This code is the latest hack among many to try to simulate our chosen guard algorithm. It's good to play with until we're sure we've got the rigor that we want!

I've stuck it at https://github.com/nmathewson/guardsim .

lib/tornet.py has a very oversimplified view of "who's a guard and who isn't", and of various network attackers.

doc/stuff-to-test.txt has a list of potentially interesting scenarios we should simulate someday.

lib/client.py has an attempt to implement the client guard selection algorithm.  I tried to follow prop259 approximately as I understood it, but I probably did a lot of stuff wrong.

The code runs, but it's littered with XXXX comments. It could probably use more logging and a better main-loop.

If you're going to focus on just one thing, I'd suggest making the client.py code more closely represent proposal 259, and extending proposal 259 to fill in any gaps that it doesn't describe which you need in order to make the code behave nicely.

If you're going to focus on another thing, I'd suggest making it report more things.