Game with Ani

So look over the above games...its not always clear to me why black is giving up its pawn by moving its bishop off the diagonal.  Looks like a blunder to me?  Wouldn't play have just continued drawishly if black had not done that?  What am I not seeing.

In the revised position, the results are the same - wins for white.  

I think when I ran it previously and was seeing draws I had the hash memory allocation messed up (the engine settings for analysis mode and for competition mode are different and I had assumed they carried over).  With memory correctly allocated, the results look good for white.

So I think it is safe to say its a win for white if white plays correctly, although I am still confused why black is giving up its pawn in some of the games above and need to look into that.  

Also, look at the length of the games.  50 to 60 moves in the end game.  Not sure how easy it would be to execute in a real game on the clock.  But I think if white did get it down to N+B v K then black needs to concede at that point that it is then a matter of white slogging through the right technique.  How easy that is and if they could do it within 50 moves before the 50 move rule applied is another story 

But I think you can put that game in your win column 

Funny how some engines opted to Queen. As if they don't know how to mate with Bishop and Knight. Thanks for this. Very interesting.

Did you see black giving up its pawn in a lot of those games?  Any thoughts on that?  I have not had a chance to look deeper into it.

From the initial position, after Kf6, I let my engine run to a depth of 31 plies and have the following analysis:

8/8/5kB1/7P/3b4/3Np3/4K3/8 w - - 0 1

Analysis by Houdini 2.0c x64:

1. +-  (3.50): 1.Be4 Kg5 2.Bf3 Kf5 3.Nb4 Kg5 4.Nd5 Bc5 5.Kd3 Kf5 6.Nc3 Ba7 7.Ne4 Bb6 8.Be2 Ba7 9.Bd1 Bb6 10.Bf3 Ba7 11.Be2 Bb6 12.Bd1 Ba7 13.Bf3 Bb6