Criminal Action No. 21-399 (RDM)
Decided: February 29, 2024
Jeffrey Pearlman, Catherine Pelker, Assistant U.S. Attorneys, U.S. Department of Justice, Washington, DC, Christopher Brodie Brown, Assistant U.S. Attorney, DOJ-USAO, Washington, DC, for United States of America. Tor Ekeland, Pro Hac Vice, Tor Ekeland Law PLLC, Brooklyn, NY, Marina Medvin, Medvin Law PLC, Alexandria, VA, Michael Hassard, Pro Hac Vice, Tor Ekeland Law PLLC, New York, NY, for Defendant.
MEMORANDUM OPINION AND ORDER
Defendant Roman Sterlingov is charged with money laundering conspiracy, money laundering, operating an unlicensed money transmitting business, and money transmission without a license, all in relation to his alleged operation of a bitcoin mixer known as Bitcoin Fog. Dkt. 43 (Superseding Indictment). Both sides have proffered multiple expert witnesses. In June, July, and August 2023, the Court held a series of Daubert hearings at which it heard testimony from nearly all of these proposed experts and considered lengthy expert reports, at least one of which was accompanied by a series of large datafiles. See, e.g., Dkt. 224 (June 23, 2023 Hrg. Tr.); Dkt. 228 (Aug. 22, 2023 Hrg. Tr.); Dkt. 229 (Aug. 23, 2023 Hrg. Tr.). In September, the Court heard argument on the admissibility of the proposed experts’ testimony. See Dkt. 232 (Sept. 7, 2023 Hrg. Tr.); Dkt. 233 (Sept. 8, 2023 Hrg. Tr.); Dkt. 235 (Sept. 15, 2023 Hrg. Tr.); Dkt. 236 (Sept. 18, 2023 Hrg. Tr.). The parties subsequently submitted supplemental briefing concerning a subset of the expert testimony issues, as well as further supporting evidence. Dkt. 191; Dkt. 192; Dkt. 193. The Court has issued multiple rulings from the bench regarding the admissibility of the proposed expert testimony. This opinion provides additional explanation regarding the Court's rejection of defendant's Daubert challenge to the reliance by two of the government's experts, Luke Scholl of the Federal Bureau of Investigations (“FBI”) and Elizabeth Bisbee of Chainalysis Government Solutions (“Chainalysis”), on a software product known as Chainalysis Reactor (“Reactor”).
Although bitcoin transactions are anonymous in the sense that each transaction is identified only by lengthy sets of numbers and letters representing the sending address(es), the receiving address(es), and the transaction ID(s), they are, at the same time, public in the sense that the amount, timing, sending address(es), and receiving address(es) of every transaction is recorded on the blockchain, which is a decentralized, immutable, public ledger available to anyone with an interest in looking. As result, bitcoin transactions are both uniquely anonymous and uniquely public. As explained further below, the public ledger permits law enforcement and others not only to trace bitcoin moving through specific transactions, but to cluster bitcoin addresses in a manner that provides a window into otherwise anonymous activity. The most widely accepted means of clustering relies on a concept referred to as “co-spend,” which occurs when the user on the sending side of the transaction draws on bitcoin held in multiple addresses. It is possible to associate those multiple sending addresses with a single sender, since the sender would need the “private key,” akin to a password, for each of the sending addresses to effectuate the transfer. When the process of identifying co-spend transactions is repeated for multiple transactions, it is possible to build a larger and larger cluster associated with the user or entity in question.
Given the volume of transactions recorded on the blockchain, investigators frequently make use of proprietary software like Chainalysis Reactor to cluster bitcoin transactions using the co-spend and other heuristics. Much of this work could be done manually given enough time, and as explained below, it is possible to corroborate (or to challenge) the results generated by the software for particular clusters with the public blockchain data, a pad of paper, a pencil, and hours of work.1 Reactor also uses other heuristics based on unique identifiers that Chainalysis has associated with particular services that have in the past or that currently transact on the blockchain.
The defense argues that Scholl and Bisbee's reliance on Reactor fails the Daubert test and that the Court should, accordingly, exclude all testimony and evidence based on clustering performed using that software. The defense contends that Reactor is “junk science,” which has not been peer reviewed and has no known error rate, and that, as a result, any testimony based on Reactor is not “the product of reliable principles and methods,” Fed. R. Evid. 702(c). For the reasons explained below, the Court is unpersuaded. Although the defense is correct that not all of the heuristics used in this case have been subject to “peer review” and that Chainalysis does not gather and record an error rate in a central location, substantial evidence supports the government's submission that the software is highly reliable—and, if anything, conservative—in clustering (and then attributing) bitcoin addresses. The defense, of course, remains free to challenge the accuracy and reliability of Reactor before the jury. But the Court is satisfied that it is “more likely than not” that the evidence and testimony at issue will help the jury to understand the evidence, that it is based on sufficient facts or data and reliable principles and methods, and that Scholl and Bisbee have reliably applied those principles and methods. See Fed. R. Evid. 702.
