Please E-mail suggested additions, comments and/or corrections to Kent@MoreLaw.Com.

Help support the publication of case reports on MoreLaw

Date: 01-03-2016

Case Style: People Of New York v. Aleynikov

Case Number: 04447/12

Judge: Daniel P. Conviser

Court: Supreme Court, New York County

Plaintiff's Attorney: Cyrus R. Vance Jr., Daniel Holmes, Jeremy Glickman, Elizabeth Roper

Defendant's Attorney: Kevin H. Marino, John Tortorella, John Boyle

Description: The Defendant was charged in this case with one count of Unlawful Use Of Secret Scientific Material in violation of Penal Law § 165.07 for conduct which occurred on June 1, 2009 (Count 1), one count alleging a violation of the same statute with respect to conduct which occurred on June 5, 2009 (Count 2) and one count of Unlawful Duplication of Computer Related in the First Degree in violation of Penal Law § 156.30 (1) for conduct which occurred on June 5, 2009 (Count 3). After a jury trial presided over by this Court, the jury was unable to reach a unanimous verdict with respect to Count 1, returned a guilty verdict under Count 2 and a not-guilty verdict under Count 3.

At the close of the People's case, the Defendant moved for a trial order of dismissal with respect to all three counts pursuant to CPL 290.10. He periodically renewed that motion during the jury's extensive deliberations. For the reasons outlined infra, the Defendant's motion for a trial order of dismissal with respect to the two counts for which that motion is outstanding (Counts 1 & 2) is granted. The Court holds that, viewing the evidence in a light most favorable to the People, the prosecution did not prove the Defendant made a "tangible reproduction or representation" of secret scientific material as required by the statute. The Court also holds,

Page 2

again under the same evidentiary standard, that the People did not demonstrate Aleynikov had the "intent to appropriate . . . the use of secret scientific material" as required by the law.

STATEMENT OF FACTS

The vast majority of the evidence during the trial was uncontested. Those general facts are first outlined here. Following that, a more detailed recitation of the specific facts relevant to the two issues in this motion are recounted with respect to key witnesses.1

General Uncontested Facts

In 2009, Sergey Aleynikov was employed as a computer programmer in the New York City office of Goldman Sachs ("Goldman"), a global financial institution. Goldman had approximately 32,000 employees in 2009. Mr. Aleynikov was one of the computer programmers responsible for writing computer source code for Goldman's high-frequency trading ("HFT") system. Computer source code is "[i]nstructions to the computer in a more human readable format so that you're instructing the computer on what you want done." (404)2 . "High-Frequency Trading" is "a trading style practiced by many trading companies whereby you use computers to make very rapid decisions that allow you to generate trades or generate orders and create prices, [at] which we [a firm, in this case Goldman] would be willing to buy and sell securities." (442). Goldman earned approximately $300 million from HFT trading during the 2009 calendar year.

Aleynikov resigned from Goldman at the end of April or early May to accept a job at a new high-frequency trading firm, Teza Technologies ("Teza"). His annual compensation from Goldman at the time was $400,000. He last worked the week ending June 5, 2009. His last date of employment at Goldman was around July 5, pursuant to a "garden period"3 where employees are paid after they cease work so they do not immediately begin work for a competitor.4 Mr. Aleynikov was not working on Goldman projects after June 5, however, was required to return Goldman property at that time and did not have permission to take or retain property or computer code from Goldman after June 5. Teza had agreed to pay Aleynikov an annual salary of $1.2 million to work as a computer programmer.

Aleynikov was permitted to access computer code from the Goldman computer code respository while he was a Goldman employee in order to work on the code. Goldman employees were also permitted to work on computer source code at home but only through a remote log-in to the Goldman computer system. On June 1 and June 5, 2009, Aleynikov uploaded approximately 32 megabytes of computer source code from the Goldman HFT system to a server in Germany.5 He then downloaded that source code from the German server to his home computer in New Jersey. When Aleynikov uploaded that computer source code he knowingly violated the Goldman confidentiality policy which he had agreed to as a condition of his employment. Goldman considered its computer source code to be intellectual property, proprietary and valuable and took significant steps to ensure that code was not transferred or used outside the firm. When he wrongfully uploaded the computer source code Aleynikov also deleted his "bash history" (the record of the commands he inputted into his computer) relevant to his unauthorized upload. This was done in a failed attempt to conceal his actions. Further, the data he uploaded was encrypted and some of the transfer dates for the scripts he uploaded were back dated.

Although Aleynikov duplicated and transferred the computer source code, Goldman continued to possess the code and never lost a single line of it. Aleynikov was arrested on July 3, 2009. He thus possessed the code for 33 days prior to his arrest. Aleynikov earned no income (outside his legitimate employment with Goldman) attributable to this conduct. After Aleynikov's unauthorized upload, Goldman continued to trade using the code and never lost any income by virtue of Aleynikov's actions. (Goldman expended resources to investigate Aleynikov's activities but the value of that work was not quantified). There was no evidence Aleynikov ever sold or attempted to sell the source code he transferred to anyone. He did, however, download a portion of the code to a server at Teza. Defendant's counsel has acknowledged that Aleynikov "made copies of the sections of Goldman Sachs's source code he thought could be helpful to him in building Teza's high frequency trading platform."6

Aleynikov's upload of the Goldman computer source code occurred after he accepted an employment offer from Teza. There was no evidence Teza was motivated to hire Aleynikov because of Aleynikov's unauthorized transfer of the code. Teza never earned any income from the source code Aleynikov obtained. It was Teza's policy that programmers who came to work for Teza were not permitted to bring proprietary source code from other firms they had worked for. Teza began high-frequency trading at the end of the third or beginning of the fourth quarter of 2010. On June 1 and June 5, 2009, Teza was in the process of developing its HFT system but was not engaged in high-frequency trading. At that time, Teza had not yet determined what kind of high-frequency trading it would conduct. After Aleynikov's arrest on July 2, 2009, his employment with Teza was terminated. There was no evidence Aleynikov ever duplicated the source code he downloaded to a piece of paper, any medium where it could be touched or any

Page 3

medium outside a computer or thumb drive.

Aleynikov was originally prosecuted and convicted in federal district court with respect to these actions for violating the National Stolen Property Act, 18 USC § 2314 (the "NSPA") and the Economic Espionage Act of 1996, 18 USC § 1832 (the "EEA"). He was sentenced to a 97 month prison term for these convictions (8 years plus one month). His convictions were reversed by the Second Circuit in a written decision published on April 11, 2012. United States v. Aleynikov, 676 F3d 71 (2d Cir 2012) (the "Second Circuit Decision"). At the time his convictions were reversed, Aleynikov had already served one year in prison for them. The Second Circuit found that Aleynikov's conduct did not violate either the NSPA or the EEA. With respect to the NSPA, the Court held that the source code Aleynikov uploaded did not constitute stolen "goods", "wares" or "merchandise" as defined by the statute. With respect to the EEA the Court held the statute was not violated because Goldman's HFT system was neither "produced for" nor "placed in" interstate or foreign commerce as required by the Act.7

Following the reversal of Aleynikov's federal conviction he was charged in the instant case in New York State court on July 31, 2012. More than 2 ½ years of motion practice before Justice Zweibel of this Court then ensued. Mr. Aleynikov has been at liberty throughout his state prosecution. The state motion practice resulted in several extended, unpublished decisions by Justice Zweibel. In a Decision and Order on May 2, 2013, he upheld the legal sufficiency of the grand jury indictment and denied the Defendant's motion to dismiss that indictment on grounds of double jeopardy, collateral estoppel, vindictive prosecution and in the interest of justice. In a 71 page decision on June 20, 2014, he found Aleynikov was arrested without probable cause because the FBI made a mistake of law when they seized him for violating the NSPA and the EEA. A thumb drive and laptop computer recovered incident to that arrest were suppressed on that basis.

The Court found that a subsequent search of Aleynikov's home at which additional computers were found was lawful because Aleynikov's wife consented to the search. However, Justice Zweibel found that the U.S. Attorney's Office wrongfully transferred this lawfully seized property to the New York County District Attorney's office after Aleynikov's acquittal in the federal case. Judge Zweibel suppressed computers found at Aleynikov's home on that basis. The combined impact of these two rulings was that all of the physical evidence recovered from Aleynikov's person and home evidencing his unlawful downloads were suppressed. Justice Zweibel finally granted the Defendant's motion to suppress a statement he made to the FBI upon

Page 4

his initial arrest but denied Aleynikov's motion to suppress a later Mirandized statement. 8

The instant trial was not lengthy but the jury deliberations were notable in a number of respects. The jury deliberated for eight days before reaching a partial verdict. The Court received and responded to 23 jury notes. Jurors early in deliberations asked for a copy of the Court's written instructions outlining the elements of the charged crimes and when the Court declined that request asked repeatedly for a read-back of the relevant crime elements. When the Court later provided a copy of its written instructions on the elements of the charged crimes to the jury after both parties consented, the jury wrote multiple detailed questions to the Court regarding the meaning of various statutory terms.9 They also requested multiple read backs of testimony regarding the content of the computer code Aleynikov obtained.

After extensive deliberations, the Defendant agreed to excuse remaining alternate jurors who were still available for jury service and the Court, with the consent of the parties, also excused two deliberating jurors. That latter excusal resulted from one juror's completely unfounded accusations that a second juror was "colluding" with one of the parties and tampering with her food and the Court's determination following these unfounded allegations that both jurors had become unqualified to continue deliberations. The Defendant then declined to exercise his right to a mistrial and consented to continue deliberations with 10 jurors.10 That ten person jury then returned the instant partial verdict.

Relevant Testimony of Key Witnesses

Testimony of Joseph Yanagisawa

Joseph Yanagisawa testified that he is a vice-president and technology and forensic officer at Goldman. On June 30, 2009, he was notified of a large transfer of computer data to a secure site outside Goldman which had occurred on June 1, 4 & 5, to a "Subversion" website. He described a Subversion website as "an area where you could actually move code. . . . You could move it, you can copy it, you can store it offsite" (76). He determined that the user who

Page 5

had sent the download was Sergey Aleynikov.

Mr. Yanagisawa determined that commands were put in by Aleynikov to collect the data, put it into something called a "Tarball", encrypt the data and then send it to the foreign site. The data which was transferred was computer source code for Goldman's HFT trading platform. Mr. Yanagisawa agreed that someone who had a thumb drive or flash drive and placed it in an appropriate USB port would be able to download the entire Goldman HFT platform. (118-120).

Testimony of Adam Schlesinger

Adam Schlesinger testified that he formerly worked at Goldman as a vice-president and was employed by the company in 2009. He said firms other than Goldman engaged in HFT in 2009 and that the HFT world was a "highly competitive environment" (219). Goldman did not share information about its HFT system with competitors. Information from Goldman's HFT system would be valuable to competitors. HFT systems have various components. Algorithms are mathematical formulas which encapsulate business logic. "Latency" expresses the delay in a system. "Resiliency" means the ability of an HFT system to be "up and running" at all times an exchange is operating. Speed is important to the success of HFT systems. If one firm receives information about market activity before another, the first firm has an advantage by being able to act more quickly on the information. Goldman participated in the market by buying and selling stocks, exchange traded funds ("ETF's") and options.

HFT systems are large but can be broken into three parts. First are algorithms or business logic. The second part is market access which is broken into two components: telling exchanges what you wish to buy with an acknowledgment that a purchase has been made and data on the market price of a security. The third category is the general infrastructure which supports trading environments. In 1999 Goldman purchased Hull Trading Company and Goldman's HFT system was thereafter largely based on this acquisition. The purchase price for this company was $450 million but Mr. Schlesinger was not aware of the extent to which this price reflected assets other than the HFT system. HFT software was updated "every day" since that acquisition but some original components of the Hull system were also in use when Mr. Schlesinger left Goldman in 2013.

Aleynikov was hired by Goldman in 2007 to work on HFT infrastructure and a particular algorithm trading on the NASDAQ exchange. He was originally paid $275,000 per year but when he tendered his resignation to work at a different firm in 2008, Goldman increased his compensation to $400,000 per year and Aleynikov decided to stay at Goldman. In April of 2009, Mr. Schlesinger supervised 25 to 30 persons at Goldman including Mr. Aleynikov, almost all of whom were computer programmers. HFT computer source code was maintained in a repository. Computer programmers had different levels of access to the code in the repository. In April of 2009 Mr. Aleynikov again tendered his resignation from Goldman to work at Teza which had offered him a $1.2 million annual salary. Mr. Schlesinger testified that: "I can't tell you that that business [Teza] is the same as Goldman's market making efforts" (238). Goldman did not try to match the Teza offer. There was no reason for an employee to transfer files out of the Goldman network in order to work from home since Goldman employees could log-on to the Goldman system remotely from home.

The code which Aleynikov uploaded was developed at Hull and then modified at Goldman so some of the software dated back 15 or 20 years. Part of the uploaded files were not

Page 6

part of the trading system but were files which could be used to build software for the HFT system. Part of the code was "an attempt to create a next generation trading infrastructure". (256). "Production code" is computer code which runs a system while "Development Code" is code which is being modified, although 99% of the content of Production vs. Development code are the same. Another uploaded directory was "Probe". Probe was a method of measuring the effectiveness of a trading system. Probe also allowed algorithms to be reactive to their environment. Probe was developed internally by Goldman. One person (not Aleynikov) developed Probe in 2008 and it took six months to develop but was probably edited over time. Mr. Schlesinger did not have specific knowledge regarding the extent to which competitors had a similar system.

A program which was part of the upload called "TV" was a theoretical value library used by the options group. Mr. Aleynikov worked in the "stock group" and did not have any job responsibilities with respect to that code. Part of the theoretical value library was purchased from Hull and then updated by others at Goldman. Mr. Aleynikov would have been able to write the program math for this program but would need experience with options and options-pricing to be able to work in the area. In response to the question of whether a Goldman competitor would be able to use source code from the theoretical value library in "some fashion" Mr. Schlesinger replied: "I think it would be interesting. I don't think you could just take it and plug it in and just start using it and think it would be rather difficult to understand unless you had a background in that area". (273).

One directory Aleynikov uploaded was called "IC" which stood for "index cash" which Mr. Schlesinger said he did not do a lot of work in. A directory named "pop" was a market-making stock algorithm used by the stock group. A directory called "TWGNMS" was a list of which securities were trading on which exchanges which was largely unchanged since its purchase from Hull. "OBB" was a component which "knew how to read the market data from stock exchanges" (277). OBB "is a way to organize all of these [market] orders into what is called a book and present these books to the trading applications so they can understand the state of the market". Id. A programmer named Navin Kumar developed OBB at Goldman from scratch. A competitor would gain an advantage from acquiring OBB. OBB was a high-quality piece of software when compared to competitors in 2009 and so a person starting an HFT system would "be able to start trading and receive information at a fairly high level" (279). It would also improve a competitor's time to market and "take away from the profits of Goldman Sachs". Id.

A competitor would be able to use OBB because it was developed outside Goldman's primary infrastructure, did not have dependencies and had "sample usage". Goldman had evaluated OBB against the comparable vendor products "SR Labs" and "Wombat" and decided to keep using OBB. OBB was not available in the marketplace and Aleynikov did not participate in OBB's development. A file called "Dense map" was used to maintain a list of orders in the order book and Aleynikov did not write the code for Dense map. Another file Aleynikov uploaded was called "data link" which was a utility which enabled the transfer of data. Part of data link was acquired through Hull and updates and modifications were subsequently made to the software. Aleynikov worked on a part of this program and there are commercial products which serve similar functions.

A number of files which Aleynikov uploaded were related to trading infrastructure which

Page 7

Mr. Schlesinger defined as computer code which is not part of the "business logic" of trading. Mr. Schlesinger said that "as a whole" the source code Aleynikov downloaded was "valuable to Goldman Sachs" (285). Programmers can check out computer source code from a Goldman repository, work on the code and then check it back into the repository. Normally, it is not necessary for a programmer to make a copy of the code for that purpose but if they were to make a copy within the Goldman system as part of their work no problem would arise.

During his employment at Goldman Aleynikov had access to all of the firm's HFT software. A USB flash drive capable to downloading Goldman's entire HFT system would likely cost between $4 and hundreds of dollars. In 2009, downloading all of the code to such a flash drive would likely take less than half an hour. Mr. Schlesinger agreed that "Sergey Aleynikov had the kind of access at Goldman Sachs that would have enabled him to purchase a USB drive . . . stick it into Goldman's system and walk out with a flashdrive in his pocket with the entire [high-frequency trading] platform" (318). The entire HFT system could also be downloaded through a script which would capture every file in the system. Mr. Schlesinger said that Aleynikov's uploads of Goldman source code were "very selective". (321). His script "backed up software into a single file, removing certain components, then taking that thing he was able to transport it". (322-323).

Testimony of Mark Freeman

Mark Freeman, a current Goldman employee, testified that in 2009 he worked in the "Strats group" and was responsible for the installation of infrastructure for software for the front office and global linked databases. The Strats group was one of two technology divisions at Goldman in 2009. The Strats group was the division dealing with securities, risk management, financial models and complex business problems as opposed to the other technology group dealing with operations. The Goldman repository for computer source code was known by the initials "cvs". The "equities quantitative trading business" at Goldman used computers to execute a large volume of trades in a short time.

All of the people in the Strats group which numbered 700-800 people as well as other teams and divisions in the technology division of the firm had access to computer source code. Mr. Freeman described the business for which the HFT system was used as market making in which the firm attempted to "capture the spread" by buying a security at one price and selling it at a different price. Mr. Freeman agreed that he did not have "any idea whatsoever" of the value of the computer source code Aleynikov downloaded on June 1 and June 5, 2009 (436-437).

Testimony of Paul Walker

Mr. Walker said he was currently the co-head of Goldman's technology division. In 2009 he was a managing director in the "core strats team" at the firm. He said the team was responsible for data and software used for risk management in the trading business and responsible for the infrastructure of the system. He said that in 2009, the "quantitative cash group" traded securities, the "volume group" or "VOL" traded options and the ETF group traded Exchange Traded Funds.

Mr. Walker explained that human beings interact with an HFT system in two ways. First, human beings create the system and provide instructions which allow the computer to generate prices or trades quickly. Second, human beings monitor and sometimes change the behavior of the system while it is operating but on a bulk level rather than with respect to individual trades.

Page 8

HFT systems have three kinds of components. "Connectivity" allows a computer program to interact with stock exchanges or receive market data. Algorithms allow decisions to be made about trades. Finally, infrastructure is the software which is needed for the system to run robustly. Goldman programmers write and revise computer source code for all three systems and it is updated on a daily basis. Goldman trades with tens or hundreds of markets around the world.

Infrastructure is like a "health meter" for the system; the goal is for the system to be "robust" meaning comprised of software that works properly and responds well when anything goes wrong. Speed is very important in an HFT system. "[Y]our software needs to make many, many decisions every second, thousands or tens of thousand or hundreds of thousands of decisions a second" (447). Latency, which he said is the time it takes a message to get from a computer program to an exchange, is also important. HFT is a competitive business and was a competitive business in 2009. The ability to be profitable is "dependent on having a good algorithm and having good infrastructure and connectivity" (448). Goldman remains competitive by hiring the best people and constantly investing in software and infrastructure.

Mr. Walker said that "[t]o be successful in [the] business of high-frequency trading, you need a fantastic system that can trade well, has a great algorithm, has great connectivity to all the markets you want to interact in and has the appropriate infrastructure so that it runs every day robustly" (450). Goldman programmers working to modify source code check out source code from the source code repository, modify the code, merge it with code from other programmers and then use it as production software to run every day. In 2009, only developers who worked on source code had access to it. The code for the Goldman HFT system is updated every week or two. Mr. Walker said that:

The [computer source] code is where all of our intellectual property about how we operate this business lives. It's the algorithm we use to choose to make prices. It's the connectivity we have invested so much in being important and it's the infrastructure that a decade of experience has allowed us to create through tens or hundreds of developers working on it that have allowed us to run a robust piece of software (454-455).

Walker testified that the directories uploaded by Aleynikov include EKEQTW which he said "contain large parts the core of our high-frequency trading system" (456). He said the code stored in the directories obtained by Aleynikov included components of all three parts of the HFT system, that is connectivity, algorithms and infrastructure. Mr. Walker said that Goldman's HFT code was a "valuable asset" which would be very useful to a competitor trying to replicate Goldman's system (459). The algorithm and Theoretical Value Library could enact the same trading patterns as Goldman and "take market share away from us and diminish the profit of our businesses". Id. The usefulness of connectivity and infrastructure code to a competitor would depend on how different a competitor's trading system was from Goldman's. With respect to the potential value of source code in the directories Aleynikov duplicated to a developer at a start-up, Mr. Walker opined such code would be useful because:

[A] developer would have the answer in the back of the book from hundreds of people, tens or hundreds of people developing a system. If you are at a start-up

Page 9

and you are asked to undertake a task and you can refer to how that was done in a system that you know works, you will be much more productive, much more rapidly able to develop a system that works for the competitor (460).

Mr. Walker said that the value of the source code which Aleynikov took derived, in part, from the fact that it is secret to Goldman.

Walker acknowledged that if an HFT firm is not trading in a particular market, connectivity to that market would be irrelevant. Mr. Walker did not know if Teza was trading in any market in June of 2009. The code Aleynikov downloaded was "integral to the operating model" of the HFT business in 2009 (465). Mr. Walker could not say the degree to which the computer code downloaded by Aleynikov accounted for the $300 million earned by Goldman from its HFT business in 2009. Roughly 45 Goldman employees, including Aleynikov, had access to Goldman's entire HFT platform. He agreed that under some circumstances, even using information a programmer retained in his brain regarding Goldman's HFT system could violate Goldman's confidentiality policy. Mr. Walker acknowledged that Aleynikov did not transfer all of Goldman's HFT computer source code but said he believed Aleynikov had taken a "substantial" part of it. He agreed Aleynikov would not have been able to use the source code he transferred to compete against Goldman on the day the transfer occurred.

With respect to Aleynikov's intent in obtaining the source code the following colloquy ensued during cross-examination:

Q: Did you ever entertain the possibility that what Mr. Aleynikov was doing was storing code, some of the code to Goldman Sachs' high-frequency trading platform so that he could study it and have it assist him in his new position at Teza?

A: I have had that opinion, yes. (505).

Testimony of Demian Kosofsky

Damien Kosofsky is a platform engineer responsible for writing computer source code for Teza's HFT system. He began working at Teza in May of 2009. At that time, Teza's HFT business was not up and running and Teza did not have any equipment or connectivity to run an HFT system. He said that Teza planned to obtain software to run its HFT system by writing it. He described the tasks which would be necessary for Teza to develop an HFT system as:

To start with, we would need a research platform. We would need to — as I said, acquire historical market data integrated into [a] research platform, provide a scheme or system where researchers can build models. Provide a container that they can trade those models within, provide connectivity to external markets to both receive data and send data to those markets. Provide a system to monitor our software, provide a system to integrate with clearing systems, a system to monitor our risk. Probably missing a few (527).

Teza's CEO, Misha Malyshev, had set a target date after late December of 2009 to begin high-frequency trading. An e-mail from Mr. Malyshev dated May 31, 2009 was sent to various Teza staff and future staff including Mr. Aleynikov. The title of the e-mail was "Let's move fast" and exhorted its recipients to move rapidly to achieve "the fastest and most scalable trading

Page 10

platform that will be our edge on the market" and "propel us ahead of competition". Mr. Kosofsky did not interpret the email as saying that "heads were going to roll" if the system was not operational by that time but was more in the nature of a "rallying cry" (553-554). He said that having "already-built" components of an HFT system would have been useful to Teza in 2009 "[a]sssuming it would be something that we would use" (534-535).

Mr. Aleynikov sent an e-mail to various Teza staff dated June 25, 2009. The e-mail indicated that Aleynikov had put some source code onto Teza's source code repository, svn.cvsdude.com. On July 2, 2009, Mr. Kosofsky attended two meetings at Teza's offices in Chicago with Mr. Aleynikov and Teza staff. At the first meeting, Aleynikov provided a tutorial in the programming language "Erlang" which Mr. Aleynikov advocated using. At the second meeting there was a discussion of what would be needed to build Teza's HFT system. Mr. Kosofsky agreed that until there was a "general sense of here is how we are going [to] approach our high-frequency trading business" "you certainly couldn't know that any of this [the source code Aleynikov transferred to Teza] would be even remotely of sustenance in writing code" (562-563).

Testimony of Konstantin Shakhovich

Mr. Shakhovich is a managing director at the Goldman strats group. He described the strats group as "the group in the securities division responsible for writing software and developing mathematical models that the security division uses to trade" (565). In 2009, he was a managing director of a group responsible for market-making operations in a highly automated fashion at Goldman. Mr. Shakhovich explained the potential value of Goldman's theoretical value library or "TV" , which was duplicated by Aleynikov, to a competitor as follows:

One potential advantage is that this library contains the accumulated knowledge and experience of tens of man-years of development of this library. So, essentially, it encapsulates a lot of our experience and things that we have learned while developing this library and that could be quite useful to a competitor. Secondly, the code itself could serve as a blueprint for somebody seeking 11 to build a similar operation. It also contains information about what our trading strategy is. It contains information about the assumptions we make about the relationships between the prices of different options and the competitor might choose to exploit that by tuning their strategy to kind of optimally enact or avoid interacting with ours. (572-573).

Testimony of Michael McSwain

Michael McSwain was the lead FBI agent who investigated the Aleynikov case. After Aleynikov's arrest, he told McSwain that he uploaded a file called "Atomic-Int" and another file called "Logger" outside of the Goldman network to a server called svn.xp-dev.com. and then downloaded that code to his home computer. He described these two files as open source files. Aleynikov also said that computer code he had obtained from Goldman was on a thumb drive in his pocket. He said that after he downloaded code to his computer, he had to transfer it to a thumb drive and then recopy the code to a different part of his home computer. Recounting his conversation with Aleynikov, McSwain testified that Aleynikov said:

he had downloaded them [the files from Goldman] to his desktop computer at home that they also might be on his wife's computer at home and that the computers were backed up to a hard drive in their home, that the files would be on his laptop computer on the Windows side and on the UNIX side and finally on the thumb drive that was with him that night (943).

Aleynikov said he had downloaded the files because "he wanted to inspect the files much like a person in college would go back and read a paper" (944). Aleynikov also made the following written statement (People's exhibit 18):

On or about June 5, 2009 I created a tarball in a [SIC] effort to collect open source work on Goldman Sach's server to which I had an account. I had previously worked on the files. I then used an encryption software to encrypt the tarball. I then uploaded the software to a repository server. The URL was sun.xp-dev.com. I then erased the encryption software and the tarball. I then erased the bash history. At a later date I down loaded the software on my home computer, laptop and thumb drive.

The reason I up loaded to sun.xp-dev.com was because it was . . . I wanted to inspect the work later in a more usable environment. At a later date I opened the files that were downloaded from sun.xp-dev.com to inspect the content. At that point I realized that I downloaded more files than I intended. These files have been not shared with any person or corporation. I[t] was not my intent [to] be involved in any malicious action. I have signed an agreement with my new employer to not to [SIC] bring in any unlicensed software. I have not violated that agreement. I have uploaded files to sun.xp-dev.com on multiple occasions.

The files that are proprietary information of Goldman Sach's have not been shared with any individual or corporation. When I was working from home I have logged into my Goldman account through remote access from my home. During the course of the session I uploaded files to sun.xp-dev.com. I will cooperate fully to find the full allocation of files on computers I downloaded the software to.

McSwain said he received three hard drives from the Subversion server in Germany. A search warrant executed at Aleynikov's home resulted in four desktop computers, two laptop computers, hard drives and thumb drives. Aleynikov said he had transferred 32 megabytes of computer source code from Goldman's HFT system in five separate transfers but there was overlap in the code he transferred so there were not 32 megabytes of unique code. McSwain said he was not aware of what the value of the source code Aleynikov duplicated was. He said Mr. Schlesinger had told him the code Aleynikov uploaded was worth more than half a million dollars to Goldman (1071).

Regarding the physical properties of computer source code, McSwain said that "[i]t takes up physical space in a computer hard drive" (1010). He also said that hard drives have finite capacities. McSwain agreed that the source code Aleynikov downloaded would not enable him

Page 11

to start an entire new HFT firm. He said that Adam Schlesinger told him it would take $3 - $5 million for Goldman to recreate the OBB system.

Testimony of Navin Kumar

Navin Kumar is a former Goldman employee who worked at the firm from 2006-2009 and was the author of OBB. He said OBB was "taking raw exchange data and the exchanges are giving at essentially bids and offers that other participants have and when they cancel them. And it's taking those sets of messages and keeping track of which ones are active and calculating which one is the best offer best bid from that." (1187-1188). Speed was very important to the program. OBB had few dependencies, meaning the need for some other code in order to create a program. This meant it would be easier to implement in a system other than Goldman's than a system with more dependencies. Densemap 1-D and Densemap 2-D are part of OBB. Mr. Kumar said that these two files shared "design decisions" with some code which was identified during the trial as having been downloaded to Teza's server by Aleynikov. This meant, in Mr. Kumar's opinion, that this code on Teza's server was based on the design for these Goldman files (1196-1197).

Mr. Kumar said that when computer files are stored in a hard drive or a CD those files are physically present on that medium. He also asserted that while it is not possible to see individual bits of data "in aggregate like in a burned CD, you'd be able to see what is written or if anything is written" (1198). He agreed that "the success of high-frequency trading systems is primarily driven by the successful and complex integration of hardware, software, data feeds, system design and connectivity, as well as expert personnel." (1199). He agreed that "components of a high-frequency trading system . . . can't fairly be judged in a vacuum . . . they can only be judged in the context of the system in which they arise" (1203). A specific dollar value cannot be ascribed to any particular component of an HFT system.

Mr. Kumar agreed that with respect to a program like OBB, "you better have it in a system that has latency addressed, that has a theoretical value system addressed that can identify the proper opportunities and that's before we even begin to talk about whether your trading options or futures or what have you." (1205-1206). He agreed that "computer software that works tremendously effectively in once scenario, in one high frequency trading system might not work at all transferred to another" (1207). He estimated that "maybe" 20% of Goldman HFT software would be rewritten on an annual basis. Trading systems if not updated will lose value over years. Describing OBB as a system which "takes exchange data and calculates what the active bids and offers are from the exchange" he agreed that "every single high frequency trading system in the world does that" (1211-1212). Most firms write their own "feed handlers" from scratch. Feed handlers are raw exchange data and processing what the exchange is sending you and building an order book from it.

A component taken from one HFT system to another may or may not have value. He agreed that "computer source code does not have physical form" (1216). Wombat was a commercially available equivalent for certain components of an HFT system. Mr. Kumar agreed that certain components of an HFT system can be taken and adapted to second system, that having an existing system as a reference would make it easier for someone to create a new system and that OBB could be beneficial and profitable in that sense. Computer code can exist on a computer without being stored physically when you are in the process of editing a file and it

Page 12

is stored in the volatile memory section of the computer.

Defense Witnesses Peter Friedman & Misha Malyshev

The Defendant called two witnesses during the trial: Peter Friedman and Misha Malyshev. Mr. Friedman currently works at Teza and in 2009 worked at his own recruiting firm. His firm recruited Aleynikov to work at Teza. He said that recruitment did not include any conversations regarding the transfer of Goldman computer source code to Teza.

Misha Malyshev said that in 2008 he was the global head of high-frequency trading at Citadel Investment Group, a hedge fund, and that his compensation that year was $150 million. He said the team he managed at Citadel earned a profit of $1.2 billion for the firm that year. He left to form his own company, Teza, which he hoped would be the best high-frequency trading firm in the world. At the time Malyshev offered Aleynikov a position with Teza in 2009, Teza had not yet determined what its trading strategy would be, had decided what exchanges Teza would begin trading with, was not "co-located" with any exchanges, had not decided whether Teza would trade options and had not determined what computer language Teza's source code would be written in.

At the time he founded Teza, Malyshev did not intend to acquire the components of an HFT system from another company. He explained that "any code gets old, the ideas get old too. If you try to build something which is the best of — the best in the world, you have to start from scratch and you have to, you have to do it all by yourself." (1406). Teza employees were not permitted to bring proprietary source code from other firms to Teza. In 2008-2009, Malyshev did not consider Goldman to be a significant competitor in the HFT field. He said that banks are not typically good at HFT. Bank HFT systems tend to be "monolithic". "Any company that depends upon the software develops the software over many, many, many years, right, changing the software is a very, very hard, right, like this, the systems and the banks tend to be stale" (1407).

He said he did not hire Aleynikov to bring Goldman's HFT software to Teza and would not have accepted Goldman's HFT code even if it had been offered to him for free. He never discussed obtaining Goldman's HFT software with Aleynikov. The compensation Malyshev offered Aleynikov made Aleynikov the second highest paid employee of Teza at the time. In June of 2009, Teza's HFT platform was in its "infancy".

CONCLUSIONS OF LAW

Under CPL 290.10 (1), at the conclusion of the People's case or the conclusion of the evidence, the court may enter a trial order of dismissal where "the trial evidence is not legally sufficient to establish the offense charged therein or any lesser included offense". The court may also reserve decision on such a motion and then determine it with respect to any count for which a guilty verdict is returned or for which the jury is discharged before the rendition of a verdict. That is the procedure the Court followed here.

"A conviction is legally insufficient where, viewing the record in the light most favorable to the prosecution, there is no valid line of reasoning and permissible inferences from which a rational jury could have found the elements of the crime proved beyond a reasonable doubt". People v. Maldonado, 24 NY3d 48, 53 (2014) (quotations and citations omitted). "A sufficiency inquiry requires a court to marshal competent facts most favorable to the People and determine whether, as a matter of law, a jury could logically conclude that the People sustained its burden of proof". People v. Khan, 18 NY3d 535, 541 (2012).

The secret scientific material count at issue here reads as follows:

A person is guilty of unlawful use of secret scientific material when, with intent to appropriate to himself or another the use of secret scientific material, and having no right to do so and no reasonable ground to believe that he has such right, he makes a tangible reproduction or representation of such secret scientific material by means of writing, photographing, drawing, mechanically or electronically reproducing or recording such secret scientific material. Penal Law § 165.07.

The Penal Law also provides four definitions for terms used in the statute, "appropriate", "benefit", "property" and "secret scientific material". The Penal Law defines the word "appropriate" as follows:

To "appropriate" property of another to oneself or a third person means (a) to exercise control over it, or to aid a third person to exercise control over it, permanently or for so extended a period or under such circumstances as to acquire the major portion of its economic value or benefit, or (b) to dispose of the property for the benefit of oneself or a third person. Penal Law § 155.00 (4).

"Secret Scientific Material" under the Penal Law:

means a sample, culture, micro-organism, specimen, record, recording, document, drawing or any other article, material, device or substance which constitutes, represents, evidences, reflects, or records a scientific or technical process, invention or formula or any part or phase thereof, and which is not, and is not intended to be, available to anyone other than the person or persons rightfully in possession thereof or selected persons having access thereto with his or their consent, and when it accords or may accord such rightful possessors an advantage over competitors or other persons who do not have knowledge or the benefit thereof. Penal Law § 155.00 (6).

The term "benefit"

means any gain or advantage to the beneficiary and includes any gain or advantage to a third person pursuant to the desire or consent of the beneficiary. Penal Law § 10 (17).

The term "property" is also defined by the Penal Law and in relevant part includes "computer data" or a "computer program". Penal Law § 155.00 (1). Finally, the Penal Law's definition of "deprive" (Penal Law § 155.00 [3]), although not part of the instant statute, is discussed infra and is thus also provided here:

To "deprive" another of property means (a) to withhold it or cause it to be withheld from him permanently or for so extended a period or under such circumstances that the major portion of its economic value or benefit is lost to him, or (b) to dispose of the property in such manner or under such circumstances

Page 13

as to render it unlikely that an owner will recover such property.

Origins of the Secret Scientific Material Statute

The sole statute at issue here, Penal Law § 165.07, was enacted in 1967.12 In the ensuing 48 years it has apparently been rarely used. There are only three reported decisions in which the validity of a conviction under the statute has ever been considered and only one where the statute's provisions have been analyzed, that case being People v. Russo, 131 Misc 2d 677 (Suffolk County Court 1986) [Corpertino, J.])13 According to the Office of Court Administration, since 2010, five cases, including the instant one, have been brought under the law, all in New York County. Those cases have resulted in one conviction under the statute, other than the instant case.

Justice Donnino's Practice Commentary to the law notes that it is intended to work in tandem with Penal Law § 155.30 (3), which makes it the Class E felony of Grand Larceny in the Fourth Degree to "steal property" which consists of "secret scientific material". As he points out, the secret scientific material statute at issue here is distinguished from the larceny statute because the former requires proof of some tangible reproduction or representation while the latter requires that the secret scientific material itself be stolen. Critically, as will be seen infra, however, Justice Donnino also notes that the two crimes share the "same larcenous intent". Put another way, both crimes require that the wrongdoer intend to steal secret scientific material.

The only legislative history applicable to the instant statute are memoranda which note that the chapter which enacted it (which also included other substantive provisions of law) was intended to amend the new Penal Law, which was to take effect on September 1, 1967, "to incorporate therein various amendments made by the Legislature to the former Penal Law during the years 1965 and 1966".14 There is, however, an extensive bill jacket outlining the legislative history of the bill which initially made the theft of secret scientific material larceny. That statute was enacted by Chapter 727 of the laws of 1964.15

That statute was intended to fill what was perceived as a gap in the Penal Law under which the theft of a secret scientific formula or process for which a value could not be readily assigned was not a felony. As the counsel for the Temporary Commission on Revision of the

Page 14

Penal Law and Criminal Code framed the issue, a theft of "secret scientific processes and formulae" could concern items "representing the product of years of scientific research and which promises a startling development in the particular industry. Losses incurred, though not measurable by legal standards, may, realistically speaking, be of fabulous proportions." 16 The case which featured most prominently as an impetus for the 1964 law was a civil and criminal proceeding involving the theft of trade secrets from a company called American Cyanamid and its Lederle Laboratories Division which manufactured antibiotic and steroid medications. It was alleged that the defendants "induced Lederle employees to steal confidential records and microorganisms from Lederle, which were turned over to the defendants and their confederates and then sold to various foreign and domestic pharmaceutical firms . . ." 17

The one reported decision which has analyzed the instant statute, People v. Russo, supra, rejected the Defendant's motion to dismiss the indictment on grounds of legal insufficiency, held that a "computer program" could constitute secret scientific material and rejected a constitutional challenge claiming the statute had not informed the Defendant a computer program could be covered by the law. The Court held that the indictment adequately alleged that the Defendants had electronically reproduced a computer program and then wrongfully sold features of it to a third party. The Russo decision, in this Court's view, is not informative with respect to the issues here. The decision did not provide the underlying facts with respect to the kind of "electronic reproduction" the Defendant was accused of making nor did it analyze the question of whether the Defendant had the intent to appropriate the secret scientific material, the two questions here.

Meaning of "Tangible Reproduction or Representation"

In arguing that his motion should be granted, the Defendant first asserts that the trial evidence was legally insufficient to demonstrate Aleynikov made a "tangible reproduction or representation" of the computer source code as required by the statute. The word "tangible" is not defined by the Penal Law and has never been defined under the instant statute in any reported decision. If the evidence during the trial viewed in a light most favorable to the People demonstrated that Aleynikov made a "tangible reproduction or representation" of secret scientific material when he duplicated the HFT source code then the evidence with respect to that issue would be legally sufficient. If the evidence (again viewed in a light most favorable to the People) indicated Aleynikov duplicated, copied or transmitted the code but not in a tangible form, then the evidence would be legally insufficient on that point and the Defendant's motion would have to be granted on that basis alone. This Court holds that the evidence was legally insufficient on this point.

The Court begins with the dictionary definitions of "tangible" when that term is used as an adjective. Dictionaries generally give three definitions for the word, two of which concern the manifestation of a thing in the physical world (the "physical definitions") and one of which construes the term with respect to thought processes. Black's Law Dictionary defines the word as follows:

1. Having or possessing physical form; CORPOREAL. 2. Capable of being touched and seen; perceptible to the touch; capable of being possessed or realized. 3. Capable of being understood by the mind. Black's Law Dictionary, 9th Ed.18

There was no evidence during the trial that the source code Aleynikov duplicated could be touched. If the word had that definition then the evidence here would be legally insufficient. On the other hand, the evidence clearly indicated the source code was "capable of being understood by the mind". If that definition is ascribed to the term, then the evidence would be clearly sufficient on this point. The construction "having or possessing physical form" and that term's related constructions present a question requiring a more detailed discussion, infra. But of course, the first order of business here is to ascribe one of multiple meanings to the term. That task is made more difficult by the lack of any legislative history indicating the meaning the law's drafters intended to ascribe to it.

Argument That "Tangible Reproduction or Representation" is Not a Physical Definition

The meaning of the term which would result in a clear finding of legal sufficiency, when read in isolation, is plausible. "Secret scientific material" as defined by the Penal Law has economic value to its legitimate holders because it may confer an advantage over competitors. Penal Law § 155.00 (6), supra. It can be argued that the "capable of being understood by the mind" construction of the word "tangible" makes sense because the drafters of the statute intended to criminalize only representations of secret scientific material which were comprehensible. In other words, it might be argued, the word "tangible" was inserted into the law to distinguish reproductions or representations which had economic value (because they could be understood) from copies of inventions or formulas which could not be understood, had no value and thus should not result in criminal liability. Thus, the People argue, "tangible" as defined by the statute may simply refer to something which is "real and not imaginary, or extant and not illusory".19

When the entire secret scientific material statute is considered, however, the logic of this analysis fails. First, the statute requires a defendant have the "intent to appropriate . . . the use of secret scientific material". As discussed in more detail infra, the intent to appropriate is the intent to steal. It is the intent "to exercise control over it [the secret scientific material], or to aid a third person to exercise control over it, permanently or for so extended a period or under such circumstances as to acquire the major portion of its economic value or benefit" or to dispose of the property for one's benefit.

It would seem odd, to say the least, however, for the Legislature to insert the word

Page 15

"tangible" into the statute to avoid punishing people for intending to obtain economic value from something which was worthless. A copy of an invention, formula, recording, document or technical process which was not "capable of being understood by the mind" would have no value. A wrongdoer could not possibly have the intent to acquire the "major portion of the economic value or benefit" of such an item. Secret scientific material is valuable only because it is cognizable. But if that is so, then the word "tangible" meaning "capable of being understood by the mind" would be a superfluous addition to the statute. To cite one of the People's proposed constructions, the Legislature surely did not insert the word "tangible" into the statute to clarify that secret scientific material had to be "real and not imaginary" in order to result in a felony conviction. We can safely presume the Legislature was not concerned about clarifying that New Yorkers should not be punished through the criminal law for their imaginations.

The definition of "secret scientific material", as noted supra, also requires that such material "accords or may accord such rightful possessors an advantage over competitors or other persons who do not have knowledge or the benefit thereof." Penal Law § 155.00 (6). It would seem impossible, however, for a wrongdoer to obtain a competitive advantage by wrongfully duplicating something which was not "capable of being understood by the mind". As with the appropriation issue, secret scientific material is only valuable if it can be understood.

It is a fundamental principle of statutory construction that a court must assume every word in a statute has a meaning and was inserted for a purpose. See Bliss v. Bliss, 66 NY2d 382 (1985); Direen Operating Corp. v. State Tax Commission, 46 AD2d 191 (3d Dept 1974); NY McKinney's Statutes § 231. "In construing a statute, no part thereof is to be considered meaningless unless that conclusion is inevitable, and words in statutes are not to be rejected as superfluous when it is practicable to give to each a distinct and separate meaning." Id. The only way to make sense of the word "tangible", meaning "capable of being understood by the mind", would be to violate that statutory maxim — twice in the same statute. But that would not only be inappropriate and, as discussed infra, contrary to the overwhelming weight of legal authority. It is unnecessary. A perfectly reasonable alternative construction exists. That comes from the physical definitions.

The People next argue that what must be "tangible" under the statute about a reproduction or representation is not the reproduction or representation itself. Rather, they argue, citing one of multiple examples, a photograph can be "represented tangibly on paper" or as an attachment to an email. "In each of these examples, the representation or reproduction is tangible' by virtue of its manifestation on a physical medium".20 This argument, again, has a degree of initial plausibility. However, like the argument that the word "tangible" means "capable of being understood by the mind" this construction would render the word "tangible" superfluous. Any "reproduction" or "representation" is associated with some physical medium. Whether information is conveyed visually, orally or through some other sense, it must be associated with something in the physical world. The only possible exception would be some kind of extra-sensory communication. But extra-sensory communications, assuming they exist, are certainly not legally cognizable.

The People next argue the word "tangible" may have been used to distinguish a person's "memory" from the manifestation of information in a more concrete form. Thus, the People argue, "[h]ad the Defendant simply used his memory to recall certain code he saw, or algorithms he created, during the course of his work at Goldman Sachs, he would not have been in violation of this particular statute."21 The New York Penal Law, however, obviously doesn't criminalize memories. There would be no reason for the Legislature to insert the word "tangible" into the statute to clarify that having a memory about confidential intellectual property is not a crime.

The People assert the word "tangible" may have been intended to distinguish concrete representations from speech. But the statute clearly does not criminalize speech. It lists the kinds of tangible reproductions or representations it covers as "writing, photographing, drawing, mechanically or electronically reproducing or recording". That list does not include speech. "It is a universal principle in the interpretation of statutes that expressio unius est exclusio alterius. That is, to say, the specific mention of one person or thing implies the exclusion of other persons or thing. . . . "[W]here a law expressly describes a particular act, thing or person to which it shall apply, an irrefutable inference must be drawn that what is omitted or not included was intended to be omitted and excluded." McKinney's Statutes § 240; See also, East Acupuncture P.C. v. Allstate Insurance Company, 61 AD3d 202 (2d Dept 2009); Caspian Realty Inc. v. Zoning Board of Appeals of Town of Greenburgh, 68 AD3d 62 (2d Dept 2009), lv denied, 13 NY3d 716 (2010).

Finding the word "tangible" was inserted into the statute because, otherwise, speech might be criminalized could also only make sense if the other words of the statute were given an unnatural construction. In common parlance, a person who orally relays the terms of a technical process, invention or formula is not said to have made a "reproduction or representation" of it. He is described as recounting, repeating, saying, talking or speaking about the subject.

The People point out that the statute explicitly covers the act of "electronically reproducing or recording" secret scientific material and that they surely demonstrated Aleynikov did that. It is certainly true, moreover, that any transmission of data through a computer system could be said to have made an "electronic reproduction" of it. As with other constructions, however, this one would render the word "tangible" meaningless. If Aleynikov made a tangible electronic reproduction in this case, what would an intangible electronic reproduction have looked like? More to the point, what would an intangible electronic reproduction have looked like when the statute was enacted in 1967?

The Court takes judicial notice of the fact that computers were at a very primitive stage of development in 1967. The drafters of the statute certainly could not have imagined that in 2009, significant lines of computer code could be transmitted in an instant across thousands of miles to "Subversion websites" then transferred back across those miles to a personal computer in one's home. The electronic reproductions the drafters of the secret scientific material statute contemplated surely had little relationship to what is at issue here.

That is not to say a criminal statute cannot cover an activity which did not exist when it was written. It surely can. But what an electronic reproduction was in 1967 is highly relevant in construing the statute today:

A statute and the words used therein are to be construed according to the meaning thereof at the time of the enactment of the statute rather than a meaning subsequently acquired, and expediency both of changing circumstances and conditions will not be permitted to alter the meaning of the plain and ordinary language therein. McKinney's Statutes § 124.

See also, People v. Litto, 8 NY3d 692, 697 (2007) ("The plain meaning of a statute must be interpreted in the light of conditions existing at the time of its passage and construed as the court would have construed it soon after its passage'") (quotation and citation omitted); Department of Finance of the City of New York v. New York Telephone Company, 262 AD2d 96, 98 (1st Dept 1999), lv denied, 94 NY2d 755 (similar).

Both the People and the Defendant note media which the drafters of the statute may have considered when it was written. The People point out that fax machines existed in 1967. The Defendant points to the existence of photocopy machines. He also notes the existence of "electronic type-composition systems" for newspapers which could set the text of a newspaper on sensitized paper or film What all of these devices had in common in 1967, however, was not only that they were capable of producing electronic reproductions. These devices also produced paper copies. They produced tangible representations. The Cyanamid case which was the impetus for the 1964 statute which originally criminalized the theft of secret scientific material also concerned the physical taking of property. It concerned the theft of microorganisms and documents.

Outcome: For all of these reasons, the Court has concluded that assigning a non-physical definition to the word tangible under the instant statute is implausible. As will be seen infra, moreover, such a construction would also be contrary to the overwhelming weight of legal authority.

Plaintiff's Experts:

Defendant's Experts:

Comments:



Find a Lawyer

Subject:
City:
State:
 

Find a Case

Subject:
County:
State: