[gcv]

One thing about this case puzzles me, and none of the news coverage I found explains it: what code did he actually take?! Details matter. Was it a tweak of something like Samba (obviously GPL)? Or was it part of a proprietary risk management system, derivatives models, or other direct money-making programs?

If he took the former, then I'd side with him — he took nothing of genuine value. If he took the latter, then, très uncool. (I witnessed several incidents of model theft while serving time on Wall Street in the mid-2000s, and none were publicly pursued like this one.)

[jkimmel]

In Flash Boys, he makes it clear that he did not take any trading strategies or in-house models.

They don't provide a ton of detail in the book, but it seems that he wanted to take some FOSS code he modified himself while at Goldman. All code ever used on a Goldman machine is licensed as proprietary, even if it was downloaded from a FOSS repository 1 µs beforehand. It's mentioned that some irrelevant infrastructure code may have been intermingled with the modified FOSS he took, though it doesn't seem Goldman was particularly incensed by any particular piece of code, simply by the fact that he took any to begin with.

Also in the book, Michael Lewis brings together a panel of HFT technologists to interview Sergey and assess whether anything he took was consequential. Their collective conclusion after meeting with him for several hours was that he took absolutely nothing of value.

[tedunangst]

Why would you cite flash boys, which doesn't provide a lot of detail, and not the court filings, which do?

[jsprogrammer]

How does Goldman relicense FOSS software as their own proprietary software?

[JoshTriplett]

By never distributing the result outside the company, and not caring about intermixing proprietary code with it? The GPL, like most other FOSS licenses, only applies to distribution; no distribution, no license compliance issues.

[jsprogrammer]

That's not relicensing then. That is just not violating the license.

The person I was responding to said they took FOSS and made it proprietary in one microsecond.

[JoshTriplett]

A comment elsewhere in the thread suggests that they put their own proprietary license header at the top. Which is legal if they never distribute the result.

[jsprogrammer]

It would depend on the license.

You can't rip the copyright notices or license text off anything under the MIT license, for example.

[gcb0]

distribute it to whom? it was certainly distributed to the machines that are rented by a company which only assets is the fact it's close to the trading servers.

[pbh101]

That is (generally) not how it works. HFTs rent rack space in which to insert machines that they themselves build, own, and maintain. The code never leaves the network/property of the company in usual circumstances.

[jkimmel]

In the book, Sergey says they simply rip the FOSS license off the header and replace it with a Goldman proprietary license. It's implied that the legality of this is a grey area.

[belorn]

Ripping software licenses of from software is extremely unlikely to be a grey area. The license is what gives a person permission, and if they remove it, so goes the permission. I would compare it to a person boarding a train, and then imminently throwing away the ticket.

[kbenson]

If possible, it would depend on the license. Do we know the software in question?

[jsprogrammer]

This appears to be the code he took:

http://code.google.com/p/svnarchive/source/browse/trunk

Source: http://zerohedge.blogspot.com/2009/07/aleynikov-code-dump-un...

[j_baker]

What he took was proprietary infrastructure-related code that didn't involve any of Goldman's trading algorithms. His claim is that it was so tangled up with the open source code he wanted that he had to take it out to get it.