I haven't posted here in a few days because I've pretty much had my head down, working with Philip trying to get the basic tests verified on our neuroevolution project.

And we've done it. We've validated that the core algorithm works. A neural network evolves to solve XOR in an average of about 30 generations with an average of about 15 nodes and connections.

So now that we've tackled the simplest benchmark out there, it's time to move on to our first game domains. We'd hoped to release the code before the end of the year, but it'll likely be quite a while more. We'd like to release it with at least two benchmarks (XOR and Tic-Tac-Toe).

Anyway, we've been working on this project for over a year, so it's nice to see some progress.