Breakthrough bot !

Yes, we have one: NataliaBot, operated by Tommah.

Enjoy :)

Brilliant news - we've been needing one for a long time!

(Confession - I've written about 50 lines of command-line parsing for a bot too, but not got beyond that stage.)

I'm still trying to improve it; it won against me but has lost against stronger players.

My opinion: the bot has quite good middle play, but the endgame is very vulnerable for forced breakthroughs.

Cool :) I would love to play it.

Welcome back, everybody. Over the weekend I tweaked NataliaBot's Breakthrough program, so she might be a little smarter now about protecting the back ranks. Please try a few games against her.

It seems that NataliaBot doesn't care too much about exchanging on the opponent's side of the board.

Unless you have a good reason to, you usually do not want to exchange pieces on your opponent's side of the board because you would trade a pawn with many moves invested in it for a pawn of less value.

@halladba: Yeah, I agree on that.

@Tommah: You should consider letting Natalia play a few rated games (or a few rated games per month). It would help estimating her actual strength.

Agree - I'd say sign her up in the championship. That way she'll play against mostly similar rated players, who probably won't be able to able to pull clever bot-beating tricks on it, and therefore it will be more likely to be a fair rating.

In my current game against NataliaBot, the bot seems to give up on material for reasons obscure to me. They are either obscure because of my limited tactical skills or because the reasons are not good... I guess we'll find out shortly.

[game;id:1442524;move:38]

[game;id:1442524;move:44]

I must have the weights out of proportion. I'll fix it now; if you play her again, hopefully she won't throw away pieces like that.

Nataliabot gave up on a few other pieces after that, but the game was over so I guess there was just no point in holding on to material.

Would it be possible that Nataliabot found a winning sequence for White as early as the first of the two above positions? That would explain its throwing pieces away. What depth does its search-tree reach?