116 ( +1 | -1 )
Nemeth Gambit - Computer Destroyer!
It goes like...
1.e4 c5 2.Na3 Nc6 3.d4 cxd4 4.Bc4 Nf6 5.h3 Nxe4 6.Bxf7+ Kxf7 7.Qh5+ Ke6 8.Qg4+ Kd5 9.c4+ dxc3 10.Be3 ...Ne5 11.O-O-O+ Nd2 12.Qe2 Qa5 13.Bxd2 cxd2+ 14.Rxd2+ Ke6 15.f4 Kf6 16.fxe5+ Qxe5 17.Qf2+ Kg6 18.Nf3 Qf6 19.Nb5 Qf4 20.g3 Qc4+ 21.Rc2 Qxb5 22.Nh4+ Kh6 23.Rc5 Qxc5+ 24.Qxc5 e6 25.Qe3+ g5 26.Rf1 Bg7 27.Nf5+ Kg6 28.Nxg7
It usually is always the same till move 10 I believe!?... but from there the wandering King is in trouble!
Trying it against my computer at a high level it did more or less the same moves!.. but didn't play ball with 10. ...Ne5, I think it played 10. ...e5 I still manage to beat it, by throwing a lot of material at it to keep pressure on the King.... only just though!
a) How did Nemeth find this?... did he stumble upon it by chance?
b) Why do lots of computers get into the same trouble?... is the gambit so bad that the moves the computer makes are easily the best ( it thinks )?
c) Are there any other openings like this that do roughly the same sort of thing ( tempting the King to the centre of the board through materialism )
I'd love to hear from somebody that knows more about how this gambit works. And if anybody has taken the gambit to absurdly hard levels of play against the strongest computer?... Post your games please.
56 ( +1 | -1 )
What computer software are you using? Fritz 8 doesn't take the obvious epawn on move five. It likes e6 and claims black is 1.66 pawns advantage. Also, if you tell Fritz to take pawn on e4, on move 8 with your queen check, it merely drops back to g8 and claims a 1.94 pawn advantage. I wouldn't take that e pawn and I'm not nearly as good as Fritz. I guess the knight on a3 is supposed to tempt black into taking e pawn, but if you don't, it sits there in limbo. Seems like the Nemeth Gambit is justifiably obscure.
160 ( +1 | -1 )
What time controls are you playing on, spurtus?
When I try it out in rapid games (Game/30), my engines do the same silly king march. However, looking closer at the engine analysis, I notice that at this time control, it's really only calculating out to 12 ply or so. So what's going on is that the engine is just searching for a short time (~60 seconds), seeing that there's no disaster or immediate loss in 5 moves, and playing the materialistic move. Admittedly, my machine isn't very fast.
If I extend the time control so that the engine searches for >120 seconds per move, it starts to play reasonable moves. The engines still want to play the greedy 5... Nxe4, but instead of marching the king out, they play 7... Kg8 intending to meet 8. Qd5+ e6 9. Qxe4 by 9... Bb4+, with ideas of ...d5 and ...Qb6 (or ...Qg5 if some form of Bd2 Bxd2 occurs) and a reasonable attack.
Computers have rather crude evaluation functions (by human standards), and they possess no real concept of intuition. So the main way they perform accurate analysis is through in-depth brute force searching Take away this in-depth search by taking away the engine's calculation time, and it's not surprising that the engine doesn't generate particularly good moves.
The most famous "drawing the king into the center with a material sacrifice" idea I can think of is 1. e4 e5 2. Nf3 Nc6 3. Bc4 Nf6 4. Ng5 d5 5. exd5 Nxd5?! 6. d4 intending Nxf7 Kxf7, Qf3+ Ke6. It might be hard to get a computer to play this unless you turn off its opening book, though; a lot of computer engines I've seen are set up to play 5... Na5 which gives Black a decent game.
70 ( +1 | -1 )
I was actually using one of my weaker chess engines Chessmaster on the gameboy!... level 7. If you have got this program then you will probably get the same results as I did. Time controls dont come into it, no idea how deep the ply it is looking. ( Chessmaster on Gameboy is very easy to beat!... mind you it took me years to get past level 5, now I beat it at level 5 all the time, and have managed a win at level 11! )
But I was really surprised that the computer played the same to move 10 as in this gambit.... made me wonder if it was perhaps a universal anti-computer opening gambit?
No doubt Nemeth had the other moves you suggest covered as part of his silicon crunching opening system?
89 ( +1 | -1 )
"But I was really surprised that the computer played the same to move 10 as in this gambit.... made me wonder if it was perhaps a universal anti-computer opening gambit?"
Nothing special is happening; this is just the horizon effect. If a move has consequences beyond the computer's ability to calculate, then to the machine, these longer-term consequences simply don't exist at all. So if you let the machine calculate long enough to reach a decent number of ply, it won't play the bad king march and as White you'll simply be busted. At least, 7... Kg8 is convincing enough for me that Black is better; I don't see what White has going for him.
"No doubt Nemeth had the other moves you suggest covered as part of his silicon crunching opening system?"
You tell me. I don't know anything about this "system", if indeed it is one. I'm just analyzing variations. I don't see that White has all that much compensation for his missing pawn, though.
37 ( +1 | -1 )
In the early days ...
The first chess computers would fall for this:
1. e4 e5 2. Nf3 Bc5 3. Nxe5 Bxf2+ 4. Kxf2 Qh4+. Now the computer, only looking a few -ply ahead, sees that Kg1 would drop the e-pawn, so 5. Ke3? Qg5+. This time the computer wants to save the N, so 6. Kd4? then c5+ 7. Kd5? (defending the N again) and then d6 wins the N, and soon the game, for Black.
I guess the Nemeth Gambit is a more sophisticated version.
73 ( +1 | -1 )
Look at this
Tak in consideration that "Nemeth used a P600 computer with 256 Mb RAM, and 64-128 Mb hash tables" -- not the fastest machine nowdays, but not a slow one too.
There are a lot of ways to destroy a computer. Mainly they are reached through trial and error method in sharp gambits, where computer takes material and is no capable of calculating all the complications, but as it was discussed in other thread here on GK - computers tend to repeat their mistakes, cause theri calculations in the same positions are the same (or almost the same). I know players who have found such winning variations in the Latvian Gambit.