Hacker News Reader

Comment by philwelch | Hacker News Reader

"1. It would be pretty simple to write a computer simulation to test this, no? Would anyone care to do so?"

I'm on it. Results in a later reply.

Preliminary results: holding sex drive and promiscuity constant, decreasing societal shyness (i.e. making single people more likely to go out and meet someone to fuck) seems to increase population infection rates rather than decreasing them. This is true if shyness is independent from promiscuity. (My simulation tends to send more promiscuous people out to meet someone to fuck more often by increasing their partner-switching rate, so shyness is indeed independent.)

Don't just take my word for it, run my code: http://github.com/philwelch/SexSim

If anyone's interested, let me know and I can work up a blog post and improve my code. (Also, if my code is wrong, let me know or send me a patch.)

EDIT: I have a bug which may invalidate my results. Stand by for fix.

EDIT 2: No, results still valid. Possible fix pushed.

I took game theory back in college. I noticed something wrong about that article: a complete lack of math. It's not that I don't believe there are counter-intuitive things that can happen. But I sure don't believe them without proof.

And your simulation is a lot more interesting than anything found in that article.

It was a fun exercise. It was great for remembering a handful of things about statistics and experiment design--for instance, it was impossible to get good results until I started doing controlled runs on the same randomly generated population rather than regenerating, because the effect of changing social shyness (while there) is way, way smaller than the effect of the random variations in promiscuity and sex drive.

I may want to play with adding condoms to the simulation, though. My transmission rate is based upon the HIV transmission rate for unprotected anal sex, which is pretty clearly an upper bound--that's why especially slutty or outgoing populations get >50% infection rates after 20 years seeding from 1% infection rate. You also see the odd run where you just have a 1% infection rate after 20 years--probably cases where the randomly seeded HIV+ individual is especially shy and has a low sex drive.

Also, it would have been a lot more interesting if I could have generated random values along different distributions, or...well, lots of things would have helped.

Reply on news.ycombinator.com