The premise of the author's argument seems to be "don't worry, the programmer won't do anything bad with the code". I don't think it's such a good idea to just implicitly trust that an employee who downloads your source code isn't going to do something unethical with it.
(especially in an environment in which the owner of that code, i.e. Goldman, has already taken great pains to prevent copying of the source code, such as removing physical access to the PCs, etc)
Goldman has to protect their shareholders in this kind of thing, and there really isn't any room for "let's just trust this guy won't do anything harmful to us" in that.
How much copied code constitutes theft these days? Arguably, a system may hinge on one to ten crucial calculations or routines, despite the many, many lines of code to make sense of that data and then present it to the user. Does copying those crucial algorithms constitute theft while copying the same quantity of procedural fluff not? I suspect there are some very grey lines in the precedential cases leading up to now. Does anybody know definitively?
Been to a live gig recently? I met more systems programmers backing up a vocalist on some instrument than anywhere else. The most recent one was a local fellow yank here in Sydney, he plays jazz for a living; this guy introduced C++ to MCI/LDDS/Worldcom back when it was a macro-processor, and he was their in-house compiler geek (btw, geeky bar talk with strangers has a peculiar taste to it; register allocation strategies over beer ftw! :-)
I also met a former Atari dude who now runs his own bar in Thailand.
Why is he so angry with Goldman Sachs? The firms survival of the crisis appears to indicate that they did not take excessive risks out of greed, and thus are not directly complicit?
Hmm, come to think about it, it reads a bit like anyone in finance is a greedy bastard who deserves bad things to happen to him.
If I were Goldman, I wouldn't be worried that he would build a better version of my software. What I would be worried about is that he would analyze it, find holes in the methodology and figure out a way to screw me on my trades that are made with this software. Either by beating me to the punch, OR trying to game the market is such a way that my software would make a trade that is not advantageous to me.
Although, one could also make the case that he already had access to the code before his departure as part of his daily responsibilities. Therefore, the fact that he was leaving had no real implications in him knowing the innards of the application and its source code.
What I don't understand is why there's no word about an NDA between Goldman and Mr Aleynikov. Had he been under contractual obligations, or bounded by a noncompete clause, Mr Aleynikov wouldn't have been incentivized to extract code from the repository to being with. It seems logical to me that if you're dealing with coveted IP, a contract between the parties is a necessity rather than a luxury.
