay, or interpreting clear statements. See Torralba-Mendia , 784 F.3d at 658.

Here, the district court adopted several safeguards to mitigate the dangers of Officer Rodriguez's dual-role testimony. First, the district court instructed the jury on dual-role testimony before Officer Rodriguez testified and at the end of the case. Second, the district court bifurcated Officer Rodriguez's testimony into percipient and expert phases. This separation was reinforced by the government's signaling of each phase when it began, and by the district court's reminder that it previously explained "the difference between expert testimony and percipient witness." See United States v. Anchrum , 590 F.3d 795, 804 (9th Cir. 2009) (risks of dual-role testimony mitigated where testimony was bifurcated and the prosecutor announced he would "shift gears" between phases). Third, the district court required the government to specify whether each question was directed at Officer Rodriguez's training and experience or his participation in the investigation. The government scrupulously complied, beginning its questions with "based on your training and experience" or "based on your participation in the investigation." See United States v. Martinez , 657 F.3d 811, 817 (9th Cir. 2011) (no error in allowing dual-role testimony where "[t]he government was nearly always exact in specifying when it was asking for [case agent's] testimony as an expert").

A. Dual-Role Instructions

Appellants first argue that the district court erred in instructing the jury on dual-role testimony. We review de novo whether a jury instruction misstates the law, but we review the language and formulation of jury instructions for abuse of discretion. See Rodriguez , 971 F.3d at 1012.

The district court twice instructed the jury on dual-role testimony: first, before Officer Rodriguez testified, and second, at the close of the case, with language hewing closely to our model instruction. See Ninth Circuit Manual of Model Criminal Jury Instructions, No. 4.15 (2018). These instructions generally distinguished testimony based on specialized knowledge, experience, and education from testimony based on a witness's own observations. The district court specifically explained that the jury should judge the credibility of expert testimony "like any other testimony" and "give it as much weight as you think it deserves, considering the witness's education and experience, the reasons given for the opinion, and all the evidence in this case." However, the district court confusingly used the blanket term "opinion testimony" to refer only to expert testimony.

Appellants' only objection to the dual-role instructions at trial was their request for an additional instruction specifically addressing undue deference. Because appellants do not identify any legal error in this omission, we review this issue for abuse of discretion.

It is good practice for a district court to give an instruction on undue deference. See Rodriguez , 971 F.3d at 1017-18 (declining to notice plain error in jury instructions which specifically addressed "that the jury should not give undue deference to the testimony of an opinion witness, just because he has been permitted to testify in that capacity"). We have frequently noted the risk that the jury will improperly consider a dual-role witness's training and experience, rather than what the witness personally observed, in evaluating the witness's percipient or lay opinion testimony. See, e.g., Freeman , 498 F.3d at 903 ("This lack of clarity regarding [the agent's] dual roles created a risk that there was an imprimatur of scientific or technical validity to the entirety of his testimony.").

But, on this record, the district court did not abuse its discretion in declining to give the requested instruction. As described above, the district court implemented numerous safeguards to ensure the jury was aware of the distinct bases for Officer Rodriguez's testimony, including bifurcation and a mid-trial instruction and reminder between phases. Most importantly, the government consistently specified whether it was eliciting testimony based on Officer Rodriguez's training and experience or participation in the investigation, so the jury was repeatedly informed that the basis for Officer Rodriguez's percipient testimony was independent from the basis for his expert testimony. See Martinez , 657 F.3d at 817. Moreover, the district court specifically instructed the jury that expert testimony "should be judged like any other testimony," and the percipient testimony of a lay witness should be judged by "taking into account the factors discussed earlier ... to assist you in weighing the credibility of witnesses." Applying these instructions, the jury would have evaluated Officer Rodriguez's lay testimony like that of any other witness – without regard to his separate testimony as an expert. Accordingly, while the defense's requested instruction may have further clarified Officer Rodriguez's distinct roles, its omission was not an abuse of discretion.

Appellants also argue that the district court should have instructed the jury on a few other aspects of the distinction between expert and lay testimony – specifically, that "the ‘facts’ on which Rodriguez based his expert opinions should not be considered for their truth but only to assess the strength of his opinions," and that "his non-expert testimony was ‘not based on scientific, technical, or other specialized [knowledge].’ " Appellants never raised these points at trial, so we review for plain error. Because these points were either not implicated in this case or were not prejudicial in light of the district court's safeguards, reversal is not warranted.

The thrust of appellants' argument, however, appears to be that the instructions failed to adequately explain the difference between Officer Rodriguez's dual roles. The district court's instructions distinguished expert testimony from percipient testimony – but it used the blanket term "opinions" to refer only to expert testimony. So the district court did not distinguish expert opinion from lay opinion. Yet much of Officer Rodriguez's non-expert testimony consisted of lay opinions – inferences drawn from observations during the investigation – rather than testimony conveying personal observations. The jury therefore may have been confused about how to evaluate Officer Rodriguez's lay opinion testimony.

We have previously "emphasize[d] that trial courts should endeavor to explain clearly the differences between lay percipient testimony, lay opinion testimony (as governed by Rule 701 ), and expert opinion testimony (as governed by Rule 702 ) in settings where all three arise." Rodriguez , 971 F.3d at 1018. "In many cases, designating an umbrella category of ‘opinion testimony’ may fail to provide an appropriate level of nuance to guide the jury's evaluation of dual role testimony." Id. Glossing over this three-way distinction may lead to the jury applying the instructions that they were given about "opinion" testimony to lay opinion even though it was intended for expert testimony. In doing so, the jury would consider the witness's experience, training, and specialized knowledge in evaluating lay opinion – exactly the kind of bolstering of lay opinion with expert credentials about which we have warned. See Freeman , 498 F.3d at 903.

However, given the other safeguards that the district court employed, the dual-role instructions were not plainly erroneous.

B. Admission of Lay Opinion Testimony

Appellants also argue that Officer Rodriguez's lay opinion testimony was admitted without proper foundation. In many instances, Officer Rodriguez offered lay opinions based on his "participation in the investigation" as a whole, not on any particular perceptions during that investigation.

However, appellants generally did not object to this limited foundation. Although appellants filed a pre-trial motion addressing dual-role testimony, the motion did not alert the district court to the specific foundational issues that arose at trial. Cf. United States v. Perez , 962 F.3d 420, 435 n.3 (9th Cir. 2020) (noting various objections at trial "raise the essence" of a Rule 701 objection). We therefore review for plain error.

" Rule 701 allows a lay witness to offer opinions that are (a) ‘rationally based on the witness's perception,’ (b) ‘helpful’ to the jury, and (c) ‘not based on scientific, technical or other specialized knowledge within the scope of’ expert testimony." United States v. Gadson , 763 F.3d 1189, 1206 (9th Cir. 2014) (quoting Fed. R. Evid. 701 ). Under Rule 701, a witness "may not ‘testify based on speculation, rely on hearsay, or interpret unambiguous, clear statements.’ " Perez , 962 F.3d at 435 (citation omitted).

Officer Rodriguez's lay opinion testimony described the roles played by individuals in the Canta Ranas organization, the gang monikers used by appellants and others, and the meaning of words and phrases encountered in the investigation. All of this can be the proper subject of lay opinion testimony.

However, lay opinion testimony is admissible only if the government lays a foundation indicating that the opinion was "rationally based on the witness's perception." Fed. R. Evid. 701 ; see also Rodriguez , 971 F.3d at 1019 (testimony was erroneously admitted where the government failed to "establish[ ] the requisite personal knowledge"). We have allowed officers to interpret communications based on the investigation as a whole and have not required all information supporting lay opinion to be placed before the jury. See Gadson , 763 F.3d at 1206-09. But our cases generally feature some testimony about the investigative activities that could have supported a witness's lay opinions. See, e.g., Barragan , 871 F.3d at 703 (affirming admission of lay opinion testimony where "[t]he agents made clear that their interpretations were based on their review of hundreds of calls and text messages during the investigation"); Gadson , 763 F.3d at 1209 (lay opinion testimony was admissible where it was based in part on law enforcement witness's "review of around 100 hours" of phone calls).

Here, Officer Rodriguez testified that he participated in searches, surveillance operations, and arrests as part of the investigation resulting in the charges tried before the jury. In some cases, he mentioned that he learned certain information by reviewing Facebook and prison correspondence by participants in the drug conspiracy, but he never explained the scope of that review. For much of his lay opinion testimony, there was no discernable connection between Officer Rodriguez's investigative activities and his conclusions. Without a specific foundation, there was a serious risk that the jury filled in the gap by looking to the basis that supported Officer Rodriguez's expert testimony – his extensive experience and training as an officer – and presumed that basis would also support his lay opinion testimony. That risk is exactly why district courts should ensure that a dual-role witness lays an adequately specific foundation. See Torralba-Mendia , 784 F.3d at 658 ("[T]he district court should require an adequately specific foundation, so that the jury has the information needed to evaluate the case agent's testimony.").

But any error here was not plain. Officer Rodriguez provided some testimony about his investigative activities. Any disconnect between his general foundational testimony and his specific opinions was not sufficient to warrant sua sponte intervention.

Appellants also argue that Officer Rodriguez's lay opinion testimony was based on hearsay. In the expert phase, Officer Rodriguez's testimony was informed by information he received from Canta Ranas members and associates and other gang investigators. But that does not mean he relied on the same information in his lay opinion testimony. "[T]he line between lay and expert opinion depends on the basis of the opinion, not its subject matter." Barragan , 871 F.3d at 704. Officer Rodriguez consistently specified that his lay opinions were based on his participation in the investigation. Appellants point to nothing in the record indicating that these opinions were instead recycled hearsay. This argument is thus better understood as another variation of appellants' challenge to the thin foundation for Officer Rodriguez's lay opinions. As we explained, this was not plain error.

AFFIRMED.

BERZON, Circuit Judge, concurring in part and dissenting in part:

I concur in the memorandum disposition, except as to paragraphs 8 and 9 and its "affirmed" conclusion as to Holguin. I also concur in the majority opinion's discussion of lay witness opinion testimony by Officer Rodriguez, see Majority Op. at 864–66, and expert testimony by Officer Rodriguez and Agent Paris, see Majority Op. at 856–57; infra p. 870 n.3. With regard to the majority opinion's discussion of expert testimony about drug jargon, I agree with much of the analysis, but, for two reasons, I disagree with the conclusion that the district court's abuse of discretion regarding Rene Enriquez's testimony was harmless as to Enrique Holguin.

First, I would hold that the district court must conduct a Daubert hearing or voir dire to assess the reliability of a police officer, detective, or other law enforcement expert who seeks to testify based on experience alone, rather than on scientific methodology.

Second, I disagree with the majority's assessment of Rene Enriquez's expert testimony as to Enrique Holguin's communications. I agree with the first holding on this issue: the district court abdicated its gatekeeping role by admitting Enriquez's testimony that Holguin sought to establish a "mesa" in Chino State Prison. But I cannot agree with the majority's conclusion that this error was rendered harmless by record evidence showing that Enriquez's testimony was reliable as to that testimony. Despite extensive testimony as to his approach to interpreting coded communications by members of the Mexican Mafia, Enriquez failed to provide any explanation of how he applied his background knowledge and experience to interpret the term "ranch" in Holguin's letter as referring to an extortion scheme pursuant to which, among other things, the Canta Ranas gang would accrue one-third of all contraband smuggled into the prison and would collect "taxes" on other prison activities. The record, moreover, contains no evidence to which the prosecution could have pointed to prove this testimony was reliable. The introduction of Enriquez's testimony regarding the "ranch" letter was not harmless, as the record contains little evidence to support Holguin's Racketeer Influenced and Corrupt Organizations Act ("RICO") conspiracy conviction in the absence of Enriquez's impermissible testimony. I would therefore reverse Holguin's RICO conspiracy conviction.

I.

A.

Courts face unique challenges when assessing the reliability of an expert who testifies about "gang membership and tenets" based only on "street intelligence," gathered from "years of experience." See United States v. Hankey , 203 F.3d 1160, 1169–70 (9th Cir. 2000). In this context, the " Daubert factors (peer review, publication, potential error rate, etc.) simply are not applicable." Id. at 1169. Unlike traditional expertise, "[t]here is no objectively ascertainable or empirically supportable measure of personal experience" with drug jargon and "no objective means of regulating or certifying gang experts." Joëlle Anne Moreno, What Happens When Dirty Harry Becomes an (Expert) Witness for the Prosecution? , 79 Tul. L. Rev. 1, 30 (2004) ; Hon. Jack Nevin, Conviction, Confrontation, and Crawford: Gang Expert Testimony As Testimonial Hearsay , 34 Seattle U. L. Rev. 857, 875 & n.120. (2011). Further, there appears to be no empirical research that "stud[ies] or test[s] the reliability of any drug jargon definitions," even though studies "to ascertain whether drug jargon definitions are accurate and current" could be undertaken. Moreno, 79 Tul. L. Rev. at 34. The absence of any empirical research-based confirmation is particularly troubling here, as Enriquez had not been a member of the Mexican Mafia since 2002. Even if his knowledge of drug jargon once was reliable, it could have become stale.

The history of law enforcement expert testimony illuminates contemporary challenges. Scholars trace the rise of law enforcement expert testimony to the 1950s and 60s. See Anna Lvovsky, The Judicial Presumption of Police Expertise , 130 Harv. L. Rev. 1995, 2018 (2017). Before then, judges allowed law enforcement personnel to testify regarding drug jargon and "vice" crimes only as lay witnesses. Id. In the 1950s, such evidence began to come in as expert testimony. Id. at 2019–2022. The source of these experts' reliability was often decades of experience working as a police officer, and, occasionally, experience attending or teaching at a police training academy. Id. at 2022. The bases of reliability roughly track the contemporary requirements for experience-based expert testimony generally. See Hankey , 203 F.3d at 1169–70.

The increased prevalence of law enforcement expert witnesses in criminal trials at once created and reinforced judicial and popular notions of law enforcement expertise. Law enforcement experts were generally unchallenged; the defense rarely introduced their own expert to counter the officer's testimony. See Lvovsky, supra , at 2062. As a result, judges faced "sustained, often-uncontested evidence of both the depth and apparent commonality of police insight into crime." Id. at 2062–63. Notably, this trust in law enforcement expertise developed even though the traditional justifications for experience-based expert testimony never quite applied.

The general justification for experience-based testimony is loosely as follows:

[I]f a source of information and data is reliable enough for an expert to rely on it in the pursuit of his or her profession,

trade or calling in the ‘real world’—where reliance on untrustworthy information can result in loss of professional standing, livelihood, and even lives—it is reliable enough to be used as a basis for expert testimony.

Clifford S. Fishman and Anne Toomey McKenna, 6 Jones on Evidence § 46:10 (7th ed. 2022). But law enforcement personnel occupy a unique institutional role: "unlike most fields of endeavor, for law enforcement and the various branches of forensics, the ‘real world’ is the courtroom," at least in part. See id. (emphasis omitted). In addition to job duties "on the beat," law enforcement and investigative personnel derive professional standing and livelihood from, and play an institutional role in, the courtroom as well. So, law enforcement expert testimony itself reinforces the legitimacy of the personnel offering it. This phenomenon is at the root of my disagreement with the majority today.

B.

Doctrinally, we have recognized some of these challenges. Precisely because law enforcement experts use no systemic methodology, see Majority Op. at 855–56, "reliability becomes more, not less, important." United States v. Valencia-Lopez , 971 F.3d 891, 898 (9th Cir. 2020). Although the district court has discretion "in deciding how to test an expert's reliability," it generally must make preliminary legal determinations about the expert's qualifications and methodology. See Hankey , 203 F.3d at 1168–69 (quoting Kumho Tire Co., Ltd. v. Carmichael , 526 U.S. 137, 152, 119 S.Ct. 1167, 143 L.Ed.2d 238 (1999) ). Requiring a Daubert hearing or formal voir dire, outside the presence of the jury, for expert testimony based on experience would preserve the district court's discretion regarding "how to test an expert's reliability," id. —which factors to consider and questions to ask—while ensuring there is a reliability assessment. After all, as the majority opinion recognizes, see Majority Op. at 853–55, the court's discretion does not extend to "whether to determine reliability" at all. Valencia-Lopez , 971 F.3d at 898.

This principle makes sense for the reasons the majority opinion outlines. See Majority Op. at 853. In the absence of a Daubert hearing or voir dire, "it may be difficult in many cases for the district court to clearly discern an expert's methodology and to evaluate how that methodology connects to the expert's opinions." Id. The connection between methodology and opinion is key. Even where an expert explains "in detail his knowledge of defendants," such as knowing they are members of a gang or drug distribution scheme, the expert still must "establish how he applied that knowledge to interpret particular words and phrases used in particular conversations." United States v. Hermanek , 289 F.3d 1076, 1094–95 (9th Cir. 2002). Allowing a law enforcement witness to testify as an expert based on experience alone, without connecting that experience through a Daubert hearing or voir dire to the specific matters on which he plans to testify, invites experts to testify even when "the basis for their expert testimony" has "gr[own] thin," and to do so without "offer[ing] an explanation for how they arrived at their interpretations." United States v. Rodriguez , 971 F.3d 1005, 1018–19 (9th Cir. 2020). And objections and cross-examination during such testimony can provide only limited means for testing the basis and reliability of the expert's particular interpretations and explanations. Such inquiries require the defendant, and the court, to respond to the witness's testimony on the fly without prior knowledge of the assumed connection between the expert's background and the specific testimony offered. The likely impact, as here, see infra Part II, is that the investigation of that connection will be incomplete and generic, and the district court will not have an opportunity to make a focused, express reliability determination. Moreover, if the expert's explanations on the stand "fail[ ] to evince indicia of reliability or methodological rigor," it may be too late to avoid undue prejudice to the jury. Rodriguez , 971 F.3d at 1019 ; Hermanek , 289 F.3d at 1095 n.7 ; see also Majority Op. at 853.

In practice, we generally do require a Daubert hearing or voir dire for experience-based testimony about gang or drug activity. Since Kumho Tire , we have, with limited exceptions, affirmed a district court's finding that a gang expert was reliable only when the court held either a Daubert hearing or voir dire, outside the presence of the jury.

United States v. Alatorre , for example, upheld the district court's admission of a Customs Service agent's expert testimony because "voir dire established that" the expert "was qualified to testify" about both the value of the drugs at issue, and the structure of the drug enterprise. 222 F.3d 1098, 1100, 1104 (9th Cir. 2000) ; see also Valencia-Lopez , 971 F.3d at 899 (describing Alatorre 's holding). United States v. Murillo upheld admission of a law enforcement agent's expert testimony because there was "clear evidence in the trial transcript that an adequate voir dire was conducted ...." 255 F.3d 1169, 1178 (9th Cir. 2001), overruled on other grounds by Muehler v. Mena , 544 U.S. 93, 125 S.Ct. 1465, 161 L.Ed.2d 299 (2005). Hankey upheld the admission of a police officer's gang-related testimony because the court had "conducted extensive voir dire to assess" its "basis" and "reliability." 203 F.3d at 1168. And United States v. Decoud upheld the admission of the government's drug expert because, before the expert testified, "the district court held a Daubert hearing at which the expert explained the methodology that he used to interpret each of the handful of disputed words." 456 F.3d 996, 1013 (9th Cir. 2006).

We have, on occasion, suggested that even without a Daubert hearing or voir dire, some other reliability determination might suffice. But a truly adequate alternative rarely materializes. See Valencia-Lopez , 971 F.3d at 899 & n.5. Often, the district court's failure to hold a Daubert hearing or voir dire coincides with "an implicit finding of reliability," or no reliability findings at all—both of which are insufficient, as the majority opinion recognizes. United States v. Ruvalcaba-Garcia , 923 F.3d 1183, 1190 (9th Cir. 2019) (quoting United States v. Jawara , 474 F.3d 565, 583 (9th Cir. 2007) ); see, e.g., Valencia-Lopez , 971 F.3d at 899 ; Rodriguez , 971 F.3d at 1018–19 ; United States v. Vera , 770 F.3d 1232, 1247 (9th Cir. 2014) ; Majority Op. at 855.

We recently declined to decide whether to formalize the requirement for a Daubert hearing or voir dire in gang-expert cases. Valencia-Lopez left for another day "whether a district court fulfills its gatekeeping role without either allowing voir dire or conducting a Daubert hearing." 971 F.3d at 899 n.5 (citing Estate of Barabin v. AstenJohnson, Inc. , 740 F.3d 457, 463 (9th Cir. 2014) (en banc), overruled on other grounds by United States v. Bacon , 979 F.3d 766 (9th Cir. 2020) (en banc)). I would reach the question today and require what, practically speaking, we generally require for a district court to adequately carry out its gatekeeping role with respect to experts basing opinions about criminal enterprises on experience alone. I would require a formal Daubert hearing or voir dire, outside the presence of the jury and focused on the particulars of the planned testimony, so that its basis and reliability can be carefully analyzed.

II.

The expert testimony in this case, Enriquez's in particular, illustrates the need for such pre-testimonial evaluation. I agree with the majority opinion's conclusion that "the district court abused its discretion by failing to make any findings that the experts' testimony was reliable." Majority Op. at 851. I depart from its harmlessness analysis as to Holguin. "The government bears the burden to show harmlessness," which it can meet "by showing either that ‘it is more probable than not that the jury would have reached the same verdict even if the [expert testimony] had not been admitted,’ or that the admitted ‘expert testimony [was] relevant and reliable’ under Daubert based on ‘the record established by the district court.’ " Ruvalcaba-Garcia , 923 F.3d at 1190 (internal citation omitted) (quoting Barabin , 740 F.3d at 465, 467 ). The majority opinion relies on the second ground; in its view, the record demonstrated Enriquez's reliability, so the district court's error was harmless. I disagree.

A.

The district court's failure to inquire specifically into the basis for Enriquez's testimony before he took the stand resulted, in my view, in the impermissible introduction of testimony that does not appear connected to any indicia of reliability based on identifiable experience. I would hold that Enriquez's testimony regarding the meaning of certain terms—terms he first encountered in this case and terms without fixed meaning—was not shown to be reliable, and there is no basis in the record to support Enriquez's reliability in testifying about those terms. A law-enforcement expert's "qualifications, including [the person's] experience with narcotics investigations and intercepted communications, are relevant but not alone sufficient" to establish a proper basis for testimony about "coded language encountered for the first time in the specific investigation at issue." Vera , 770 F.3d at 1241. For this type of testimony, "the proffered expert must establish that reliable principles and methods underlie the particular conclusions offered. " Hermanek , 289 F.3d at 1094 (emphasis added). Our cases show some ways experts can properly lay this type of foundation:

• Reed , 575 F.3d at 923 "approv[ed] expert testimony interpreting terms the agent ‘knew to refer to the reagent used in the PCP manufacturing process’ ";

• Decoud , 456 F.3d at 1013–14 & n. 6 "approv[ed] the agent's explanation that he interpreted ‘diznerty’ as slang for ‘dirty’ based on his familiarity with a common speaking style that creates slang versions of specific words by adding ‘e’ or ‘ez.’ "

Vera , 770 F.3d at 1242 (citing Reed , 575 F.3d at 923, and Decoud , 456 F.3d at 1013–14 & n.6 ).

The majority recites the wrong standard when considering testimony about newly encountered terms. After properly citing the requirement from Vera that testimony about new terms must have a more robust foundation, the majority relies on Hankey for the proposition that reliability "in the context of gang or drug experts ‘depends heavily on the knowledge and experience of the expert, rather than the methodology or theory behind it.’ " Majority Op. at 858 (quoting Hankey , 203 F.3d at 1169 ). But that quoted language from Hankey concerned expert testimony about "gang memberships and tenets," precisely the type of testimony for which general police experience is sufficient. Hankey , 203 F.3d at 1169–70 (considering expert testimony about gang names, the defendant's gang membership, and the gang's "code of silence"). Contrary to the majority's assertion, for testimony about new or uncommon terms, the methodology and theory behind each specific conclusion offered is more important than background knowledge and experience. Thus, Hermanek explained, an expert's "knowledge and prior investigation of defendants" is "too vague and generalized" to suffice as showing reliability regarding the interpretation of new or uncommon terms used by alleged conspirators. 289 F.3d at 1094. Instead, "[u]nder Rule 702, the proffered expert must establish that reliable principles and methods underlie the particular conclusions offered—here, the interpretation of particular words as referring to cocaine." Id.

Enriquez's testimony did evince general knowledge that members of the Mexican Mafia communicated with each other in coded language. But his testimony did not establish any such "reliable principles and methods" as applied to his specific testimony regarding Holguin's "ranch" letter. Id. As the majority opinion recounts, Enriquez testified that the term "the ranch" in Holguin's letter referred to a "mesa," an "ad hoc commission comprised of soldiers for the" gang, which will "control that prison and generate revenue for the Mexican Mafia member they represent." Asked what that opinion was based on, Enriquez simply quoted a sentence from Holguin's letter and then said, "That in itself right there is saying that we formed a mesa because it wasn't running right, so we got together and formed this commission for you. We're creating a mesa for you." Enriquez's response provides no basis at all; it is simply a restatement of his conclusion.

Enriquez went on to testify about how a "mesa" works in general. Enriquez utterly failed to explain how he discerned those gang-specific meanings from the particular words in the letter. And his testimony was not general, but quite specific, stating even the percentage of profits that a "mesa" receives from gang activity He drew no connection whatsoever between the term "the ranch" and a "mesa," let alone one whose structure requires anyone who brings contraband into prison "to pay a third" to "the house, which is the mesa." In short, Enriquez's testimony was improper under Rule 702. See Hermanek , 289 F.3d at 1094.

Concluding otherwise, the majority "appear[s] to misapprehend the parameters of expert testimony in the gang expert context, assuming that" Enriquez's "general qualifications sufficed to support the full range of opinion testimony" he gave. Rodriguez , 971 F.3d at 1018.When testifying about other terms, Enriquez did adequately explain his methods; the contrast highlights the absence of a sufficient basis for his "mesa" testimony.

For example, Enriquez explained why he believed "Palm Springs" referred to the prison building called "Palm Hall": "In each—each housing unit or cell block is named, like Mariposa Hall or Birch Hall and Palm Hall. They're all named after trees. Palm Hall was headquarters for the Mexican Mafia. This was home front for the organization, and it's usually referred to as ‘Palmas,’ Palm Springs." Similarly, when explaining the meaning of images of a frog and a spider in Holguin's letter, Enriquez stated: "The spider is David Gavaldon's nickname, and he's from Canta Ranas, Singing Frogs. So it's really a—it's just—in a symbolic way, theyre saying Spider from Canta Ranas." Again, this explanation provided an appropriate "link" between his knowledge and testimony; it shows "how he applied" his background "knowledge to interpret particular words and phrases used in particular conversations." Hermanek , 289 F.3d at 1094–95 (emphasis added).

Enriquez failed to offer any similar link when testifying about Holguin's purported reference to a "mesa"—never mentioned in the letter—and when extrapolating from that reference to a scheme to generate revenue for the Mexican Mafia, and from that scheme to the distribution of a specific percentage of drug revenues to the Mexican Mafia. The basis for the asserted connection between Holguin's letter and the very particular drug-trafficking extortion scheme Enriquez described remains opaque. Enriquez presented the court with no basis for adjudging the reliability of the posited connection or the reliability of his ultimate conclusion about the meaning of the letter. His testimony failed to satisfy Rule 702's requirements.

In my view, nothing in the record can render harmless the district court's error of admitting Enriquez's key testimony about a "mesa" without a reliability finding, on the ground that reliability was otherwise demonstrated. See Ruvalcaba-Garcia , 923 F.3d at 1190. Simply put, the record contains no explanation or other facts demonstrating his testimony as to the "ranch" letter was reliable.

B.

A district court's error in improperly admitting expert testimony will still be harmless if "it is more probable than not that the jury would have reached the same verdict even if the [expert testimony] had not been admitted." Ruvalcaba-Garcia , 923 F.3d at 1190 (quoting Barabin , 740 F.3d at 465 ).

The government has not shown " ‘it is more probable than not that the jury would have reached the same verdict even if the [expert testimony] had not been admitted.’ " Id. (alterations in original) (quoting Barabin , 740 F.3d at 465 ). To meet this standard would require evidence in the record independent of Enriquez's testimony showing Holguin met all four elements of the RICO conspiracy charge under 18 U.S.C. § 1962(d). Those elements, as relayed in the jury instructions, are as follows:

[1] One, an enterprise existed as alleged in the indictment, namely, the Canta R[anas] organization, and it was engaged in, or its activities affected, foreign or interstate commerce;

[2] number two, that the agreement between two or more persons to conduct or to participate in the conduct of the affairs of the enterprise through a pattern of racketeering activities;

[3] three, that the defendant became a member of the conspiracy knowing of its object and intending to help further or facilitate the scheme; and

[4] number four, that the defendant knew or contemplated that one or more members of the conspiracy, not necessarily the defendant, would commit at least two acts of racketeering in furtherance of the conspiracy.

To demonstrate a "pattern of racketeering activities," the government must prove that at least two predicate acts occurred. 18 U.S.C. § 1961(1), (5). The government need not have proved that Holguin committed the predicate acts, but it must prove that he "knew about and agreed to facilitate the scheme" to commit those acts. Salinas v. United States , 522 U.S. 52, 66, 118 S.Ct. 469, 139 L.Ed.2d 352 (1997).

Here, the jury was instructed that the predicate acts could include: (1) drug trafficking offenses—specifically, possession with the intent to distribute a controlled substance and conspiracy to distribute a controlled substance; (2) money laundering; (3) extortion; and (4) robbery. See 18 U.S.C. § 1961(1)(5). The government argued in closing that "drug trafficking is the primary business of the enterprise." From those predicate offenses, the government explained, the other acts flowed. The enterprise engaged in extortion through "tax collection" from enterprise members and non-members operating in the enterprise's territory, and in money laundering—"what actually happens to the taxes once they are collected." The "enterprise also engaged in robberies."

The special verdict form for Holguin's RICO conspiracy charge did not specify the predicate acts that the jury found Holguin knew about or contemplated. But the jury did indicate that "the pattern of racketeering activity included drug trafficking" to find that Holguin was liable for the distribution of over 50 grams of methamphetamine. Importantly, the jury separately acquitted Holguin of an independent charge of conspiracy to distribute a controlled substance in violation of 21 U.S.C. § 846. The jury also convicted Holguin of assault in violation of the Violent Crimes in Aid of Racketeering Activity ("VICAR") statute, 18 U.S.C. § 1959(a)(6). Viewing Enriquez's testimony against that background, I would hold that without that testimony, there is insufficient record evidence to support Holguin's RICO conspiracy conviction, in particular, with regard to the known or contemplated predicate acts. The prosecution's closing argument is instructive as to how the jury likely viewed the evidence against Holguin. The lynchpin of the government's racketeering theory as to Holguin was that he was involved in running a "mesa" in Chino state prison. That theory was dependent upon Enriquez's interpretation of the word "ranch" as a reference to a "mesa," and, in turn, as a reference to an in-prison extortion scheme connected to drug trafficking.

Specifically, the prosecution pointed to Holguin's letters to Gavaldon, as interpreted by Enriquez, to argue that Holguin had "a very important role in that Canta Ranas organization." The prosecution quoted this statement from Holguin's letter: "I was staying in Chino for the last six months. It's all messed up in that city." Even without Enriquez's testimony, a jury could infer that the letter meant Holguin had been in Chino state prison. But only Enriquez's testimony could extrapolate from "It's all messed up in that city," the elaborate, detailed explanation the government summarized: "Now, you heard from Rene Enriquez that that phrase had a hidden meaning. Enrique Holguin was telling David Gavaldon that he was actually incarcerated in Chino state prison for the last six months and that there was no one there controlling the prison, meaning no one was in charge of the drug trafficking, the extortion, and the other illegal activities in the prison yard." Nothing in Enriquez's testimony explained how he knew that "all messed up" was connected to problems with drug trafficking and extortion.

The government then described other excerpts of the letter, including the statement that Holguin was in "East Chino" where "there was no one there to take care of the ranch, so we as a group got together and looked out for the ranch so it runs smooth." From this statement, a lay person could discern that some prisoners got together to see that the gang's activities in the prison were coordinated. But Enriquez testified to the specific details of how the group ran a "mesa," a term that never appears in the correspondence. The government in closing summarized Enriquez's "mesa" interpretation of the excerpt, stating, "Now, you heard that this portion of the letter was very important. This is actually the heart of the message ... [Holguin]'s telling David Gavaldon that no one was running the prison, so he and a few other Mexican Mafia associates got together to set up a mesa, to make sure the ranch runs smooth. And you heard [from Enriquez] that ‘mesa’ is a Spanish word for ‘table,’ and it refers to a group of Mexican Mafia associates who set up a sort of committee in the prison, and they then oversee all of the illegal activities in that prison, like collecting taxes and smuggling and drugs, and they do that on behalf of particular Mexican Mafia members." In the absence of Enriquez's testimony, it is highly doubtful that this letter alone would have been understood to convey the necessary details of racketeering activity—i.e., possession with intent to distribute, money laundering, robbery, or extortion—to the jury.

Next, the government described Holguin's other "role in the enterprise," asserting that he was "directly involved in the drug trafficking portion of the enterprise." It reminded the jury of balloons of heroin found in a car in which Holguin was a passenger, and more heroin found in the house where he and his brother lived with other family members. The government asked, "Now, how do you know that Enrique Holguin knew about this heroin?" The answer, it turns out, is again "the letter where [Holguin] was saying that he set up a mesa on behalf of David Gavaldon." And again, the term "mesa," and all that the concept entails, appears only in Enriquez's testimony. The government presented no additional evidence connecting Holguin to the drugs in the car or the drugs and paraphernalia found in the residence.

The government also pointed to two other letters in which Holguin asked the recipient to send someone his "hookup." Based on those letters, a reasonable jury could have believed that "hookup" referred to a drug supplier. However, "[m]ere sales to [or purchases from] other individuals do not establish a conspiracy to distribute or possess with intent to distribute." United States v. Mendoza , 25 F.4th 730, 738-740 (9th Cir. 2022) (alteration in original) (quoting United States v. Lennick , 18 F.3d 814, 819 n.4 (9th Cir. 1994) ). Evidence of having a drug supplier is compatible with being "a mere drug user" making "a casual sale [or purchase] of drugs." Id. (quoting U.S. v. Moe , 781 F.3d 1120, 1125 (9th Cir. 2015). No other evidence was presented connecting Holguin to any drug purchases or sales. Without more, the "hookup" letters are insufficient to demonstrate that Holguin engaged in any predicate acts of drug trafficking. Indeed, the jury acquitted Holguin of conspiracy to distribute a controlled substance.

Granted, even if Enriquez's improper testimony were excluded, some portions of his testimony regarding gang tenets and commonly used phrases would remain in the record. For such testimony, general experience, without an explanation of the precise methodology, is sufficient. See Hankey , 203 F.3d at 1169–71. Thus, even though the district court erred in admitting the testimony without making explicit reliability findings, see supra Part II.A; Majority Op. at 853–55, that error was harmless under the first prong of Ruvalcaba-Garcia. But that testimony, and the evidence it interprets, is still insufficient to convict Holguin of the RICO conspiracy charge.

One of Holguin's letters to Gavaldon states that "me ... and a few other camaradas were on the east chino ...." Enriquez testified that the word "camarada" in the Mexican Mafia refers to "a trusted individual who's committed violence, who's killed for the organization, who's on the cusp of being inducted into the organization." The prosecution reminded the jury in closing that "Holguin actually refers to himself as a camarada, and you heard that a camarada is the top level of Mexican Mafia associate." However, gang membership, alone, without evidence of knowledge of specific illegal acts, is insufficient to support a conspiracy conviction. See United States v. Perez , 962 F.3d 420, 445 (9th Cir. 2020) ; United States v. Bingham , 653 F.3d 983, 997 (9th Cir. 2011). Perez upheld a conspiracy conviction based on evidence sufficient to demonstrate that the defendant was "a core member of [the] drug-trafficking operation" who "supervis[ed] drug sales," "protected it with violence," and "helped launder its profits." Perez , 962 F.3d at 445. In contrast, Enriquez's own testimony suggests that a "camarada" is someone who is just "on the cusp of being inducted into the organization." Holguin's use of the word "camarada" in a letter, divorced from Enriquez's testimony interpreting the contents of the letter as referring to Holguin's operation of a "mesa" and of the activities of a "mesa," is insufficient to demonstrate that Holguin was actually aware of or involved in any of the specified predicate offenses.

The government also invoked Holguin's description of his assault of another person in the prison. The evidence connecting the assault to the RICO elements is as follows: Records from a call Holguin made from jail show that he stated that he was "in the hole" because he got into a fight. Explaining how it occurred, he stated "they took me and uh, a friend of mine to the medical.... And, they put somebody down there that wasnt supposed to be near us. You know? One of them PC's.... You know what a PC is? ... Them guys that are no good. They protect them. Protective custody.... Yeah there was one down there so ... I ended up here."

Enriquez interpreted this jail call for the jury. He testified that "PC" is "an acronym for ‘protective custody’ " and is an example of a term used to describe people who have broken one of the Mexican Mafia rules. Enriquez further stated that the rules of the Mexican Mafia are that, "[a]s soon as a Mexican Mafia associate sees" a "PC," the associate, "at first opportunity, regardless of the consequences, he's supposed to kill him."In closing, the government reminded the jury of this testimony, asserting, "[Y]ou actually heard from Rene Enriquez that being no good means that someone has violated one of the rules of the Mexican Mafia and they are now on the list, meaning that any Mexican Mafia associate can and should attack that individual on sight. So Enrique Holguin did just that." A reasonable jury could have concluded, based on Enriquez's testimony, that Holguin did commit the assault in furtherance of some objective of the Mexican Mafia. Indeed, the jury separately convicted Holguin of an independent charge for violent crime in aid of racketeering under the VICAR statute. But that conviction relates only to one act, and it does not establish how that act is related to the specified RICO predicate acts of possession with intent to distribute, money laundering, extortion, or robbery.

Thus, the government's theory as to Holguin's involvement in the RICO conspiracy was dependent upon Enriquez's opinion that, based on the one letter to Gavaldon as Enriquez interpreted it, Holguin was involved in running a "mesa" that engaged in drug trafficking, extortion, and money laundering. The government provided no other independent evidence of Holguin's knowledge of or involvement in those activities.

Crucially, outside of Enriquez's testimony, the government nowhere advances or supports Enriquez's assertions. The government's brief states generally that the defendants extensively cross-examined Enriquez, and that Enriquez's "testimony was corroborated by abundant other evidence, including eyewitness testimony, prison correspondence, prison videos, wiretap calls, jail calls, and physical evidence seized from defendants homes and elsewhere." There is no such corroboration for Enriquez's "mesa" testimony. In fact, other testimony suggested the government declined to pursue additional evidence that could have corroborated the existence of a "mesa." Special Agent Castrilla, one of the investigators assigned to this criminal case, testified as a lay witness and conveyed that he never "ask[ed] anyone at Chino whether the Mexican Mafia or Canta Ranas was running any extortion scheme there in 2007," nor did he "ever interview any former inmates from Chino to ask them if they had ever been extorted back in 2007 by the Mexican Mafia or the Canta Ranas gang."

CONCLUSION

The use of Enriquez's expert testimony in this case is illustrative of the analytical sloppiness that results from the failure to insist on a formal hearing or voir dire, outside the hearing of the jury and before the testimony, for expert testimony by law enforcement-related personnel. As I have demonstrated, Enriquez's key "mesa" testimony was not in fact reliable, at least on any basis revealed on this record. During his mid-trial testimony, the question how Enriquez knew that the vague letter was referring to a very specific "mesa" scheme (even though a "mesa" was never mentioned) by which one-third of the prison drug trafficking profits would be collected by the Mexican Mafia (though neither drug trafficking nor extortion nor a payment amount was mentioned) was never asked or answered. That Enriquez had been a Mexican Mafia member years ago, was knowledgeable about the structure and activities of the Mafia generally, and knew that the gang members communicated in code did not fill that gap.

The admission of Enriquez's testimony was erroneous and not harmless. I would vacate Holguin's conviction of RICO conspiracy involving at least 50 grams of methamphetamine.