Can Houdini 3 be beaten?

I got my lessons through the mail. My lessons consisted of me asking Lev to analyze whatever opening I found interesting. Most of his analysis was of the Evans Gambit, 4.Ng5 in the Two Knights' Defense and the King's Gambit (1.e4 e5 2.f4 exf4 3.Qe2). We never had face to face lessons.

I just gave you four games---two were published in Chess Life and a third was published in Inside Chess. That is empirical data. See post 182

Here is a quote from GM Alburt, "I've been consulting Richard Moody on various chess openings for about 25 years. Today, Richard is an unrelenting opening researcher, well skilled in using all the tools of the trade---books, data bases, Fritz. His book is another  proof that one doesn't have to be a Master to make important discoveries. And making your own opening theory is one of the surest paths to success in chess."

Quoted in what publication?  I'd like to look it up.  And my previous post was that you had no empirical proof of actually playing chess against Lev Alburt.

For 25 years Lev saw me as his student; for the vast majority of that time I was a weak player and theoretician, so it is not surprising that Lev is skeptical; I'll post his comments about the Maroczy Bind when I get them.

It takes Lev (and post members) a great deal of proof that I can play consistently at a high level. Roman, on the other hand, has only seen me play at a high level (we only started corresponding a couple of months ago, so he has no  difficulty believing that I am playing much beyond Class B level.) 

I am working with Senior Life Master Russ Potter and will provide his analysis of some of my games. 

The only reason I know this is true is that I can sometimes outplay Houdini 3 in complex middlegames. Did you see my game against Houdini where I shepherded an advantage of +/= to +- over the span of 20 moves? Houdini was out to a depth of 22 for most moves and failed to see many of the key moves.

In a purely developmental attack (as I define development), I don't calculate, I just count moves i.e. I just focus on making the maximum number of developing moves in a row that I can make. I know where my pieces and pawns belong and just place them on those squares, and,  when my game is on, the computer makes no attempt to stop me.

Perhaps you can use a different term than developing moves?

To any other chessplayer there's a finite number of developing moves you can make, 21 sounds ridiculous and it doesn't really mean anything to anyone else.

Perhaps it would be better to distinguish developing moves from "useful" moves. Here is my latest draw, just relying on my way of looking at development. I think that 7.Bf1, book, is suspect i.e. it is an undeveloping move whose sole purpose is to win material. This is suspect because after 7.Bf1 Nxd5 8.Ne4 Ne6 =  This is not surprising in view of the strong outposts occupied by the Knights which gives full compensation for the pawn, so 7.cxd4 gaining time is the preferred move when just considering development.

Here is a way to annotate the game in view of development/useful moves: (U stands for useful move; +1 is a tempo gained either developing a piece or capturing a developed piece. 0 if for moving a piece to another square it could occupy in one move e.g. Qe2 +1, but then playing Qf3 one move later is worth 0.

By contrast the move sequence Qe2/Qe3 is worth +2 because the Queen needs two tempos to access the e3 square. It should be apparent that if you capture a piece that has moved multiple times, you gain all the time it took for the captured piece to get to that square. By contrast if you use a developed piece to capture an undeveloped piece that is worth 0.

1. e4 (U) e5 (U) 2. Nf3 (+1) Nc6 (+1) 3. Bc4 (+1) Nf6 (+1) 4. Ng5 (+1) d5 (U) 5. exd5 (U) Nd4 (+1) 6. c3 (U) b5 (U) 7. cxd4 [+1---Book is Bf1 and it is not surprising that Black probably is equal because Bf1 is worth -1, so after 7...Nxd5 (+1) 8.Ne4 (+1) Ne6 (+1 =)] bxc4 (+1)  8.dxe5 (U) Qxd5 (+1) 9. O-O (+1) Bb7 +1)10. Nf3 (-1 [Qf3 is +1, but White needs to keep the Queens on the board as after Qxf3 11.Nxf3 Nd5, Black has no problems] Nd7 (+1) 11. Nc3 (+1) Qd3 (0---because from the starting position the Queen can reach d3 in one move) 12. Re1 (+1) O-O-O (+1) 13. Re3 (+1) Qg6 (+1) 14. e6 (U) fxe6 (U)15. d4 (U) Bd6 (+1)16. Qe2 (+1) Rhf8 (+1) 17. Nh4 (+1) Qf6 (-1) 18. Nf3 (-1) Qg6 (+1)19. Nh4 (+1)Qf6 (-1) 20. Nf3 (-1)  Qg6 (+1) 21. Nh4 (+1) Qf6 * 1/2-1/2

When we add up development, useful moves, material and pawn structure we get.

White  Development +8    Useful moves +6    Knight for Bishop   3 pawn  islands.

Black Development +8   Useful moves +4  Bishop for Knight 5 pawn islands. But since neither side can deviate without taking risks the draw is the best result.

Why listen to a B class player ?  

pull out the plug GG

sloughterechess, almost every schoolboy knows by now that Black stands very badly after 1.e4 e5 2.Nf3 Nc6 3.Bc4 Nf6 4.Ng5 d5 5.ed5 Nd4 6.c3 b5 7.Bf1 Nxd5 8.cd4 Qxg5 9.Bxb5+ Kd8 10.0-0.

You either don't know that, or you clumsily try hiding it under the carpet. It won't work though.

Hi Pfren!

Good to see you joining the conversation. We can agree to disagree. Yes White can stay a pawn up but I have played a dozen or more games against Houdini in this variation. Here is my latest attempt. Look how Houdini uses a slight initiative for the pawn to eliminate a pawn island leading to a dead drawn position a pawn down. I let Houdini crunch out to a depth of 28 in the final position. It could see no way for White to make progress.

1. e4 e5 2. Nf3 Nc6 3. Bc4 Nf6 4. Ng5 d5 5. exd5 Nd4 6. c3 b5 7. Bf1 Nxd5 8.cxd4 Qxg5 9. Bxb5+ Kd8 10. O-O Rb8 11. Bc6 exd4 12. d3 Qf5 13. Re1 Be6 14. Nd2 Qxd3 15. Nb3 Qxd1 16. Rxd1 Kc8 17. Nxd4 Bd6 18. Nxe6 fxe6 19. Bxd5 exd5 20.Rxd5 Re8 21. Kf1 Be5 22. Rb1 Bc3! 23. Rd1 a5! 24. b3 a4 25. Be3 axb3 26. Rxb3 Rxb3 27. axb3 Bf6 28.Rd5 Re4 29.Ke2 Rb4 30.Rd3

This game provides a case for changing the rules of chess. From my experience in simple endgames with no special circumstances, Houdini can hold most pawn down endings. Maybe we should redefine draws to escape "draw death". What do you think that we should say that the player with the last remaining piece or pawn is the winner? If a player knew that trading down into a pawn down ending was a dead loss, this would provoke much greater aggression in the middlegame.

"This game provides a case for changing the rules of chess"

Or perhaps playing against people.

Computers have earned the nickname the "beast" for their superb handling of middlegames. Humanity has lost some of its humanity when computers were beating humans. What if we tweak the rules so that humans could, once again, beat computers? We have gone out of our way to skew the rules to benefit computers. Isn't it about time we right rules to benefit humans over computers?

GM Alburt has suggested that we return to the "old" rules i.e. 120 or 150'40 followed by adjournment. Wouldn't it be uplifting to not just chess players but to humanity as a whole if humans could regain supremacy in chess over computers?

What if we eliminate time delay and have as an iron clad rule that no computer ever again can make a move in less than one second? Since computers have unlimited access to opening and endgame theory, don't have to move their pieces so they never make a touch move violation, this one rules change would level the playing field.

In simple Rook and pawn endings, the human player can win by making a hundred mindless moves to run the computer out of time.

If this one rules change were implemented and we follow the sage advice of GM Alburt, humans could again beat computers in chess. Isn't that a worthwhile goal?

No, seems pretty irrelevant to me.

Putting in limitations on computers so humans can beat them is worthwhile?

Houdini cheats when I play it in blitz chess. In a five minute game it will play book for 10-12 moves and not take even one second off its clock. Why is leveling the play field a limitation? He who writes the rules wins. Is it a limitation, for example, for GM Judit Polgar and World Champion Garry Kasparov play Houdini 3 at a time limit of 120'40 followed by adjournment? If we change no other rules I predict that in consultation games humans could beat computers because they will make far fewer simple tactical oversights, the bane of human contestants.

You're limiting it. If you are going to limit the software, you're not playing the best computer, so humans beating it doesn't count for much. Why not play against Chess Titans?

And why does it even matter if humans can beat computers? There are few that put as much emphasis on them as you.

Pfren,

It is far more dangerous for Black to keep the pawn:

1. e4 e5 2. Nf3 Nc6 3. Bc4 Nf6 4. Ng5 d5 5. exd5 Nd4 6. c3 b5 7. Bf1 Nxd5 8.cxd4 Qxg5 9. Bxb5+ Kd8 10. O-O Rb8 11. Bc6 exd4 12. d3 Qf5 13. Re1 Nf6?! 14. Be4 Nxe4 15. dxe4 Qd7 16. Nd2 f6 17. Nf3 d3 18. a3 Ba6 19. Be3 Rb7 20. e5 Qd5 21.b4 Kc8 22. Bd4 Be7 23. Rc1 Rd8 24. Qd2 Bb5 25. Bc5 c6 26. h3 h5 27. e6

There is far more play here for White than there is after 13...Be6. According to my theories this is simply because 13...Nf6 is an undeveloping move based on tactics to hold the pawn whereas 13...Be6 is a developing move, sacrificing a pawn to reach a drawish endgame a pawn down. By some "strange  coincidence" Black is able to eliminate a pawn island and arrive at at book draw while 13...Nf6 leads to a complex position where White has better practical chances to win.

I think you can beat Houdini 3, have you tried to run it in a 386 or a 486?