[Ghostavo]

Though you could have used a supercomputer to generate different games to cover most possibilities. If I was writing a chess game, I'd presimulate every game on every level so to speed it up overall.

 

I hope you have a very large hard-drive, because just a complete 6 piece endgame tablebase would occupy over 1 TB. A game of chess has 32 pieces, each addicional piece increases the size of the tablebase exponencially, so that you have any idea of the size increase with a piece added, a 5 piece tablebase occupies a bit over 6 GB and a 4 piece tablebase occupies 30 MB. And that's dismissing the fact that you'd have to  calculate every position with the freaking 32 pieces. So you have any idea how dificult it is, I just have to say that complete 7 piece endgame tablebases do not exist and probably will only be completed beyond 2010.

[Bob-san]

Or just have it plan 2-3 moves in advance... anyways--whatever really. The processor takes the strain.

[Ghostavo]

My brother tried fiddling around with an Atari 2600 version of chess on GameTap a while back.  He's no grandmaster, but he's at least decent in the game, so he figured he'd have a good shot of winning.  He must have had the game mode set on the hardest difficulty level, because that damn ancient AI obliterated him every time he played.  The most amusing part for the non-chess-playing me was watching the game screen flash psychedelic rainbow colors every time the computer "thought" about its next move.  Who knew those cartridges were so advanced?

I don't want to be insulting, but the hardest setting in a chess playing game from a console from the late 1970's shouldn't be difficult. Most likely your brother wasn't as good as he thought, or he was having a really bad day.

Well, I never said he was good good, but it at least surprised him that he wasn't able to make any headway against the program.  I'll leave it to someone who actually understands chess and happened to play the thing themselves to determine how objectively difficult it really was.

That's the thing, you don't have to be a good good player to beat even a fairly good chess game from that time. For example, Kaissa was an average engine at that time, it had an ELO rating below 1700. In 1980, Belle had a performance around 2250, master level, but just with one problem, besides being by far the strongest computer in the world at that time, it used dedicated hardware. From what I gather, Chessmaster 2100, one of the strongest engines available for the Amiga, had an ELO rating below 1800 (this take from the fact that Chessmaster 3000 had an ELO rating of 1800).

The average (club) player have an ELO rating around 1800. So to stand equal to that chess engine you'd have to... average. To be slaughtered...

Damn, I never knew it was that hard to get ELO ratings of engines from that long ago. 

P.S.
I have no idea how I confused the Atari 2600 for the Amiga, but I assume the point still stands. 

[Admiral_Stones]

My brother tried fiddling around with an Atari 2600 version of chess on GameTap a while back.  He's no grandmaster, but he's at least decent in the game, so he figured he'd have a good shot of winning.  He must have had the game mode set on the hardest difficulty level, because that damn ancient AI obliterated him every time he played.  The most amusing part for the non-chess-playing me was watching the game screen flash psychedelic rainbow colors every time the computer "thought" about its next move.  Who knew those cartridges were so advanced?

I don't want to be insulting, but the hardest setting in a chess playing game from a console from the late 1970's shouldn't be difficult. Most likely your brother wasn't as good as he thought, or he was having a really bad day.

Well, I never said he was good good, but it at least surprised him that he wasn't able to make any headway against the program.  I'll leave it to someone who actually understands chess and happened to play the thing themselves to determine how objectively difficult it really was.

That's the thing, you don't have to be a good good player to beat even a fairly good chess game from that time. For example, Kaissa was an average engine at that time, it had an ELO rating below 1700. In 1980, Belle had a performance around 2250, master level, but just with one problem, besides being by far the strongest computer in the world at that time, it used dedicated hardware. From what I gather, Chessmaster 2100, one of the strongest engines available for the Amiga, had an ELO rating below 1800 (this take from the fact that Chessmaster 3000 had an ELO rating of 1800).

The average (club) player have an ELO rating around 1800. So to stand equal to that chess engine you'd have to... average. To be slaughtered...

Damn, I never knew it was that hard to get ELO ratings of engines from that long ago. 

P.S.
I have no idea how I confused the Atari 2600 for the Amiga, but I assume the point still stands. 

Hydra has an Elo (yes, Elo, not ELO) rating of over 3000. It just massively pawns everyone.

[Ghostavo]

But Hydra uses modern algorithms and modern (dedicated) hardware to accomplish that.

And it can be written as Elo or ELO.

Oh and it still isn't enough not to get owned in correspondence chess against human players.

[Mongoose]

I've never heard of the term "ELO rating" before now, and I don't think my brother has ever played in a club, so I wouldn't have any idea how he stacks up with the average player.  And maybe "slaughtered" was a generous term; all I really know is that he failed to beat the thing after trying several times.  Honestly, I would have been impressed if a 2600 game managed to outplay a 5-year-old, much less someone who knows at least a small amount of strategy.

[DarthWang]

I can beat chessmaster 8000 using only 5 takebacks in the whole game

[Admiral_Stones]

But Hydra uses modern algorithms and modern (dedicated) hardware to accomplish that.

And it can be written as Elo or ELO.

Oh and it still isn't enough not to get owned in correspondence chess against human players.

Kasparov had a rating slightly over 2800. So yes, it IS pretty enough to pawn everything, except if it's some strange autisctic hyper-intelligent child from outer space.

[Ghostavo]

Yes, but that's over the board chess. Hydra loses in correspondence chess. Correspondence GM Arno Nickel for example played a 2 game match with Hydra, result? 2-0 in favor of the human.

And even then, most chess engine creators are suspicious of Hydra's true strength since it's only indication was it's match with Michael Adams. Rybka for example plays regularly and has an elo around 3000 in most computer chess ratings lists and it's creator offered to play against Hydra, IIRC, and it's still waiting for a response.

Finally, I feel that I should add that although the rating is supposed to represent the strength of the player, it doesn't stop players with lower rating from winning.