Nightmare on LSS

      This interesting position recently came up in one of my cc games and I spent a lot of time trying to figure out (as White on the move) what I should play next.  

      My database had three games in it with this position with the results of +1 -0 =2 so that wasn’t much help.  The continuations were 23.Nxh7, 23.e5 and 23.Qg4.  Houdini suggested 23.e5 and my other “go to” engine, Fire 1.5 xTreme liked 23.Qg4. Still, I was fascinated by 23.Nxh7 and spent a great deal of time examining it especially because in my db of correspondence games by top-level players white had won. I reasoned that 1) white was a strong cc player and 2) he surely had carefully analyzed the position before playing 23.Nxh7 and 3) the engines evaluated the resulting positions at nearly 0.00.  So, not being able to find anything wrong with the move, I decided to go ahead and play it and the game continued 23.Nxh7 Kxh7 24.f6 g6 25.e5 Kg8 27.Bxg6 fxg6 28.f7+ Kg7 29.Bxd8 Rxd8 30.Qd2 Rf8 31.Rae1 reaching the following position which I realized was critical.

      In the game in my db Black continued 31…Bd4+ and eventually lost.  However, Houdini was suggesting 31…b4 as the best move and evaluating it at about 0.20 P’s in Black’s favor, so I decided to spend some time examining its suggestion. I ended up not liking what I was seeing; White always seemed to be slowly but surely drifting into slightly inferior positions!  Eventually after much analysis on my part we reached this position:

      Black has just played 54…Kxh5 and the engines are evaluating his position as about 3.5 P’s in Black’s favor but I realized that if I can keep checking the game will be drawn by the 50-move rule.  However Black discovered a plan where he boldly played his K over to the Q-side and captured my b-P arriving at the following position:

      Black has just played 75…Kxb2 so now I have to keep checking until move 125 and not allow his b-P to move.  Can it be done?  The answer is, “No.”  If you remove Black’s b-Pawn and setup the position in the Shredder Endgame Database, Black wins in 34 moves and if my math is correct, that puts us at move 122.  Still, all is not lost because if Black relies totally on engine moves to make his selection, it’s highly unlikely that any engine will recognize this and will possibly suggest less than optimal moves…or so I reasoned.  I ran several shootouts from this position and found some positions where black was threatening mate on the next move and white was compelled to capture the black B which would start the count all over again.  But…this situation always occurred slightly the other side of move 122.  Of course Black could probably find an improvement and save a couple of moves.  It was going to be very close!  Here’s the position after Black’s 88…Ke4:

      Shootouts were suggesting that Black was going to mate within the 50 move rule, but I’m still going to hang in there and play 89.Qe2+ and see what happens.  So what did happen? I know you’re dying to find out.

       I made a mouse slip on the server and played 89.Qd2 and resigned after 89…b2.  Disgust can’t describe my feelings.  How does one make such a stupid mistake on a server where you have to verify your move by answering the question “Do you really want to pay 89.Qe2+”  I don’t know, but this was a very painful post.