Forums

Engines have no idea how to play chess ...

Sort:
gspaulsson

Maybe the moral is that if an engine finds a refutation, then refutes its own refutation, then changes its "mind" at depth 30, a human at any level would have trouble with it. I remember a blitz game that I played 50 years ago against the top player in our club, whose ELO was at least 500 points higher than mine. I inadvertently offered a queen "sacrifice" (i.e., didn't notice it would be hanging), and he was befuddled by it long enough to lose on time.  A successful sac (sound, unsound or insane) is one that sets your human opponent a problem that he fails to solve.     

Cherub_Enjel

It is relative, but setting the standard too low is delusional.

It should be at least 2200-2300, master level, where sacs are judged based on ability to respond. Anything is good against a 400 player for example.

gspaulsson

Don't be snooty. My opponent played pretty well for an 1176 player: he outplayed me in the opening, and I give him credit for it. Neither of us is a master, so there were errors and inaccuracies on both sides - actually, I made more of them than he did, according to the engine. But positions offer problems, regardless of how they were arrived at, and this game has a few good ones. 

The first key moment in the game is 19...Ne5, which the engine "thought" was unsound and gave a refutation; but when I try out the refutation and let it "think" some more, it changes its "mind". In other words, Ne5 would have been a good sac against a 2200-2300 player as well. A small victory for human intuition.

The second key moment is 27... Bc8, which sets a good problem for a player at his level. He did resist the temptation to grab the Rc6, but he didn't find the refutation, Rh1.

The position after 28.g4 is a good problem too. 28... Bxg4, the move I was setting up, offers another piece while leaving the rook hanging. To see the necessity of Rh1, he would have had to anticipate Bxg4 - easy when it's set up as a problem; not so easy OTB.

Apparently, Bxg4 leads to mate in 15: the challenge is to find the move and the mate.

The position after 29.Rh1 (too late) sets another good problem: black has to resist the temptation to grab white's queen and find the mate in 3. Again not hard if set as a problem, but easy to miss OTB, and I did miss it. 

Finally, I was looking for comments on my analysis (of both players' moves), not a pat on the back. 

Cherub_Enjel

My comment was just my opinion, nothing personal. The ...Ne5 move is playable imo. Against an 1100 it's especially good since most 1100s would just take it. 

DiogenesDue

Your subject line is basically fishing for "snooty" replies, so why act surprised when you get them?

tondeaf

you are a fool

HorsesGalore

what time limit was this game ?   I would not have rushed 19....Ne5  as White seems OK to take that piece and defend his K-side with Nf1 and if needed  g3.   I can't see how that is a sound sac as Black does not have more pieces on the K-side and it will take time to manuever others over.   I would have continued to get more compensation for your pawn minus with 19 ....R-c8

 

I believe your Rating intimidated your opponent and he gave little thought to accepting your sacrifices.   often low Rateds can play like that -- respecting who you are.....even though they might not see the reason why you played something.   

 

Once  Emanuel Lasker's opponent took over a half hour on a piece sacrifice,  Lasker ignored it and did not take long on his move.   When asked afterwards, Lasker said his opponent is a master and did not try to see if it was unsound !   He Just continued playing solidly, had a big time edge and gradually won.

Cherub_Enjel

Well, in my experience, most 1100s blunder immediately when under just a little pressure (and they blunder reasonably often without it even) or threat, so that's why I thought ...Ne5 would be playable against an 1100. Obviously against a player of equal level the OP would probably have gotten the sacrifice refuted.

president_max

how to prevent snooty replies

null

lofina_eidel_ismail

😊 so cute😉

Cherub_Enjel

Black can just move out of the pin, with Qe7, and white has two hanging pieces. 

Cherub_Enjel

Well, the move ...Qe7 I offered is just an idea - you can play ...Qb8 or something, to avoid the tricks. The point is that when you have such hanging knights that can't move, and the only thing stopping it is a pin, and you can break the pin with 5 different moves, something is very off - and that's why I just said "move the queen".

It's a bad sacrifice that isn't as bad against an 1100. 

Cherub_Enjel

Well, if you want an engines perspective, all of blacks moves after dxe5 are flat out losing.