Appeal from the United States District Court for the District of Columbia (No. 1:20-cr-00222-1)

Molly E. Runkle , Assistant Federal Public Defender, argued the cause for appellant. With her on the briefs was A. J. Kramer , Federal Public Defender. Tony Axam Jr. , Assistant Federal Public Defender, entered an appearance.

Eric Hansford , Assistant U.S. Attorney, argued the cause for appellee. With him on the brief were Matthew M. Graves , U.S. Attorney, at the time the brief was filed, and Chrisellen R. Kolb and Elizabeth H. Danello , Assistant U.S. Attorneys.

Before: HENDERSON, PAN and GARCIA, Circuit Judges .

Opinion for the Court filed by Circuit Judge HENDERSON.

Opinion concurring in part and concurring in the judgment filed by Circuit Judge GARCIA.

Page 2KAREN LECRAFT HENDERSON, Circuit Judge : On January 20, 2020, reports of gunfire took police to a rowhouse in Southeast Washington, D.C. There, officers found spent shell casings but no witnesses and no suspect. They reviewed footage from a nearby pole-mounted surveillance camera, installed days earlier for an unrelated investigation. The video showed someone stepping from the rowhouse, firing a gun into the air and retreating inside. Later that day, officers executed a search warrant at the residence. Inside, they recovered a large quantity of narcotics, digital scales and a firearm. As the officers entered, Demetrius Green tried to flee through the back door but was arrested. A jury later convicted him of several federal drug and firearm offenses based on the surveillance footage and the evidence recovered from the house.

Green now challenges his convictions on three grounds. First, he contends that the use of the pole camera violated the Fourth Amendment, asserting that it constituted a warrantless search infringing on his reasonable expectation of privacy. Second, he argues that the evidence at trial was insufficient to establish that he constructively possessed the drugs found in the rowhouse. Third, he claims that the district court erred by admitting two exhibits—a photograph of a bag of powder on top of a digital scale and a text message referring to a drug sale—arguing that both amounted to impermissible character evidence and any probative value was substantially outweighed by the risk of unfair prejudice. We disagree. The use of the pole-camera footage did not constitute a search under the Fourth Amendment. The evidence at trial was sufficient to establish constructive possession, given Green’s documented connection to the residence and the items recovered. And the challenged exhibits were relevant and properly admitted. Alternatively, even if the exhibits were admitted in error, any error was harmless. Accordingly, we affirm.

In the early morning hours of January 20, 2020, an officer of the Metropolitan Police Department (MPD) heard gunshots in her patrol area in Southeast Washington, D.C. ShotSpotter (a gunshot-detection system) also alerted nearby, close to 917 Wahler Place, Washington, D.C. (917 Wahler), a rowhouse in a public-housing complex. When officers responded, they discovered several spent shell casings scattered on the rear steps of the residence but saw no immediate suspects or eyewitnesses.

Seeking more evidence, MPD officers reviewed surveillance footage from a pole-mounted camera that had been installed two days earlier by the Bureau of Alcohol, Tobacco, Firearms and Explosives (ATF), which was conducting a separate investigation. An ATF agent testified that the camera was monitoring the open-air courtyard behind the rowhouses in response to reports of drug trafficking and gun violence in the area. The camera continuously recorded activity in the shared courtyard and the rear entryways of several rowhouses, including 917 Wahler. See Gov't Ex. 52.

When MPD officers examined the footage from the time of the reported gunfire, they observed an individual emerging from the rear door of 917 Wahler at around 4:45 a.m. The individual—a male dressed in a dark hooded sweatshirt, white pants and tennis shoes-raised a firearm and appeared to discharge multiple rounds into the air before retreating inside. Id. The handgun's muzzle flash was clearly visible on the recording. See Gov’t Exs. 53 & 54. The footage also showed a male stepping out of the same door at around 12:25 p.m. See Gov’t Ex. 52. An ATF agent later identified him as Demetrius Green.

Based on that investigation, officers applied for and obtained a search warrant for unlawful firearms or ammunition Page 4at 917 Wahler. The warrant was executed later the same day. As officers approached, Green attempted to exit through the back door but, after seeing police, retreated into the residence. Officers entered the residence and took Green into custody without incident. No one else was present at the house. In the kitchen, officers discovered a clear plastic bag containing 700 oxycodone pills, sorted into smaller bags based on strength, hidden inside a toaster oven. In a nearby cabinet, officers recovered three bags containing 288 hydromorphone pills (another opioid). In other cabinets, they discovered sandwich bags, two digital scales, a Pyrex cup containing white residue and a bank card bearing the name “Demetrius Green.”

Upstairs, officers found a loaded firearm-a Glock 23 fitted with a “conversion device” that allowed the handgun to fire automatically.¹ DNA testing on the gun revealed a major male contributor and a minor contributor. A DNA expert later excluded Green as the major contributor but concluded that there was “moderate support” that he was the minor contributor. App. 870-71. A firearms examiner determined that the recovered shell casings were from rounds fired by that gun. There were three bedrooms upstairs (identified at trial as Bedrooms A , B and C ). Of the three, Bedroom A appeared to be the most occupied room. There, officers found a driver’s license, a learner’s permit and a bank card, all bearing Green’s name. They also found a shoebox containing thirty-three small packages of crack cocaine, several articles of clothing consistent with those worn by the shooter appearing on the pole-camera footage and a set of green-and-black headphones. In Bedroom B , they found a shoebox containing a DHL label addressed to Green at 917 Wahler. There was also a preemployment drug testing form in Green’s name and a notarized letter-dated several months earlier-from his sister stating Page 5that Green was temporarily residing with her at 102 Irvington Street. Green’s identification cards listed the same address. In Bedroom C , officers found a “significant amount” of cash but did not seize it. App. 690-91.

When officers arrested Green, he had a Motorola cellphone and another pair of green-and-black headphones. Officers later extracted information from the phone, which included an array of photographs, GPS data, text messages and emails, some of which appeared to link Green to 917 Wahler. Relevant here, the government later introduced two exhibits from that data extraction that Green challenges: (1) a photograph of a bag filled with white powder resting on a digital scale reading '13.4g' (Exhibit 101), and (2) a text message Green sent on December 30 saying: “I got sum tree come support my hustle” (Exhibit 110).

On October 15, 2020, a grand jury indicted Green on unlawful possession of a firearm by a convicted felon in violation of 18 U.S.C. § 922(g)(1) (Count 1); unlawful possession of a machinegun in violation of 18 U.S.C. § 922(o) (Count 2); three counts of unlawfully possessing with intent to distribute cocaine base, oxycodone and hydromorphone in violation of 21 U.S.C. § 841(a)(1), (b)(1)(C) (Counts 3-5); and possessing a machinegun in furtherance of a drug trafficking offense in violation of 18 U.S.C. § 924(c)(1)(B)(ii) (Count 6). The government later voluntarily dismissed Count 2. App. 79.

Before trial, Green moved to suppress the footage obtained from the pole camera, arguing that its use violated the Fourth Amendment prohibition on unreasonable searches. App. 40-48. The district court denied the motion. App. 37384. It was not convinced that Green had established any Fourth Amendment standing in relation to 917 Wahler, given his unclear connection to the residence. App. 382-83. But even assuming Green had standing, the court found that he did “not have an expectation of privacy in the particular exposed, Page 6undifferentiated space captured [by the pole camera], particularly given the limited period of time of the surveillance and the manner in which the surveillance was being conducted.” App. 383. And the court determined that, to the extent Green did have any expectation of privacy in that space, 'he voluntarily waived it' by firing the handgun there. Id.

The government moved in limine to admit two exhibits from Green’s cellphone: the photograph of a digital scale with a bag of powder (Exhibit 101) and the text message referencing “tree” (Exhibit 110). The district court admitted both under Federal Rules of Evidence 404(b) and 403. As to Exhibit 101, the court found that the photograph's presence on Green's cellphone was 'relevant to show identity, knowledge and intent” and concluded that Green’s objections went “largely to [the exhibit’s] weight and not its admissibility.” App. 431-33. As to Exhibit 110, the court acknowledged that the message referenced marijuana rather than the drugs charged in the indictment but held it was nevertheless 'relevant to the question of whether the defendant had the intent to distribute [narcotics] on January 20, 2020.” App. 103. The court noted that any risk of unfair prejudice could be mitigated by a limiting instruction under Rule 105, clarifying that the exhibits were not to be used for an improper purpose. App. 103, 432. At trial, however, Green’s counsel opposed giving such an instruction, explaining that doing so might unduly emphasize the exhibits—“a strategic trial decision made by thoroughly vetted counsel.” App. 1183. The district court required counsel to confirm that, by declining the instruction, he was “waiving the right to argue later that the jury must have used it for an improper purpose.” App. 1182-83. With that understanding, no limiting instruction was given as to either exhibit.

After a five-day trial, the jury convicted Green on four of five counts, failing to reach a verdict on Count 6. App. 10708. The government later dismissed that charge. App. 105. On June 16, 2023, the district court sentenced Green to 84 months of imprisonment, followed by 36 months of supervised Page 7release. Green timely appealed. The district court had jurisdiction under 18 U.S.C. § 3231 and we have jurisdiction under 28 U.S.C. § 1291.

Green’s challenges require us to apply different standards of review. First, in reviewing a denied motion to suppress, we review 'legal conclusions de novo and factual findings for clear error.” United States v. Brown , 125 F.4th 1186, 1201 (D.C. Cir. 2025).

Second, in assessing a challenge to the sufficiency of the evidence, we examine the evidentiary record de novo but 'consider it in the light most favorable to the government, and . . . will affirm a guilty verdict where ‘ any rational trier of fact could have found the essential elements of the crime beyond a reasonable doubt.'' United States v. McGill , 815 F.3d 846, 917 (D.C. Cir. 2016) (quoting United States v. Wahl , 290 F.3d 370, 375 (D.C. Cir. 2002)).

Third, we review the district court’s admission of evidence under Rules 403 and 404(b) for an abuse of discretion. Id. at 880. But '[a]n erroneous admission of other crimes evidence must be disregarded as harmless error unless it had a substantial and injurious effect on the jury's verdict.' Id. (citation modified).

Green raises three issues on appeal. First, he contends that the pole-camera evidence violated his Fourth Amendment rights and should have been suppressed. Second, he asserts that there was insufficient evidence that he constructively possessed the drugs found at 917 Wahler. Third, he challenges the admission of Exhibits 101 and 110 as impermissible character evidence and improperly risking unfair prejudice.

Green’s primary claim is that the district court erred by denying his motion to suppress the video evidence recovered from the pole camera with an unobstructed view of the rear of 917 Wahler. Accessing that footage, he argues, was an unreasonable search that violated his reasonable expectation of privacy as protected by the Fourth Amendment. He contends that because the pole-camera evidence was obtained unlawfully and was the foundation for the later warrant, the rest of the evidence against him should also have been excluded as tainted by that unconstitutional search.

The Fourth Amendment provides: “The right of the people to be secure in their persons, houses, papers, and effects, against unreasonable searches and seizures, shall not be violated.” U.S. Const. amend. IV. Warrantless searches are ordinarily ' per se unreasonable under the Fourth Amendment-subject only to a few specifically established and well-delineated exceptions.” Katz v. United States , 389 U.S. 347, 357 (1967). If the government oversteps that constitutional boundary, the remedy is generally exclusion— courts must suppress the unlawfully obtained evidence and any derivative evidence tainted by the violation unless an exception applies. Utah v. Strieff , 579 U.S. 232, 237 (2016).

The fundamental question Green poses is whether the government’s use of a pole camera under the circumstances here constituted a “search” within the meaning of the Fourth Amendment. The U.S. Supreme Court has set forth two tests to assess whether government conduct constitutes a search. First, the “common-law trespassory test” considers whether the government has physically intruded on private property. United States v. Jones , 565 U.S. 400, 409 (2012); see also Florida v. Jardines , 569 U.S. 1, 5-10 (2013). That test is not relevant here because the pole camera did not physically intrude at 917 Wahler. Second, the government performs a search when it intrudes upon a defendant’s reasonable Page 9expectation of privacy. See Katz , 389 U.S. at 361 (Harlan, J., concurring); Jones , 565 U.S. at 406. Determining whether such an expectation exists involves a two-step inquiry: first, whether the defendant exhibited an actual, subjective expectation of privacy; and second, whether that expectation is one society is prepared to recognize as objectively reasonable. Smith v. Maryland , 442 U.S. 735, 740-41 (1979). If both prongs are satisfied, the conduct qualifies as a search and any resulting evidence is subject to suppression absent a warrant or a valid exception. See, e.g. , Carpenter v. United States , 585 U.S. 296, 304 (2018). Here, Green’s challenge implicates two strands of Fourth Amendment jurisprudence: the public-view doctrine and the mosaic theory. We consider each in turn.

The Supreme Court has consistently held that individuals have no reasonable expectation of privacy in areas exposed to the public. See Katz , 389 U.S. at 351 ('What a person knowingly exposes to the public, even in his own home or office, is not a subject of Fourth Amendment protection.'); Kyllo v. United States , 533 U.S. 27, 32 (2001) (“[E]xamining the portion of a house that is in plain public view . . . is no 'search' at all' under the Fourth Amendment.). It is nearly axiomatic that “Fourth Amendment protection of the home has never been extended to require law enforcement officers to shield their eyes when passing by a home on public thoroughfares.' California v. Ciraolo , 476 U.S. 207, 213 (1986); see also Dow Chem. Co. v. United States , 476 U.S. 227, 239 (1986) (taking aerial photos of an industrial plant was not a search); Collins v. Virginia , 584 U.S. 586, 600 (2018) (affirming the “ability to observe inside curtilage from a lawful vantage point”). Our court has similarly recognized that “there is no reasonable expectation of privacy in the movement of objects outside a residence where they can be viewed from a public route or adjoining premises, nor in activities conducted in the curtilage of a home, even behind a hedge or fence, if they may be viewed by naked-eye observation.' Brennan v.

Page 10Dickson , 45 F.4th 48, 62 (D.C. Cir. 2022) (citation modified) (discussing a constitutional challenge to an agency rule regarding drone flights). Courts have referred to this principle as the 'public-view doctrine': an observation of an area exposed to public sight from lawful vantage points does not constitute a Fourth Amendment search.²

Both Carpenter and Jones also noted that the public-view doctrine remains good law. See Carpenter , 585 U.S. at 316 ('We do not . . . call into question conventional surveillance techniques and tools, such as security cameras.”); Jones , 565 U.S. at 412 ('This Court has to date not deviated from the understanding that mere visual observation does not constitute a search.'). Apart from Judge Barron's concurrence in MooreBush , every circuit court to consider whether a pole camera observing a public area violates a reasonable expectation of privacy has held it does not.³ Instead, pole cameras 'qualify as Page 11a conventional surveillance technique” to observe areas open to public view, as specifically blessed by Carpenter . Tuggle , 4 F.4th at 526 (citation modified).

Under the standard application of the public-view doctrine—that is, without considering the mosaic theory—this case is not a close one.

The first question under Katz is whether Green exhibited an actual, subjective expectation of privacy in the rear of 917 Wahler Place.⁴ A defendant manifests such a desire if he, at a minimum, takes 'normal precautions to maintain his privacy,' such as installing a high fence to prevent 'casual, accidental observation' from sidewalk traffic. Ciraolo , 476 U.S. at 211-12 (quoting Rawlings v. Kentucky , 448 U.S. 98, 105 (1980)); see also Florida v. Riley , 488 U.S. 445, 450 (1989) (holding that a defendant who had taken “precautions . . . against ground-level observation” exhibited a subjective expectation of privacy). Several circuits have noted the relevance of defendants' failures to exhibit a subjective expectation of privacy in areas surveilled by pole cameras. See, e.g. , United States v. House , 120 F.4th 1313, 1317 (7th Cir. 2024); Harry , 130 F.4th at 348.⁵

Page 12Green exhibited no expectation of privacy in the rear of 917 Wahler. As the district court observed, the area was “well lit,” “entirely open to public view” and was not differentiated “from the space behind any of the other adjacent townhouses.” App. 377. Although a chain-link fence surrounded the rear area of all the townhouses, no fence or hedge separated 917 Wahler from its neighbors. Id. Green has also not established—and indeed disclaims—that 917 Wahler was his residence or abode, where a subjective expectation of privacy might be more readily established. Granted, Green is correct that none of those facts precludes him from establishing a subjective expectation of privacy, see Appellant Br. 37, but he has offered nothing in their stead. Moreover, if Green ever did show a subjective expectation of privacy, he plainly did not when he stepped outside to fire a handgun into the air in full view of the neighborhood. See App. 378.⁶

Green fares no better under Katz ’s second prong. The rear of 917 Wahler was plainly visible from multiple public vantage points—including the adjacent parking lot, a public walkway and the windows of nearby residences. See App. 377. The pole camera was mounted in a lawful location and recorded only what was already in plain sight. It did not enter the property, Page 13peer through barriers or use any technology to reveal what was otherwise hidden. Cf. Kyllo , 533 U.S. at 40 (holding that use of “a device that is not in general public use” can sometimes constitute a search). It captured what any passerby or neighbor could have seen with the naked eye. That the observation was recorded and extended over time does not alter that analysis. The camera saw no more-and, except for its elevation, no differently-than the public could have seen all along. See Gov't Ex. 52. Under the public-view doctrine, then, the surveillance here did not violate any objectively reasonable expectation of privacy.⁷

The crux of Green’s challenge—indeed all of the recent challenges to the use of pole cameras—is that the aggregation of surveillance over time violates a reasonable expectation of privacy, even if any brief or isolated observation would not. The argument rests on the so-called 'mosaic theory,' which suggests that the government’s collection of numerous discrete data points over time can create an impermissibly invasive picture of an individual’s private life, even if any individual Page 14data point, standing alone, would not constitute a search. See generally Tuggle , 4 F.4th at 517 (collecting academic discussions of the theory).

The mosaic theory first emerged in our court’s decision in United States v. Maynard , in which case we concluded that tracking a car using a planted GPS device constituted a search. 615 F.3d 544, 560-62 (D.C. Cir. 2010), aff’d sub nom. Jones , 565 U.S. at 400. We held:

[T]he whole of a person’s movements over the course of a month is not actually exposed to the public because the likelihood a stranger would observe all those movements is not just remote, it is essentially nil. It is one thing for a passerby to observe or even to follow someone during a single journey as he goes to the market or returns home from work. It is another thing entirely for that stranger to pick up the scent again the next day and the day after that, week in and week out, dogging his prey until he has identified all the places, people, amusements, and chores that make up that person’s hitherto private routine.

Id. at 560. The court continued: 'Prolonged surveillance reveals types of information not revealed by short-term surveillance, such as what a person does repeatedly, what he does not do, and what he does ensemble. These types of information can each reveal more about a person than does any individual trip viewed in isolation.' Id. at 562.

The Supreme Court reviewed that decision in United States v. Jones , 565 U.S. at 400. There, the Court’s majority avoided the question of whether aggregated data could ever amount to a search, relying instead on the trespassory test. Id. at 404-07. Concurring, Justice Alito-joined by Justices Ginsburg, Breyer and Kagan—noted that, although “relatively short-term monitoring of a person's movements on public Page 15streets accords with expectations of privacy,” it was possible that 'longer term GPS monitoring' could impinge on such expectations. Id. at 430 (Alito, J., concurring). Writing separately, Justice Sotomayor similarly voiced a concern that the capabilities of GPS monitoring—including its inexpensiveness, precision, efficiency and limitless storage— posed serious concerns: “GPS monitoring generates a precise, comprehensive record of a person's public movements that reflects a wealth of detail about [a defendant's] familial, political, professional, religious, and sexual associations.” Id. at 415 (Sotomayor, J., concurring). Those attributes, she wrote, should be considered when assessing societal expectations of privacy. Id. at 416.

Since Jones , the Supreme Court has signaled a continuing willingness to consider the aggregation of data as distinctively problematic. See, e.g. , Riley v. California , 573 U.S. 373, 394 (2014) (noting that the “sum of an individual’s private life” can be adduced from a warrantless search of a cellphone). In United States v. Carpenter , the Court concluded that collecting seven days of cell-site location information (CSLI) was a search under the Fourth Amendment (despite precedent suggesting a contrary result). 585 U.S. at 315, 310 n.3.⁸ The Court endorsed the theory propounded by Justices Sotomayor and Alito and found that 'individuals have a reasonable expectation of privacy in the whole of their physical movements,' even if exposed to public view, and that accessing the CSLI data contravened that expectation. Id. at 310-11. It noted that the CSLI data presented “even greater privacy concerns” than the vehicle GPS data in Jones because a cellphone 'follows its owner beyond public thoroughfares and into private residences, doctor’s offices, political headquarters, and other potentially revealing locales,” allowing Page 16the government to create a comprehensive map of a person’s movements with “just the click of a button.” Id. at 311-12.

At the same time, the Court cautioned that its decision was “a narrow one' and that it did not “call into question conventional surveillance techniques and tools, such as security cameras.' Id. at 316. It also left undisturbed the holding in Knotts , in which case it held that using a more “rudimentary tracking” device that simply “augmented visual surveillance' for discrete intervals was not a search. Id. at 306 (citing United States v. Knotts , 460 U.S. 276, 281-85 (1983)) (citation modified).

Pole cameras pose a special challenge to the mosaic theory. In one sense, they are among the most common forms of surveillance. They rely on a public, unobstructed vantage point and off-the-shelf technology, not unlike an agent with binoculars perched atop a telephone pole. But unlike that unfortunate agent-who will get bored, blink or need to stretch—a pole camera never looks away. It records everything, 24/7, for weeks or months, even years, preserving everything it sees. By aggregating that data, critics worry, the government can reconstruct not only what happens at a location, but also the patterns and relationships of the individuals who pass through it. Little about the underlying camera technology has changed in recent years but Carpenter ’s embrace of the mosaic theory has made pole-camera challenges newly relevant to the Fourth Amendment. And as other technologies like artificial intelligence and facial recognition improve, the potential capabilities of ubiquitous cameras may grow exponentially.

Still, other circuit courts have consistently rejected attempts to extend the mosaic theory to pole cameras. See supra n.3. Those decisions primarily rely on the continuing vitality of the public-view doctrine as the Supreme Court has articulated it, including Carpenter ’s reassurance that it did not invalidate the use of traditional surveillance techniques likePage 17 'security cameras.' See, e.g. , Tuggle , 4 F.4th at 525-26 (quoting Carpenter , 585 U.S. at 316). They also sometimes note that, even if pole-camera surveillance could violate the Fourth Amendment, the duration of the evidence in their respective cases was not sufficient to establish such a violation. See, e.g. , id. (18 months permissible but noting the “obvious line-drawing problem: How much pole camera surveillance is too much?”); Harry , 130 F.4th at 348 (50 days); Hay , 95 F.4th at 1316-17 (68 days); see also Carpenter , 585 U.S. at 322-23 (Kennedy, J., dissenting) (deriding the arbitrariness of the majority’s six-day cutoff).

More fundamentally, there seems a material difference between the types of data the Supreme Court has found to implicate mosaic-type concerns—such as omnipresent location tracking-and the more limited information a fixed pole camera can capture. The cell-site location data in Carpenter , like the GPS data in Jones , 'provide[d] an all-encompassing record of the holder’s whereabouts,” revealing “not only his particular movements, but through them his ‘familial, political, professional, religious, and sexual associations.’” Carpenter , 585 U.S. at 311 (quoting Jones , 565 U.S. at 415 (Sotomayor, J., concurring)). Those kinds of data have a 'retrospective quality” that allows the government to reconstruct a suspect’s past-surveilling him before he was ever a suspect-and to access “a category of information otherwise unknowable.” Id. at 312.

By contrast, the observational power of a single pole camera is both fixed and limited. The Fourth Amendment 'protects people, not places,' Katz , 389 U.S. at 351, so the simple fact that a public-facing camera records a location continuously is not itself constitutionally suspect. The question is what the government in fact learns about an individual from that camera’s limited perspective. The information may still be meaningful—agents might see when a person comes and goes, who visits him or how often he mows the lawn—but it would tell them nothing about him outside the frame. The Page 18footage, in other words, 'only depict[s] one small part of a much larger whole.' Tuggle , 4 F.4th at 524. There is a difference between location-tracking technologies, which follow a person broadly but shallowly, and fixed surveillance tools, which observe narrowly but in greater depth. As an analogy, if an individual’s daily patterns were the surface of the ocean, CSLI or GPS data would be a buoy drifting across the water, reporting any contacts along the way. But a pole camera is anchored in place—it might provide complete information about the reef it rests on but say little about the sea beyond.

In any event, this case is a poor candidate for applying the mosaic theory to pole-camera surveillance. The footage here spanned only two days—far shorter than the weeks or months involved in other cases where courts have had reservations about cumulative observation. Given that brief duration, the government had no opportunity to compile a retrospective record of Green’s movements or reconstruct his patterns of life. Nor did the footage itself reveal much—the camera captured just two fleeting moments in which Green stepped outside, offering no insight into his associations, routines or private conduct in the manner condemned in Carpenter . Whatever the outer bounds of the mosaic theory may be, they are not approached here. This was short-term, public-facing surveillance, limited in scope. It did not implicate the privacy concerns the mosaic theory is intended to address.⁹

We emphasize, however, the limits of our holding. We do not suggest that pole-camera surveillance could never amount to a Fourth Amendment search. In another case, the technology might be used over longer periods, with more cameras, or in combination with other tools-such as facial recognition, Page 19automated tracking or artificial intelligence-to build a far more comprehensive portrait of an individual’s life. Whether such surveillance would raise constitutional concerns, however, is a question left for another day.

In sum, the rear of 917 Wahler was exposed to public view and the surveillance was brief and unsophisticated. Under the public-view doctrine, Green lacked any objectively reasonable expectation of privacy. And because the observation was both limited and discrete, the mosaic theory does not change that result.¹⁰

Green next challenges the sufficiency of the evidence supporting his constructive possession of the drugs undergirding three of his convictions. To convict Green, the government was required to prove that Green possessed crack cocaine for Count 3, oxycodone for Count 4 and hydromorphone for Count 5. See App. 37-38. Green contends that there was insufficient evidence that he had constructive possession of any of those drugs.

In a challenge to the sufficiency of the evidence, we consider the evidentiary record de novo but 'consider it in the light most favorable to the government, and . . . will affirm a guilty verdict where ‘ any rational trier of fact could have found the essential elements of the crime beyond a reasonable doubt.'' McGill , 815 F.3d at 917 (quoting Wahl , 290 F.3d at 375). That is a “highly deferential standard,” reflecting that the jury is 'entitled to draw a vast range of reasonable inferences from evidence, but may not base a verdict on mere Page 20speculation.” United States v. Slatten , 865 F.3d 767, 792 (D.C. Cir. 2017) (citation modified).

Possession of contraband may be either actual or constructive. See United States v. Alexander , 331 F.3d 116, 127 (D.C. Cir. 2003). Actual possession requires 'direct physical control.” Henderson v. United States , 575 U.S. 622, 626 (2015). Because Green was not found in actual possession of any controlled substances, the government had to prove that he constructively possessed them. “Constructive possession is established when a person, though lacking such physical custody, still has the power and intent to exercise control over the object,' thus 'maintaining control' over the object. Id. We generally permit an inference of constructive possession in two circumstances. First, it may be inferred that a “sole occupant” of a residence exercises dominion and control over its entire contents. United States v. Morris , 977 F.2d 617, 620 (D.C. Cir. 1992) ('A jury is entitled to infer that a person exercises constructive possession over items found in his home.'). Alternatively, if the home is shared or if the defendant is simply discovered close to the contraband, there must be other evidence connecting him to it. See United States v. Dorman , 860 F.3d 675, 679-81 (D.C. Cir. 2017). For example, “connection with a gun, proof of motive, a gesture implying control, evasive conduct, or a statement indicating involvement in an enterprise—coupled with proximity may suffice.' Alexander , 331 F.3d at 127 (citation modified).

Green argues that no reasonable jury could find that he was the sole occupant of 917 Wahler because evidence suggested a link to at least one other person, including a bank card in the name of 'Roneka Eaton.' See Appellant Br. 45-47. He is incorrect.

A jury could have permissibly found that Green was the sole occupant of 917 Wahler. As described above, one room of the residence was the most lived in and it was full of Green’s possessions. Indeed, Green’s possessions were found Page 21throughout the house, including in multiple bedrooms and the kitchen. Cf. United States v. Dykes , 406 F.3d 717, 722 (D.C. Cir. 2005) (holding that a defendant constructively possessed contraband where his personal belongings were found in the same bedroom as drugs and paraphernalia). There was further evidence that Green had used 917 Wahler for some time, including a shipping label bearing his name along with several text messages and emails from his cellphone linking him to 917 Wahler. See, e.g. , App. 795-802. Law enforcement also observed Green in the pole-camera footage as the only occupant for the day and night before his arrest and no one else was seen coming or going during that time.

It is true that Green was neither the owner nor lessee of 917 Wahler and some items were discovered that did not appear to be his. For example, Green points to Eaton’s bank card and a letter addressed to “Ronnika Jennings.” Appellant Br. 45-46. There was also evidence that Green was temporarily living with his sister on Irvington Street. Id. at 46. But much of that evidence was contradicted. See, e.g. , App. 688 (ATF agent’s testimony that she had found no evidence of a woman living at 917 Wahler). And, as the government points out, although that evidence might have led a jury to find that Green was not an occupant of 917 Wahler, none of that evidence would require it to do so. See Dykes , 406 F.3d at 722 (noting that the jury may infer dominion and control over a residence even if it is shared, although the inference may be less strong); see also Morris , 977 F.2d at 620. Because the jury could have reasonably concluded that Green exercised dominion and control over the whole residence, that is enough.

In sum, there was sufficient evidence for the jury to conclude that Green was the sole occupant of 917 Wahler and thus was in constructive possession of all of the contraband discovered therein.

Green finally challenges the district court’s admission of Exhibits 101 and 110 under Federal Rules of Evidence 404(b) and 403. Exhibit 101 consists of a photograph found on Green’s cellphone depicting a white powdery substance on a digital scale and Exhibit 110 contains a text message in which Green apparently inquired about selling marijuana (“I got sum tree come support my hustle”). See App. 207, 236.

Relevant evidence is admissible, subject to certain limits. Fed. R. Evid. 402. Whether a piece of evidence is relevant turns on whether it tends to make any fact of consequence to the determination of the action more or less probable. Fed. R. Evid. 401. Whether a piece of evidence is relevant may depend on an underlying factual predicate. For example, a photograph is relevant only if there is sufficient evidence to establish that it depicts what the proponent claims it does. In such circumstances, a proponent must produce enough evidence to permit a reasonable jury to find the predicate fact by a preponderance of the evidence. Fed. R. Evid. 104(b); Huddleston v. United States , 485 U.S. 681, 689-90 (1988).

Rule 404(b)(1) separately restricts one type of otherwise relevant evidence—that offered to prove a criminal character or propensity ( i.e. , that because a defendant committed a previous crime, he more likely committed the charged one). The prohibition does not apply, however, if the same evidence is offered for other purposes “such as proving motive, opportunity, intent, preparation, plan, knowledge, identity, absence of mistake, or lack of accident.' Fed. R. Evid. 404(b)(2). Even if the evidence could conceivably be used in an improper way, it is still admissible under Rule 404(b)(2) provided it has any purpose other than seeking to prove a criminal propensity. See United States v. Cassell , 292 F.3d 788, 792 (D.C. Cir. 2002).

Page 23Otherwise admissible evidence may also be excluded “if its probative value is substantially outweighed by a danger of . . . unfair prejudice, confusing the issues, misleading the jury, undue delay, wasting time, or needlessly presenting cumulative evidence.” Fed. R. Evid. 403. A risk of “unfair prejudice” can arise when “some concededly relevant evidence [could] lure the factfinder into declaring guilt on a ground different from proof specific to the offense charged.' Old Chief v. United States , 519 U.S. 172, 180 (1997). That said, 'Rule 403 'tilts, as do the rules as a whole, toward the admission of evidence in close cases,’ even when other crimes evidence is involved.' Cassell , 292 F.3d at 795 (quoting United States v. Moore , 732 F.2d 983, 989 (D.C. Cir. 1984)).

Green was charged with three counts of possessing with intent to distribute certain controlled substances— hydromorphone, oxycodone and crack cocaine. To convict him of those charges, the government was required to prove that, on January 20, 2020, Green possessed the controlled substance, knowingly and intentionally, with the specific intent to distribute it. See 21 U.S.C. § 841(a)(1), (b)(1)(C). Green's identity, knowledge and intent were thus all relevant as facts of consequence to the verdict.

Combining the applicable standards, the district court was charged with answering three questions: was the proffered evidence relevant to at least one of the charged offenses, was it improperly offered to show Green’s criminal character and was its probative value substantially outweighed by the risk of unfair prejudice?

a. Exhibit 101

The district court determined that Exhibit 101 was relevant because it could be used to show Green’s identity as well as his knowledge or intent to distribute illegal drugs. App. 431-33. Because it was being offered for those purposes, the court reasoned, it did not breach Rule 404(b). Id. The court also found there was a low risk of any unfair prejudice, particularly Page 24with a limiting instruction, and thus the photograph’s probative value outweighed that concern under Rule 403. Id. Green argues that Exhibit 101 should have been prohibited under Rule 404(b) because a picture of one powdery substance “proved nothing” about any other drugs concealed in 917 Wahler, the identity of the person in the photo was not at issue at trial, the government did not prove the powdery substance was any kind of narcotic, the government did not prove the photo was taken by Green and a two-month-old photograph was too remote to bear on Green’s knowledge and intent at the time of his arrest. See Appellant Br. 59-61.

Relevance. Exhibit 101 was relevant but its relevance was conditioned on several predicate facts. Although couched in Rule 404(b), Green’s objections primarily contest the photograph’s relevance—or conditional relevance—not its use as improper character evidence. To illustrate, if the photograph was taken by Green and the substance pictured was cocaine packaged for sale, there is no dispute that it would be relevant to show his identity (as the person in possession of the cocaine found at 917 Wahler), knowledge (of cocaine or other illegal narcotics), or intent (to distribute those drugs). None of those purposes makes it impermissible propensity evidence. Green attacks the predicate factual bases of those conclusions (who took the photograph and what it shows), which bear on conditional relevance under Rule 104(b), requiring only evidence sufficient for a reasonable jury to find the predicate facts by a preponderance of the evidence. Huddleston , 485 U.S. at 689-90.

The district court determined the photograph was relevant because it was taken on Green’s cellphone and it showed what appeared to be illegal narcotics packaged and weighed for distribution on one of the scales recovered during the government's search of 917 Wahler. App. 431. It then properly decided (albeit implicitly) that a reasonable jury could find those predicate facts by a preponderance of the evidence. See Huddleston , 485 U.S. at 689-90. At least some evidence Page 25suggested that Green had taken the photograph, given that the image was recovered from his cellphone, contained a distinctive green-and-black headphone cord resembling ones found in his possession and depicted a digital scale similar to those seized from 917 Wahler. Likewise, the powdery substance pictured could have reasonably been found to be an illegal drug packaged for distribution, based on its appearance, the expert testimony describing how drugs are typically weighed and prepared for sale and the presence of identical scales found alongside the recovered narcotics. Of course, the jury could also have rejected those conclusions and disregarded Exhibit 101 but that fact did not make the photograph irrelevant.

Green also suggests that, even if the photograph depicts him with a bag of suspected narcotics, it was taken too long before his arrest to be relevant. Once again, that is not a challenge to improper character evidence under Rule 404(b) but a straightforward relevance objection. But time does not entirely strip evidence of its relevance, provided it is not excessively remote and the context is sufficiently similar. See United States v. Douglas , 482 F.3d 591, 597-98 (D.C. Cir. 2007) (holding that evidence of an earlier arrest for possession with intent to distribute crack cocaine was relevant to a new charge more than a year later); Cassell , 292 F.3d at 793 (threeyear-old arrest on the same crime relevant to new charge). Although the passage of time may bear on the probative value of the evidence, that consideration is properly the domain of Rule 403 or fodder for cross-examination, not the relevance inquiry. The district court was therefore well within its discretion in determining that a two-month interval was not so remote as to render Exhibit 101 irrelevant, especially given its connection to narcotics distribution activities involving similar (or even the same) paraphernalia. App. 431-32.

Page 26Propensity. Because Exhibit 101 was relevant, the next question is whether it was offered to show Green’s criminal propensity and thus prohibited under Rule 404(b).¹¹ It was not. The government introduced Exhibit 101 to establish Green’s knowledge of and intent to distribute illegal narcotics—both disputed elements of the charged offenses. This court has repeatedly found that evidence of past drug distribution is admissible to establish the charged offense, provided at least some of the characteristics are the same. See, e.g. , United States v. Pettiford , 517 F.3d 584, 590 (D.C. Cir. 2008) (evidence of earlier distribution of crack cocaine admissible to show knowledge of the substance and intent to distribute it again); United States v. Crowder , 141 F.3d 1202, 1208-09 (D.C. Cir. 1998) (en banc); Douglas , 482 F.3d at 600. Here, the government used Exhibit 101 for a permissible purpose and thus Rule 404(b) did not bar its admission.

Risk of Unfair Prejudice. Green finally argues that even if Exhibit 101 was relevant and otherwise admissible, it should have been excluded because its probative value was outweighed by the risk of unfair prejudice under Rule 403. The district court did not gravely abuse its discretion in finding the contrary. See Douglas , 482 F.3d at 596. The photograph was at least somewhat probative of Green's identity, intent and knowledge. App. 431. On the other side of the ledger, the risk of unfair prejudice posed by the photograph Green raises was the potential that it would be used as improper propensity evidence. Yet the district court acknowledged that potential hazard and found that any such prejudice could be eliminated Page 27through a limiting instruction. App. 432. Such an instruction would ordinarily be sufficient to alleviate the risk identified by Green, absent some showing of “compelling or unique evidence of prejudice.' Cassell , 292 F.3d at 796 (citation modified); Douglas , 482 F.3d at 601. Here, Green's defense asked that no such instruction be given at trial. App. 1182-83. In such cases, we 'only look to that prejudice which would have accrued despite the giving of a proper limiting instruction.” Moore , 732 F.2d at 990. But the question at this point is not whether Exhibit 101 in fact resulted in unfair prejudice; it is whether the district court correctly conducted the Rule 403 balancing test at the time the issue was raised. From that perspective, the district court was well within its discretion to consider the effectiveness of a limiting instruction to mitigate any risk of unfairness and the fact that Green later made a tactical decision not to request one does not affect our review. The district court therefore did not abuse its discretion in finding Exhibit 101 passed Rule 403.

In sum, the district court properly concluded that Exhibit 101 was relevant, was offered for a permissible purpose and passed Rule 403 balancing.

As to Exhibit 110, the district court found the text message relevant to Green’s intent to distribute drugs (of a different sort from those charged) because its language suggested distribution, rather than personal use, and it had been sent shortly before the search warrant was executed. App. 10203.¹² It also determined, under Rule 403, that any risk of unfair prejudice would be mitigated by a limiting instruction to the Page 28jury. Id. Green argues, in effect, that evidence of his possible distribution of another drug (marijuana) was irrelevant to his alleged intent to distribute the drugs at issue here. See Appellant Br. 61-64. With no permissible use, he suggests, its only remaining use would be as prohibited propensity evidence. Id. He also asserts that the evidence should have been excluded under Rule 403. Id. at 61.

Relevance. Exhibit 110 was likely at least marginally relevant. As discussed, evidence of past drug sales of one drug is usually relevant to show knowledge or intent to sell the same drug. See, e.g. , Douglas , 482 F.3d at 600; United States v. Washington , 969 F.2d 1073, 1081 (D.C. Cir. 1992), cert. denied , 113 S. Ct. 1287 (1993); United States v. Clarke , 24 F.3d 257 (D.C. Cir. 1994). As our colleague discusses in more detail, Concurring Op. 1-9, whether past sales of one drug are still relevant to show knowledge or intent to sell a different drug can be a closer question. In United States v. Mitchell , we noted that 'we have frequently upheld the admission of evidence regarding other drug transactions as relevant to intent in a charged drug transaction.” 49 F.3d 769, 775 (D.C. Cir. 1995). There, the court upheld the admission of evidence of an uncharged methamphetamine deal as relevant to two coconspirators’ efforts to obtain cocaine and cocaine base. Id. at 775-76. Green seeks to distinguish Mitchell on the ground that the defendants there were part of an ongoing conspiracy and both were involved in the previous methamphetamine deal. See Appellant Br. 61. But it is unclear why either of those distinctions is material. The essential question there and here is the same: is a previous sale of Drug A relevant to show a later intent to sell Drug B ? 13 Green also cites our court's decision in United States v. Watson , 171 F.3d 695 (D.C. Cir. 1999). In dicta, the court suggested that the district court on remand might wish to reconsider its admission of the Page 29defendant’s seven-year-old conviction for selling cocaine (rather than the cocaine base at issue in the case), as its probative value under Rule 403 might not reach the contested issues. Id. at 702-03. In dissent, Judge Garland remarked that neither the seven-year interval nor the differences between the drugs rendered the conviction inadmissible and the district court had committed no error. Id. at 708 & n.8 (Garland, J., dissenting). In any event, that disagreement does not bind us here. Moreover, our court has previously upheld the admission of evidence involving different drugs on different occasions if it has some relevance other than the forbidden character inference. See United States v. Gaviria , 116 F.3d 1498, 153233 (D.C. Cir. 1997) (evidence of uncharged heroin transactions relevant to conspiracy to distribute cocaine); see also Moore , 732 F.2d at 987-88 (evidence of uncharged transactions involving multiple drugs relevant to the defendants’ intent to sell cocaine).

Multiple circuit courts have also squarely considered whether past distribution of one drug is relevant to show knowledge or intent to distribute another and have generally determined that it is. See, e.g. , United States v. McLean , 581 F. App’x 228, 234-35 (4th Cir. 2014) (heroin in a cocaine case); United States v. Carpenter , 30 F. App'x 654 (8th Cir. 2002) (cocaine in a methamphetamine case); United States v. Hernandez , 896 F.2d 513, 522 (11th Cir. 1990) (marijuana in a cocaine case); United States v. Broussard , 80 F.3d 1025 (5th Cir. 1996) (same); United States v. Rubio-Villareal , 927 F.2d 1495, 1503 n.9 (9th Cir. 1991) (“[W]here evidence is offered to show knowledge and intent, it is not necessary that the illegal drug involved in the prior offense be identical to the illegal drug involved in the charged crimes.”).

Ultimately, the bar for relevance is low and the district court reasonably found that evidence of Green’s marijuana sale just three weeks before his arrest was relevant to his intent here. As the court noted, the fact that the drugs were different 'diminished' its probative value but did not eliminate it.

Page 30App. 103. The mental states required for both are the same, thus “evidence that [Green] previously had the state of mind— the knowledge and intent-to distribute illegal drugs is probative and thus relevant to whether he had the knowledge and intent to commit the crime charged here.' McLean , 581 F. App’x at 235. The proximity of the time the message was sent further supports the same conclusion. Cf. id. (a six-year interval did not render a past conviction irrelevant). Thus, the district court did not abuse its discretion in concluding the proffered text message was relevant.

Propensity. Exhibit 110 was used for a permissible purpose-showing Green's knowledge and intent to sell controlled substances-not to show any criminal propensity proscribed by Rule 404. Indeed, the government specifically stated in its closing argument that the text message was “good proof of [Green’s] possession with the intent to distribute the drugs that were seized in this case,' not that Green had a propensity to deal drugs or was a “drug dealer.” App. 1237. Because the evidence was offered for a permissible purpose, it was admissible under Rule 404(b).

Risk of Unfair Prejudice. Exhibit 110's probative value was not substantially outweighed by the risk of unfair prejudice under Rule 403. Granted, the probative value of Exhibit 110 was limited. The text message involved a different drug with no clear link between the message and 917 Wahler and the message was unmoored from any physical evidence. But as the probative value was low, so too was the risk of unfair prejudice. The message was neither inflammatory nor likely to mislead the jury and its weight could be readily challenged by the defense. As before, the district court also properly considered using a limiting instruction, even if none was ultimately given due to defense counsel’s tactical judgment. See Moore , 732 F.2d at 990. On balance, although Exhibit 110's probative value was not particularly strong, neither was its potential for undue prejudice and the district court was justified in concluding that it met the Rule 403 test.

Page 31In sum, the district court correctly determined that Exhibit 110 bore some relevance to the charged offenses, was introduced for a permissible purpose and survived scrutiny under Rule 403.

Even if we concluded that Exhibits 101 or 110 were admitted in error, however, that alone would not compel reversal. See Fed. R. Crim. P. 52(a) ('Any error, defect, irregularity, or variance that does not affect substantial rights must be disregarded.”). “For nonconstitutional errors like the one[s] [Green] asserts here, ‘an error is harmless’ and thus does not compel reversal 'if it did not have a substantial and injurious effect or influence in determining the jury’s verdict.’” United States v. Milligan , 77 F.4th 1008, 1012 (D.C. Cir. 2023) (quoting United States v. Powell , 334 F.3d 42, 45 (D.C. Cir. 2003)). Because Green timely objected to the introduction of both exhibits, the “burden of showing the absence of prejudice” is on the government. United States v. Olano , 507 U.S. 725, 741 (1993). Only if we are 'in grave doubt about the harmlessness of the error' must the conviction be reversed. United States v. Smart , 98 F.3d 1379, 1390 (D.C. Cir. 1996) (citation modified).

“The introduction of other crimes evidence to illuminate intent carries an inherent risk of . . . prejudice' because the permissible inference (intent) is “very close” to the impermissible one (propensity). Mitchell , 49 F.3d at 777. The “most significant factor” that can negate an error’s impact on the verdict is “the weight and nature of the evidence against the defendant.” Milligan , 77 F.4th at 1012 (quoting McGill , 815 F.3d at 886). Limiting instructions that 'guard the space between the permissible and impermissible inferences' can also reduce the impact of any erroneously admitted evidence. Mitchell , 49 F.3d at 777. Conversely, “dramatic [or] compelling” evidence that might “rivet the jury’s attention on [the defendant’s] bad character” could increase the chance of a Page 32harmful error. United States v. Brown , 597 F.3d 399, 405 (D.C. Cir. 2010); see also United States v. Sheffield , 832 F.3d 296, 309 (D.C. Cir. 2016).

Here, to the extent either exhibit was incorrectly admitted, any error was harmless. Because Green constructively possessed the contraband inside 917 Wahler, there was substantial evidence of his involvement in drug distribution even without the photograph or text message, including numerous digital scales, sandwich baggies, cutting agents and significant amounts of cash. The jury also heard testimony from the government's witnesses about Green's possessions (including identification cards) in the house, his occupancy there and the typical operations of stash houses, all of which further minimize the impact of both exhibits. Moreover, the quantity of the drugs found and the fact that they were packaged for street distribution further supports the reasonable inference that Green had an intent to distribute them. See United States v. Williams , 233 F.3d 592, 595 (D.C. Cir. 2000) (a reasonable jury may infer from 'the quantity of drugs possessed” and “the fact that [the] drugs were segregated into ‘baggies’” that a defendant had an intent to distribute). Neither the photograph nor the text message was dramatic or compelling such that the jury's attention would have been riveted on it as evidence of Green’s character. Instead, they were small pieces of the evidence connecting him to at least some of the drugs in 917 Wahler. Although Green argues that the exhibits were the only direct evidence of drug distribution, see Appellant Br. 63-64, the volume of the indirect evidence is sufficient to render any misstep in admitting them harmless.

The government further suggests that, for the purpose of the harmless-error analysis, we should assume that the district court gave a limiting instruction in light of Green's tactical decision to decline one. See Gov't Br. 61. Green responds that he was “not required to reiterate this damaging evidence to the jury in the form of an instruction to preserve his claim that it should never have been introduced in the first place.” Reply Page 33Br. 31. Green’s argument is misplaced—the relevant issue is not preservation (which is uncontested), but whether the purportedly inadmissible evidence resulted in prejudicial error, a risk that could have been mitigated by an instruction.

Although the caselaw on declined limiting instructions is sparse, several circuit courts have held that declining such an instruction at least weakens a later claim of undue prejudice. See, e.g. , United States v. Wheeler , 540 F.3d 683, 693 (7th Cir. 2008) (“Because [defendant] waived the opportunity to alleviate the risk of unfair prejudice, we decline to reverse the district court’s evidentiary ruling on the grounds that the [Rule 404(b)] evidence was unfairly prejudicial”); United States v. Monks , 774 F.2d 945, 955 (9th Cir. 1985) (“[W]e will uphold a district court’s refusal to grant a mistrial where the prejudice resulting from introduction of the improper character evidence is minimal, and the judge’s offer to give a limiting instruction is rejected.” (citation modified)); United States v. Tejeda , 974 F.2d 210, 215 (1st Cir. 1992) (harmless Rule 404(b) error after defendant declined limiting instruction); cf. Moore , 732 F.2d at 990. Relatedly, a tactical decision by defense counsel not to seek a limiting instruction-and a district court's resultant failure to give one—does not preclude finding harmless error. See United States v. Williams , 212 F.3d 1305, 1311 (D.C. Cir. 2000) (concluding that an officer’s prejudicial statement was harmless error given the strong evidence of defendant’s guilt and the fact that defense counsel did not seek a limiting instruction to address the comment). Taken together, these cases reflect a common-sense principle that a defendant who declines a limiting instruction bears at least some responsibility for the risk of prejudice the instruction might have averted.

Accordingly, although we uphold the admission of Exhibits 101 and 110, in the alternative, any error was also harmless and their admission does not warrant reversal.

Page 34For the foregoing reasons, Green’s convictions are affirmed.

So ordered.

Page 35GARCIA, Circuit Judge , concurring in part and concurring in the judgment:

I join the court’s opinion except for Part III.C.1.b. There, the majority affirms the district court’s ruling that Federal Rule of Evidence 404(b) did not bar Exhibit 110’s admission despite powerful arguments that the exhibit amounted to impermissible character evidence. That holding is unnecessary to resolve this case, as we all agree that any error in admitting Exhibit 110 was harmless. See United States v. Burnett , 827 F.3d 1108, 1118– 19 (D.C. Cir. 2016) (deeming any error harmless and declining to address 404(b) issue). By nonetheless reaching the question of Exhibit 110’s admissibility, the majority needlessly wades into a complex area of the law that has deeply divided the circuits. And it extends our precedent in ways that contradict Rule 404(b)'s text and core purpose. I write separately to identify the difficult issues our cases and now the majority here have breezed past, and to explain why I do not join the court’s holding on this issue.

Rule 404(b)(1) prohibits the introduction of propensity evidence: “Evidence of any other crime, wrong, or act is not admissible to prove a person’s character in order to show that on a particular occasion the person acted in accordance with the character.' Rule 404(b)(2) then states that evidence of other bad acts “may” nonetheless “be admissible for another purpose, such as proving motive, opportunity, intent, preparation, plan, knowledge, identity, absence of mistake, or lack of accident.”

In criminal cases, Rule 404(b) serves the important purpose of ensuring that a defendant is tried for his alleged conduct, and not based on a character-related assumption “that he is by propensity a probable perpetrator of the crime.' Michelson v. United States , 335 U.S. 469, 475 (1948). Put another way, the Rule prohibits evidence suggesting 'that because the defendant committed another bad act, he is more likely to have committed the charged act.' United States v.

Page 36Mitchell , 49 F.3d 769, 774 (D.C. Cir. 1995); see also United States v. Caldwell , 760 F.3d 267, 276 (3d Cir. 2014) (explaining that the Rule ensures “an accused is tried for what he did, not who he is”).

Reconciling that principle with the “Permitted Uses” listed in Rule 404(b)(2) can sometimes prove difficult. “Intent,” for example, is a permitted use. So, focusing myopically on Rule 404(b)(2), one might say that the Rule poses no bar to admission whenever the government ostensibly introduces a prior bad act as evidence of a defendant’s “intent” rather than his “character.” But what if the only way the act is probative of a defendant’s “intent” is through an inference that he has a propensity to commit bad acts, the very 'character'-based reasoning that Rule 404(b)(1) prohibits? Is that truly sufficient to evade Rule 404(b)(1)?

This case illustrates the dilemma. Demetrius Green was tried for possessing with intent to distribute oxycodone, hydromorphone, and cocaine base. To help establish his intent to distribute, the government introduced Exhibit 110, a text message in which Green purportedly offered to sell someone marijuana (it said: “I got sum tree come support my hustle,” App. 236). As the majority correctly notes, the text message does not suggest anything except that Green had once offered to distribute drugs: 'The text message involved a different drug with no clear link between the message and 917 Wahler [Place,] and the message was unmoored from any physical evidence.” Maj. Op. 30. The text message did not show, for example, that Green had previously attempted to sell drugs in a manner similar to how he allegedly intended to distribute the drugs at issue here. Nor did the text message insinuate that he had used similar paraphernalia to sell drugs, or that he had sold drugs in the same location or to the same person.

Page 37Logically, the only way a bare allegation of Green’s intent to sell marijuana on one occasion could be thought relevant to show he intended to distribute other drugs on this occasion is by reasoning that Green is the type of person who intends to distribute the drugs he possesses—in other words, that he is a drug dealer. This means the conclusion the government asked the district court to draw (that Exhibit 110 was relevant to Green’s intent to distribute) rested exclusively on an intermediate inference sounding in propensity reasoning (that because Green had acted like a drug dealer once, he was more likely to have acted that way again). Cf. United States v. Matthews , 431 F.3d 1296, 1313 n.1 (11th Cir. 2005) (Tjoflat, J., concurring in the judgment) (“It is difficult to argue that a person had an intention to do something on a particular occasion because he or she demonstrated that intention previously without implicitly suggesting that the person has a proclivity towards that intent.”).

The government’s view is not easy to square with Rule 404(b). There is a strong intuition that it must be improper for the government to introduce evidence whose relevance to the defendant’s intent rests entirely on a propensity inference: here, that the defendant has a propensity for dealing drugs. The text of the Rule suggests as much. Rule 404(b)(1) prohibits the use of prior bad acts “to prove a person’s character” and to show that the person “acted in accordance with the character.” If the government’s only theory of relevance relies on the inference that a defendant “acted in accordance with” how he has acted before, the evidence is seemingly inadmissible, full stop.

Rule 404(b)(2) does not say otherwise. The “Permitted Uses” it lists do not constitute exceptions to Rule 404(b)(1), but rather examples of ways to admit other-act evidence 'for another purpose'-that is, a purpose other than the one forbidden by Rule 404(b)(1). A natural reading of Rule 404(b), and one that fits its commonly accepted purpose, is therefore Page 38that the proponent of the other-act evidence must establish its relevance to a permitted use without relying purely on character-based inferences.

Indeed, at trial, the government seemed to share that intuition about the Rule. As the court recounts, the government was careful to assert that it did not introduce Exhibit 110 to show “that Green had a propensity to deal drugs or was a ‘drug dealer.’” Maj. Op. 30. The government said it introduced the text message for the supposedly different purpose of showing Green’s “intent to distribute the drugs that were seized in this case.” Id. (quoting App. 1237). I fail to see the distinction. The government offered no explanation for how the text message could show that Green intended to distribute the narcotics at issue except insofar as it suggested that he had a propensity for drug dealing.

To frame the concern in more practical terms, imagine that a juror received a limiting instruction telling her not to use the text as evidence of Green’s “character,” but that she could use it as evidence of his “intent.” It strikes me that any reasonable person would have no earthly idea how to proceed. The juror might gather that she may not look to Green’s one-time offer to sell drugs as a reflection of his “character” and reason that he is the type of person who more likely intended to distribute drugs this time around. But if that is so, how else could she conclude the evidence is relevant to Green’s intent?

Commentators have repeatedly criticized the government's question-begging logic. As one puts it, 'it is hard to see how this use avoids the propensity inference from character; namely, that a person with a history of selling drugs has a propensity to sell drugs and that is therefore what the defendant intended to do with the drugs in this case.” Paul S. Milich, The Degrading Character Rule in American Criminal Trials , 47 Ga. L. Rev. 775, 786–87 (2013). Another explains:

Page 39'[C]ourts too often fail to demonstrate how the mental leap from possessing a particular state of mind on one occasion to possessing the same state of mind on a different occasion does not involve the use of a character-propensity inference.' David P. Leonard, The New Wigmore: Evidence of Other Misconduct and Similar Events § 7.5.2 (2d ed. Supp. 2025). Similar criticisms from respected authorities abound. See, e.g. , Julius Stone, The Rule of Exclusion of Similar Fact Evidence: America , 51 Harv. L. Rev. 988, 1033 (1938) (describing permissive admission of other-act evidence as “utter perversions of the object of the original rule'); 1 Edward J. Imwinkelried, Uncharged Misconduct Evidence § 5:2 (2025 ed.) (explaining that evidence should not be admissible to prove intent under Rule 404(b) if “the prosecution must rely on an intermediate bad character inference”); Daniel J. Capra & Liesa L. Richter, Character Assassination: Amending Federal Rule of Evidence 404(b) to Protect Criminal Defendants , 118 Colum. L. Rev. 769, 789–95 (2018).

As these scholars lament, some courts have (mostly without acknowledging these concerns) admitted other-act evidence when it is relevant in any way to a defendant’s intent. See Maj. Op. 29 (collecting cases). But several othersincluding at least the Third, Fourth, and Seventh Circuits, and one panel for the Sixth Circuit—have articulated limits on such use of other-act evidence. Those courts guard against the danger of 'intent . . . blend[ing] with improper propensity uses' by 'not just ask[ing] whether the proposed other-act evidence is relevant to a non-propensity purpose but [also asking] how exactly the evidence is relevant to that purpose . . . without relying on a propensity inference.” United States v. Gomez , 763 F.3d 845, 856 (7th Cir. 2014) (en banc); United States v. Miller , 673 F.3d 688, 698 (7th Cir. 2012); 1 Robert P. Mosteller et al., McCormick on Evidence § 190.11 (9th ed. Feb. 2025 update) (endorsing Seventh Circuit’s approach); see also Caldwell , 760 F.3d at 277 (requiring “a chain of inferences that Page 40does not contain a propensity link”); United States v. Hall , 858 F.3d 254, 277 (4th Cir. 2017) (demanding 'propensity-free chains of inferences'). Applying that requirement, some of these courts exclude evidence of other drug activity unless the government can show a meaningful, non-propensity 'linkage . . . in time, manner, place, or pattern of conduct' between the charged and uncharged acts. Hall , 858 F.3d at 261 (citation modified); see id. at 272-75; see also Miller , 673 F.3d at 700; United States v. Bell , 516 F.3d 432, 443 (6th Cir. 2008) (permitting evidence of other distribution activity only if it was 'part of the same scheme or involved a similar modus operandi '). But see United States v. Hardy , 643 F.3d 143, 151 (6th Cir. 2011) (casting doubt on Bell but requiring that the other act be “substantially similar and reasonably near in time” to charged conduct (quotation omitted)).

Courts of appeals, in short, “sharply disagree” over how to reconcile Rule 404(b)’s prohibition against propensity reasoning with its list of permitted purposes. Capra & Richter, supra , at 794. Some of the basic questions underlying that disagreement include: Is other-act evidence admissible when its relevance to a permitted purpose rests entirely on propensity inferences? If not, what counts as a permissible non-propensity inference? These questions are no doubt difficult to resolve in a way that facilitates practical application. To my eye, however, our cases to date have not grappled with them in the slightest. Instead, our cases fall into the group that seems to proceed as if the only question when a Rule 404(b) objection is raised is whether the evidence is relevant in any way to a permitted use. See Maj. Op. 28–30.

Still, our precedent by no means dictates the majority's conclusion here. For one thing, the majority uncritically endorses the government’s assertion that it is not “propensity” reasoning to use Green’s alleged offer to distribute marijuana as evidence that he subsequently intended to distribute Page 41oxycodone, hydromorphone, and cocaine base. Id. at 30. I have already explained why that logic, without more, is indistinguishable from “the very kind of reasoning— i.e. , once a drug dealer, always a drug dealer—which 404(b) excludes.” Bell , 516 F.3d at 444 (citing Old Chief v. United States , 519 U.S. 172, 180-81 (1997)); see also Miller , 673 F.3d at 700.

For another, our cases have not blessed the introduction of evidence whose relevance is so purely based on propensity inferences as the text message here. To illustrate the point, consider the in-circuit cases cited by the majority. In most of them, the other-act evidence featured the defendant distributing the same type of drug as the one charged in the indictment. See Maj. Op. 26–28. And in each, the other-act evidence shared some meaningful similarity with the charged conduct that made it probative of something beyond a generic proclivity to distribute drugs. See id. For example, in United States v. Pettiford , 517 F.3d 584 (D.C. Cir. 2008), the defendant was indicted after officers recovered cocaine from his car’s center console; the other-act evidence was that officers had recovered cocaine from another car’s center console after observing the defendant reach inside the car and then conduct a cocaine sale. See id. at 586-87. In United States v. Crowder , 141 F.3d 1202 (D.C. Cir. 1998) (en banc), the other-act evidence involved the defendant selling cocaine “on the same block” where he was alleged to have possessed heroin and cocaine with intent to distribute. Id. at 1203-04. And in United States v. Douglas , 482 F.3d 591 (D.C. Cir. 2007), the other-act evidence 'involved sale of the same substance in almost the same neighborhood.' Id. at 601; see also id. at 599 (noting that charged drug sales occurred around the time and place of prior sales introduced under Rule 404(b)); United States v. Washington , 969 F.2d 1073, 1081 (D.C. Cir. 1992) (emphasizing “similarity of the transactions” where other drug sale occurred around the same time and place as charged offense, and involved same paraphernalia); United States v.

Page 42Clarke , 24 F.3d 257, 263-66 (D.C. Cir. 1994) (admitting evidence of other drug sales at same drugstore parking lot and to same middleman).

So too in Mitchell , the principal case the majority highlights from our court in which the prior act involved a different type of drug from the one charged in the indictment. The similarities between the charged act and the uncharged drug-distribution activity there were also notable. For one of the two defendants, the charged and uncharged conduct occurred contemporaneously, and each incident involved him allegedly providing transportation services for his coconspirator. 49 F.3d at 775–76. For the other, the uncharged conduct showed him orchestrating a complex drug transaction in ways that mirrored the charged offense. Id. at 772-75; see also United States v. Gaviria , 116 F.3d 1498, 1532–33 (D.C. Cir. 1997) (describing contemporaneous drug sales arranged by same co-conspirators using code words just as in the charged incident); United States v. Moore , 732 F.2d 983, 988 (D.C. Cir. 1984) (describing the other acts evidence as showing “a pattern of drug possession and dealing taking place immediately before the conspiracy alleged in the indictment”).

In each of these cases, the other-act evidence helped demonstrate the defendant’s intent to distribute narcotics in a particular place, or in a particular manner. Each piece of evidence would, at least arguably, be admissible under the more careful approach taken by some of our sister circuits.

No such similarity is present here. The text message is an unadorned statement of intent to sell one drug (marijuana) on one occasion, and its sole purpose at trial was to invite the inference that Green was more likely to distribute different drugs (oxycodone, hydromorphone, and cocaine base) on a different occasion. Cf. Maj. Op. 28 (framing the question as “is a previous sale of Drug A relevant to show a later intent to Page 43sell Drug B?'). We have never affirmed the admission of evidence under Rule 404(b) that sounds so loudly in propensity reasoning. *

To be sure, precisely because our cases thus far have not grappled with the difficult questions this case illustrates, the majority’s decision today may well be a defensible application of our precedent. But that does not mean it is a sound one. The majority reaches its holding by reading our cases broadly and extending their logic to new terrain, all while giving short thrift to the countervailing interests underlying Rule 404(b). The result is to place our precedent in greater tension with the text and basic purpose of the Rule, the weight of informed scholarship, and the decisions of other courts of appeals. Particularly because doing so was unnecessary to resolve this case, I do not join that analysis.