Evil finance should contribute to open source
This is a guest post from an friend who wishes to remain anonymous.
I’ve been thinking lately that a great way for “evil finance” to partially atone for its sins is to pay a lot of money to improve open source libraries like numpy/scipy/R. They could appeal to the whole “this helps cancer research” theme, which I honestly believe is true to some extent. For example, if BigData programs ran 10% faster because of various improvements, and you think there’s any value to current medical research, and the cancer researchers use the tools, then that’s great and it’s also great PR.
To some extent I think places like Google get some good publicity here (I personally think of them as contributing to open source, not sure how true it really is), and it seems like Goldman and others could and should do this as well — some sliver of their rent collection goes into making everybody’s analysis run faster and some of that leads to really important research results.
Personally I think it’s true even now; Intel has been working for years on faster chips, partially to help price crazy derivatives but it indirectly makes everybody’s iPhone a little cheaper. Contributing directly to open source seems like a direct way to improve the world and somewhat honestly claim credit, while getting huge benefit.
And it simultaneously has nice capitalist and socialist components, so who wouldn’t be happy?
There are actual problems banks care about solving faster, a lot of which uses pretty general purpose open source libraries, and then there are the maintainers of open source libraries doing this nice altruistic thing but I’m sure could do even better if only they didn’t have a “day job” or whatever.
And if the banks don’t want to give away too much information, they can just make sure to emphasise the general nature of whatever they help fund.
mathbabe 
“Evil finance” ? I could tell from the title of the blogpost that it was going to be an extremely biased one. I was right. The worst part is that the coward who wrote it wouldn’t even reveal their identity. I’m nominating this as worst post in the history of Mathbabe.org.
LikeLike
Wow, so much to say, so much to ridicule. Majordomo, you may just be the guy who makes me enjoy being mean to a fellow human. First of all, the “evil” was clearly meant with tongue in cheek. Too bad you don’t get humor. Second, how can you possibly call someone out for being anonymous and cowardly when you post anonymous nasty comments? You are unbelievable.
LikeLike
Pardon me if this question is out of line, but how close is (for example) Sage (http://www.sagemath.org) to being in the right weight class to reimplement the kinds of algorithms you were using in finance? If you’re unfamiliar, it has the type of symbolic manipulation you get from things like Maxima, and the linear algebra tools of things like numpy/scipy. It also supports and links to code written in “R”.
I’m trying not to ask anything that would be likely to create conflicts with any possible NDA you may still be under.
LikeLike
Anonymous commented: “…Intel has been working for years on faster chips, partially to help price crazy derivatives…”
Some 18 or so years ago, Intel put out a severely flawed chip (P5 Pentium) that some good stiff open sourcing could well have prevented. Intel tried to suppress its revelation, but ended up charging off some $500 million in losses due to its bad quality control. As I recall, one quant admitted to losing some $44 million on a decimal point disaster due to the P5.
Some “conspiracy theorists” have linked part of cause of the world-wide debacle of Long Term Capital Management (besides mathematically ignorant use of volatility assumptions, among others) to non-upgraded P5 PC’s and servers, also along with an appallingly sloppy and inaccurate coding of M$ Excel spreadsheet algorithms.
LikeLike