quarantine 'em

These Tetris-playing AIs are not prepared for highly adversarial enemy AIs like HATETRIS and I think they are likely to perform very poorly under these conditions.

Under normal circumstances I think a good human can play about as well as the best Tetris-playing AIs. The AIs probably win out eventually, due to superior reaction times and a reduced tendency to get bored or make mistakes

Under adversarial conditions I believe a good human can far outperform the best Tetris-playing AIs. The reason I say this is that I simply can't find any examples of a Tetris-playing AI being trained under adversarial conditions.

This said, I don't see an easy way to train up a fitter, stronger HATETRIS unless it has a credible AI opponent. I can't train the thing manually. I'm not even that good at regular Tetris.

To cut a long story short ("TOO LATE") I think I need to create a Tetris-playing AI of my own.

Probably the easiest way to get started here is to take some existing code and pit it against HATETRIS in the training phase, instead of a more forgiving random piece generator.

HATETRIS is a fixed target and as far as I know it has never really been hit hard before, at least by anything other humans.

Recall that since HATETRIS is deterministic, it may be possible to catch it in a loop and score unlimited lines. It would be really exciting if that loop could be found.