Second Circuit Criminal Law Blog

Second Circuit Reaches for Affirmance in Child Pornography Case

In United States v. Clarke, the Second Circuit ventured once more into the thicket of internet crime, tangling with statutory interpretation and discovery issues complicated by their technological setting.  The panel (Walker, Leval, Carney) gave an expansive read to a child pornography statute, ruling that the defendant “transported” child pornography through an online peer-to-peer network, even though the government moved the files from the defendant’s computer, and the defendant was not aware of the file transfer.  In addition, the panel refused the defendant’s discovery request for inspection of law enforcement software based on the facts presented here, and declined to announce a standard for future cases involving similar law enforcement programs.  The panel also rejected a bevy of challenges relating to the sufficiency of the evidence and the reasonableness of the sentence, to ultimately affirm the judgment of the district court.   

The decision was released nearly a year after post-argument briefing was completed—a timeframe that is not unheard of, but that may suggest closer than usual attention to the case.  Tensions are evident in the decision’s reasoning, with the panel avoiding literal and textualist interpretations, and at times explicitly acknowledging the limits of its rationale, opening the door to the future development of the law.  The decision underscores the recurring difficulty of mapping statutes and discovery rules onto online conduct, and further raises concerns about courts’ willingness to broadly construe those statutes and rules in order to reach serious and disturbing conduct involving child pornography.


In 2015, federal agents initiated an investigation into child pornography crimes involving a BitTorrent peer-to-peer file sharing network.  To assist, agents used law enforcement software called “Torrential Downpour,” which identified downloads of specific files known to contain child pornography.  Using this software, agents ascertained that hundreds of child pornography files were downloaded through an IP address traced to Clarke’s residence.  In addition, agents were able to download two videos of child pornography from Clarke’s computer. 

Agents then obtained a search warrant for Clarke’s house, and found thousands of child pornography files on Clarke’s computer.  Agents also spoke to Clarke, who waived his Miranda rights.  When asked about the BitTorrent program’s operations, Clarke gave somewhat equivocal answers, but stated that he “knew how” “peer-to-peer programs work,” and answered in the affirmative when asked if he knew that he “was essentially sharing files of child pornography.”

In discovery, Clarke requested production of the Torrential Downpour software or its source code, arguing that it was necessary to determine whether the software had been used to access non-public information on his computer.  The government refused, citing the need to keep the technology private, lest other distributors learn how to evade detection.  The district court rejected Clarke’s motion for production.

The case ultimately proceeded to trial.  Clarke was convicted on all counts, and subsequently sentenced to 120 months.

Transportation Charges

Clarke was charged with and convicted of, inter alia, two counts of “knowingly transport[ing]” child pornography, in violation of 18 U.S.C. § 2252(a)(1).  The two charges were based on the two instances when agents were able to download child pornography from Clarke’s computer.  On appeal, Clarke raised several challenges to these charges.  The panel quickly rejected arguments relating to the intent standard and the sufficiency of intent-related evidence, before proceeding to the closest question: what does it mean to “transport” child pornography on the internet? 

In post-argument briefing requested by the panel, Clarke asserted that he had not “transport[ed]” any materials within the meaning of the statute.  Clarke himself did not transfer the files; instead, government agents accessed his files and downloaded them through the peer-to-peer network.  Clarke pointed out, and the panel agreed, that there was “neither evidence (nor allegation) that, at the time of the downloading by the government agents, Clarke took any action or was aware of the government’s downloading.” 

Nonetheless, the panel found that Clarke, by using the peer-to-peer network, “enabled others to download those files to their computers,” and “implicitly invited other participants in the file sharing network to share his files.”  Then, “as anticipated, another user of the file-sharing network caused the files to be downloaded and sent from his computer to the other user’s computer.”  The panel viewed this as an “almost inevitable consequence” of using BitTorrent, and thus the government agents “completed Clarke’s commission of the transportation offenses” by moving the files. 

This reasoning leaves some room to doubt Justice Kagan’s conclusion that “we’re all textualists now.”[1]  The panel paid little heed to the definition of the word “transport”—“to move something from one place to another”—which it referenced only in a footnote.  The definition would seem to favor Clarke, since he did not “move” the files “from one place to another”: the government did.  The panel, however, did not grapple with that issue.  Also notable was the panel’s extensive analogizing to cases where similar conduct was found to constitute “distribution” of child pornography.  As the panel itself observed, “distribution precedents are not squarely pertinent to the transportation statutes,” since “distribution can be accomplished by mere transfer of ownership without movement from one place to another.”  The panel nonetheless felt that the “logic of these cases” was “consistent” with the conclusion that Clarke’s enabling the downloads should suffice to show he “transported” them.  Thus, the textual distinction between “distribution” and “transportation” appeared to carry comparatively little weight. 

Software Nondisclosure

Clarke also argued that the government should have been required to produce the Torrential Downpour software, or its source code, pursuant to Fed. R. Crim. P. 16(a)(1)(E)(i).  That rule provides that the government “must permit the defendant to inspect and to copy” any “documents” or “data” within the government’s control that are “material to preparing the defense.”

The panel noted that, on one hand, information about the software was material to Clarke’s defense.  On the other, the government disclosed certain information about the program’s operation, short of turning over the software or source code.  And the government held a “substantial” interest in preventing child pornographers from learning how to avoid detection.  Nonetheless, the panel declined to resolve these conflicting interests, holding instead that Clarke had not demonstrated prejudice from the nondisclosure. 

The panel explained that Clarke had sought production of the software to prove his expert’s theory that his external hard drive was not publicly accessible, and that the agents could not have downloaded Clarke’s child pornography files over the open BitTorrent network.  But the government presented evidence that, once a user specifies a location to store downloaded files, those files will be accessible to others—regardless of whether they are located on an external hard drive.  The government also provided evidence that, in a separate instance, agents saved files to an external hard drive, and those files were accessed by other BitTorrent users.  The panel viewed this evidence as sufficient to reduce Clarke’s expert’s theory to mere “speculation,” and concluded there was no prejudice from the nondisclosure.

By leaving the competing interests among the prosecution and defense unresolved, and instead ruling only that no prejudice was shown, the decision declined to offer a standard for district courts to consider when presented with future challenges.  The panel also made no mention of whether the law enforcement privilege protected materials like Torrential Downpour (as the government had argued in its brief).  The decision appeared to readily find that the government’s proof refuted Clarke’s expert—but did not explain how or to what degree a defendant would need to show that the government’s factual assertions are incorrect or otherwise vulnerable before more specific discovery would be warranted (or prejudice after the fact would be found).  This is all the more interesting given that the panel seemed to accept that the information sought was indeed material, and Rule 16 announces a defendant-friendly standard, by requiring that the government “must” disclose “material” evidence.  Again, however, the panel did not linger on these textual requirements. 


The panel also rejected Clarke’s procedural and substantive reasonableness challenges to his sentence.  The panel noted that Clarke’s 120-month sentence was 90 months below the bottom end of the Guidelines range.  In addition, the panel emphasized that the district court had explicitly considered United States v. Dorvee, 616 F.3d 174 (2d Cir. 2010), and its requirement that courts take particular care in child pornography sentencing, where the Guidelines may lead to excessive punishment.  As in Dorvee, the Circuit has repeatedly expressed concern about the length of sentences for child pornography crimes, but this sentence represented a significant reduction from the Guidelines range. 

Concluding Thoughts

Child pornography cases often involve difficult legal issues surrounding appalling conduct, and produce unusual results.  The conduct causes harm to innocent children and is disturbing to consider.  In addition, some of the statutes were written for a time prior to the widespread transmission of child pornography on the internet—the “transportation” statute’s language dates to at least 1978, when the current statute was enacted.  Here, the panel may have been motivated to resist a strict, literal construction of the “transportation” statute and discovery rules, in order to make sure that the statute reached the alarming behavior in which the defendant engaged.  The panel did not appear to harbor much doubt that the defendant’s purpose in collecting the child pornography was to make it available to others over the internet.  It may be that Congress should rewrite the statute to clarify or update the language so that it is applicable to the harmful conduct that society needs to prevent.

The discovery request issue also is one to which courts may return.  Similar discovery requests have received divergent treatment among the various circuits.  Compare United States v. Budziak, 697 F.3d 1105, 1112 (9th Cir. 2012) (compelling disclosure), with United States v. Pirosko, 787 F.3d 358, 366 (6th Cir. 2015) (affirming nondisclosure).  In the context of internet crime, it seems likely that the government will wish to keep secret the details of some of its law enforcement tools and technologies.  This is an understandable impulse, but one that may be hard to reconcile with the rights of the accused.  Nor is it a new impulse: when the government uses an informant to make a case, it seeks to keep the informant’s role and identity out of the case, arguing that concerns of safety and future enforcement efforts are paramount.  Yet if the government cannot justify a traffic stop made to seize drugs from a car without disclosing the role of an informant, the government must choose between prosecuting the case and protecting the informant.  The government’s right to secrecy often must yield in criminal matters, and the adversary system depends on sufficient disclosure to permit the defendant to test the government’s proof.  Thus this decision is likely not to be the last word on this topic, and future decisions may bring greater clarity to the appropriate standards.

by Abigail E. Marion and Harry Sandick

[1] Justice Elena Kagan, The Scalia Lecture: A Dialogue with Justice Kagan on the Reading of Statutes at 8:20 (Nov. 17, 2015), available at