Do the bots go for 'draw by repetition' because real players would? Or is it just silly AI?

I've noticed the bots are quite easy to get to "draw by repetition" and it makes me wonder how often it happens with real people. Do pros ever really go for draws? Particularly by repetition?

I'll show my position here:

Black (AI) moves queen to H4. I defend with Queen to G3. So now the position is like this:

So then the bot goes back to Queen F6 and I follow back to Queen G6 and it repeats for a draw.

I'm just curious why the AI wouldn't go Queen to E7 or D8 and have the game continue. Yes, they'd lose a rook or knight, but given that they're in a much better position to win the game it surprises me that they just repeat the attack position on H4 and essentially giveaway the match. (Which is what the summary says at the end. I was getting beat pretty bad (from some early blunders) and then Black just gave it away to a draw for, essentially, no reason.)

I assume it's just silly AI seeing the attack as the best move even though it'll result in the game ending? From my perspective I'm repeating the move because I don't have a lot of other options. I need to press the attack and try to bring the game back towards my favor. Plus the draw is a more favorable outcome than a loss. But from Black's position retreating the Queen would have kept the game going and avoided the draw and probably resulted in a win.

Any reason a real human would do this from Black's perspective? Is it safer to take the draw result than to risk a loss, no matter how low the odds are?

To me, it look quite a bit as though you were playing against an incompetent computer opponent. Computers think differently than we do. A  computer opponent does what looks best to itself, whether or not it's actually the best. This means that when it played Qh4, and you played Qg3, it probably thought that, by its perspective, Qf6 was probably still the "best" move, even if, in the long term, it just leads to a draw. And yes, if a human didn't catch Qe7 or Qd8, they may very well try for perpetual check. It is safer to go for a draw, but the reward for avoiding it would, arguably, be much higher.

Yeah, that's what I suspected as well. Just the computer being silly and repeating the best move it could make. It's not the most incompetent computer by any stretch. It was a 1200, so kind of in the mid-range. Around the low end of Class D.

It's just an unfortunate side-effect of playing the computer I suppose. I have to consciously be aware that they'll likely repeat a position and try to avoid it myself. Just isn't as good to give up a favorable position, particularly when I'm losing in the overall scheme of things. Really wanted to keep pressing the Knight/Rook and use the split to take one.

Thanks for your input!

There's something off about this whole game. In the initial position, 1...Rg7 wins instantly, And after 1...Qh4 2.Qg3, 2...Rg7 still wins instantly

Not sure what you mean? Like it's a glitch or something? Because Rg7 definitely isn't a win... It's not even a check in either position lol

OscarWilde1854 wrote " Rg7 definitely isn't a win... It's not even a check in either position lol"

This is an excellent learning opportunity for you. Take your first diagram as an example: White's queen is on the same open file as his king. So when Black plays 1...Rg7, the White king is trapped. It can't move aside without exposing the king--which is illegal. The best White can do is to exchange his queen for the black rook. This is winning for Black

I think I'm just confused what you mean by "win" It's certainly a positive for black in these two frames, but it doesn't conclude the game. It's not a checkmate. It's definitely an effective pin and almost guarantees the black queen is gone. But again, not checkmate.

In the 2nd frame if black went Rg7 then white could go Ng6 and block the pin. Then black goes Qg3, and white goes Kg3 (or hxg3) and now both queens are gone. 

So yes, the first position Rg7 would have been a better move for black, but not necessarily a "win". Unless you're referring to it in a "win the battle not the war" type way. Where it wins the exchange but not the entire game?

And in the 2nd position it's still a better exchange of material for black, but loses a queen for both.

I can appreciate your confusion. The term "winning" is not always clear. In your first example, after 1...Rg7 pinning the queen, the best White has is to trade his rook for Black's queen, giving Black a large material advantage. This alone should constitute a winning advantage for Black. But there's more! Remember, Black was already a rook and a piece ahead. So after the smoke clears, White is down a queen and two pieces, with no compensation. Black has a winning advantage
In your second diagram, Black also gets a winning advantage after 1...Rg7, but it's much more complicated, and hard to calculate

My engine doesn't want to draw

Thank you for the insight! I'm definitely very new to chess (aside from playing occasionally as a kid) and trying to take it more seriously and learn. So I appreciate the feedback and clear explanation!

You are more than welcome! Sometimes we forget what it was like to be a relative beginner. If you have any puzzling questions, send me a message and I'll see if I can help

I guess when any other move has a negative evaluation, it goes for the repetition.

Yeah the bots are specifically programmed to not repeat moves, I just forced martin to do it because I was repeatedly checking him

no way it missed rg7