Thank you, AmirES

A big thank you to AmirES for playing against me. This has lead my author to look more deeply into my algorithm for playing in situations where I am behind.

First, he noticed that, during the first four moves, I was always playing near the corner, which used to be important, in order to get into my endgame database quickly (back when my database wasn't complete).

But even once this was noticed, it was not fixed correctly.

In one recent game, for example, I sacrificed a box on turn 3, which even Bill was able to tell was probably not the most challenging move.

It turned out that, for the first four moves, all moves which only lost by one point were being weighted equally, giving me a 2 in 59 chance of sacrificing that box, instead of about a 4 in 17000 chance.

In other words, "Watch out! I think I've gotten stronger as player 2!"

I haven't had enough time to play with Shark again Unitl Now, but when I did it again today, I noticed that he really got stronger in the second player. So that I lost many of the games that I Play first Player.
And this surprised me
I did not leave much time and think for my moves
But before, with the same conditions, I easily won A Lot Against shark in the first player
It was an interesting work, thank you too

i guess its gotten a bit better, just challenged it a couple times and it beat me in 3 of 7 games. when i tried it half a year ago it only won 4 out of 11 games. i think it still could become a lot better, some of these games were much too simple to win. causing a human blunder by making the game extremely complex should be possible in really almost all games i believe. humans are especially bad in endgames involving 2 loops, for example this game i figured i was most likely gonna lose by move 4 just because of the top right becoming this isolated area that would likely turn into a 4-loop. on the other hand none of the 4 games i won felt even remotely difficult.