Comment Re:hmm (Score 1) 278
Good question, but the answer is that it does matter. I'm actually a huge fan of Python, but I can indeed see a big advantage in having some element of provability in the language chosen, and I do have some concerns about my favorite language being used for this. Python is a pragmatic, readable language but it isn't a great one for provability.
To address your insightful point: You don't have to trust the compiler. The SEC wants source code to be given. Since you can run the source code on any correct compiler, including one you yourself just wrote, and expect to get the same result, there's no way to exploit it through that.
The unraised point is that you can, in fact, exploit the program code this way. Given innocuous code that in certain situations--some of which may be intentional on the part of the author of the code, some of which may not be--you can get surprising and highly lucrative corner cases. A clever coder can create these intentionally; a bad one can do it unintentionally, and either way it can be hard to spot.
I think the answer to the latter issue is "unit tests". A third party can write tests for a bunch of inputs, run it through your code and make sure your code produces the expected outputs. Discover bugs and backdoors this way, and report them to the code author or SEC, whichever is appropriate.