Computer Chess Question

[Tenorman]

The ad for this game mentions that "The board can be set up for a particular chess problem or situation". Does anyone know how to actually accomplish this? Basically, I have been playing it against other computer chess games and since it doesn't have any way to save a game, another way to do this might be to set the board up like it was before I had to turn the computer off and go from there. If anybody has a working link to a scan of the instructions for this game, that would be great too. I suppose I could just use an emulator and a save state, but where's the fun in that. Thanks.

[+orpheuswaking]

http://www.atarimania.com/game-atari-400-800-xl-xe-computer-chess_14030.html

 

Instructions from the UK release of Computer chess can be found at the above link

[Rybags]

An interesting experiment might be to see how smart the computer player really is at the top level.

IIRC a move takes something like 20 hours at the highest setting, but under emulation that could be sped up to something like 10-20 minutes.

 

I'd try it myself, except Chess for me is about as interesting as watching paint dry.

[+therealbountybob]

You might find this interesting, our relations at atarionline.pl battled Sargon III vs Chesmaster 2000

[Tenorman]

http://www.atarimania.com/game-atari-400-800-xl-xe-computer-chess_14030.html

 

Instructions from the UK release of Computer chess can be found at the above link

 

Thanks, I hadn't ran across that one. I must have skipped it because it was on tape, but I see that they are the same games. Setting the board up looks like kind of a pain, but it should be workable as I really hate to just leave 25 year old electronics on.

[Tempest]

Hmm... Now that would be interesting. Which classic system plays the best game of Chess? Almost every system had a Chess program, so you could have tons of systems involved (Atari 8-bit, 2600, Vic-20, C-64, Intellivision, O2 (with the module), Aquarius, etc.).

 

Tempest

[Rybags]

Fairly sure Atari's Chess is only 4K - so it's not exactly going to contain much in the way of super algorithms.

 

Probably why the top level moves takes 20 hours - rather than having much in the way of strategies, the computer just evaluates every single possibility a certain number of moves into the future depending on level.

[+orpheuswaking]

Yeah it's a quick load from tape. It was one of the first games I ever had for the Atari 800xl (It came as a pack in with my 800xl and 1010)

[Tenorman]

Fairly sure Atari's Chess is only 4K - so it's not exactly going to contain much in the way of super algorithms.

 

Probably why the top level moves takes 20 hours - rather than having much in the way of strategies, the computer just evaluates every single possibility a certain number of moves into the future depending on level.

 

I've been playing it against the OS X Chess game with the difficulty slider set to the middle. Computer Chess on the Atari just now seems to be putting up a halfway decent fight, but I had to set it to Level 6. It was beaten very easily on anything under that.

[+orpheuswaking]

I think that's why I liked playing it as a kid. It was right about my level of play (i.e. crap - lol)

[JohnBuell]

I was bored enough once to do Fidelity Ultimate Chess Challenge on the Lynx vs Chess on the Atari and the Lynx won VERY quickly....

[+thegoldenband]

Hmm... Now that would be interesting. Which classic system plays the best game of Chess? Almost every system had a Chess program, so you could have tons of systems involved (Atari 8-bit, 2600, Vic-20, C-64, Intellivision, O2 (with the module), Aquarius, etc.).

I'd love to see this! Doing it with real hardware could be extraordinarily time-consuming, but under emulation it could work really well. The VCS would potentially forfeit a lot of games -- doesn't it have a penchant for cheating, and/or an inability to understand en passant captures?

 

The Intellivision is surprisingly decent for its vintage, and on the higher levels it plays a pretty robust game, though it has a tendency to collapse unpredictably. Either way it sounds like they put a lot of effort into it -- according to the BSRs, it was the only Inty game ever to have extra onboard RAM.

 

I saw some footage of the two Aquarius chess programs on Youtube, and both seemed impressive. IIRC, one program went about 10 moves deep into a Sicilian Dragon variation before leaving "book" (departing from known opening moves). For a system of that age, that's pretty good.

[Tenorman]

I'm not sure how far I'm going to take this, as it is very time consuming, but I'll post any interesting results that I get. The thing to would probably be to figure out the best chess program on a few different platforms and then play them against each other. This guy has already played quite a few classic programs against each other, mostly for Commodore 64 but there is info on a few others, and posted tournament results if anyone is interested: http://www.spacious-mind.com/html/tournaments.html.

[Tempest]

Well we can start with systems that have a known chess program.

 

Keeping it classic (pre-85):

 

Atari 2600

Intellivision

Atari 8-bit

Aquarius

Odyssey 2 (with module)

Channel F (Saba cart)

Vic-20

C-64

TI-99/4a

Apple II

TRS-80

CoCo 1/2

 

Tempest

[+CharlieChaplin]

Yep,

but even on the A8 only there are a dozen or so of Chess programs:

 

- Chess by Atari

- (Odesta) Chess 7.0

- Mychess II

- Sargon 2

- Sargon 3

- Chessmaster 2000

- Colossus Chess 3

- Colossus Chess 4

- Masterchess

- Chess by Compute! (with 6 or 7 dif. levels)

- Chess mod./german (with 9 dif. levels)

- Querg Chess

- "Forth" Chess (or was it First Chess ?!?)

- Parker Chess

- if you like the opposite of chess, i.e. instead of keeping/defending your figures as long as possible, killing your figures as fast as possible, then there are Negation of Querg, Killer Chess and maybe a few others.

 

And I guess this list of chess programs for the A8 is still incomplete. Would be a long fight just for every A8 chess program to play against all other A8 chess programs (not to think of the dozens of chess programs of other 8Bit systems)... -Andreas Koch.

 

 

 

 

 

Well we can start with systems that have a known chess program.

 

Keeping it classic (pre-85):

 

Atari 2600

Intellivision

Atari 8-bit

Aquarius

Odyssey 2 (with module)

Channel F (Saba cart)

Vic-20

C-64

TI-99/4a

Apple II

TRS-80

CoCo 1/2

 

Tempest

[Divya16]

Hmm... Now that would be interesting. Which classic system plays the best game of Chess? Almost every system had a Chess program, so you could have tons of systems involved (Atari 8-bit, 2600, Vic-20, C-64, Intellivision, O2 (with the module), Aquarius, etc.).

I'd love to see this! Doing it with real hardware could be extraordinarily time-consuming, but under emulation it could work really well. The VCS would potentially forfeit a lot of games -- doesn't it have a penchant for cheating, and/or an inability to understand en passant captures?

 

The Intellivision is surprisingly decent for its vintage, and on the higher levels it plays a pretty robust game, though it has a tendency to collapse unpredictably. Either way it sounds like they put a lot of effort into it -- according to the BSRs, it was the only Inty game ever to have extra onboard RAM.

 

I saw some footage of the two Aquarius chess programs on Youtube, and both seemed impressive. IIRC, one program went about 10 moves deep into a Sicilian Dragon variation before leaving "book" (departing from known opening moves). For a system of that age, that's pretty good.

 

Ahm, the better chess program is more dependent on the programmer than the computer. If you were smarter than others and wrote a good algorithm, you can have an 8-bit Atari defeating a Pentium. So emulation or real doesn't matter. Checkmate.

[Tenorman]

Ahm, the better chess program is more dependent on the programmer than the computer. If you were smarter than others and wrote a good algorithm, you can have an 8-bit Atari defeating a Pentium. So emulation or real doesn't matter. Checkmate.

 

While that is true to a certain extent, the best algorithms rely heavily on brute force computation. More CPU speed and more memory mean the computer can look much further ahead, making it far better. The 8-bit Atari could win against a really crappy chess AI on a Pentium (probably like if I were to write one), but anything even half-way competent would be very tough.

[MaPa]

And I guess this list of chess programs for the A8 is still incomplete. Would be a long fight just for every A8 chess program to play against all other A8 chess programs (not to think of the dozens of chess programs of other 8Bit systems)... -Andreas Koch.

I think that the most of chess programs from your list can be left out just on experience of someone who played that games and knows that it has not high ELO rating and select only a few to test.

Edited October 14, 2010 by MaPa

[Divya16]

Ahm, the better chess program is more dependent on the programmer than the computer. If you were smarter than others and wrote a good algorithm, you can have an 8-bit Atari defeating a Pentium. So emulation or real doesn't matter. Checkmate.

 

While that is true to a certain extent, the best algorithms rely heavily on brute force computation. More CPU speed and more memory mean the computer can look much further ahead, making it far better. The 8-bit Atari could win against a really crappy chess AI on a Pentium (probably like if I were to write one), but anything even half-way competent would be very tough.

 

Just from example of Kasparov defeating some brute force computers specifically designed for chess shows that the best algorithm has yet to be written. Humans don't look ahead millions of moves per second; they only compute a few moves and the fact that a human can defeat a computer using brute force shows that the AI is crappy compared to real intelligence. Check. Mate in 1.

[+thegoldenband]

Just from example of Kasparov defeating some brute force computers specifically designed for chess shows that the best algorithm has yet to be written. Humans don't look ahead millions of moves per second; they only compute a few moves and the fact that a human can defeat a computer using brute force shows that the AI is crappy compared to real intelligence. Check. Mate in 1.

Unfortunately, your information seems to be out of date: for several years now, the strongest programs like Rybka have been equal to or better than the strongest humans. It's not that the computers win every single game, but their ability to calculate short-term tactics is now far beyond any human's, and their strategic play (long-term planning) has vastly improved. The computers can even play at odds -- giving up a pawn before the game, or a rook for a knight -- and still beat leading grandmasters with little difficulty.

 

You simply don't see the kind of blind spots that were commonplace 10-15 years ago, where a clever human could still exploit the "horizon effect" to launch a winning attack before the computer could see it coming. And it's not just brute force -- the algorithms have improved massively, and while the scenario of an Atari defeating a modern CPU is unlikely, the Wikipedia article on computer chess points out that a recent program that runs on mobile phones is stronger than the Deep Blue supercomputer, which was able to search 10,000 times as fast (!!).

 

By the way, Kasparov hasn't won a match against a computer since 1996. He lost a famous match in 1997 against Deep Blue when he was the reigning world champion; since then he's had a couple of drawn matches, I think, though the computers in question may well have been weaker than Deep Blue. In fact, I don't think any world champion has won a match against a computer since then -- Kramnik was crushed by Deep Fritz a few years ago.

 

The good news is that computers still haven't figured out go. Then again, neither have I!

Edited October 14, 2010 by thegoldenband

[Divya16]

Just from example of Kasparov defeating some brute force computers specifically designed for chess shows that the best algorithm has yet to be written. Humans don't look ahead millions of moves per second; they only compute a few moves and the fact that a human can defeat a computer using brute force shows that the AI is crappy compared to real intelligence. Check. Mate in 1.

Unfortunately, your information seems to be out of date: for several years now, the strongest programs like Rybka have been equal to or better than the strongest humans. It's not that the computers win every single game,

You can stop right there. You are AGREEING that computers don't win every game thus you are NOT even addressing my argument regarding the algorithm. You were already checkmated and rather than address the argument you became a soar loser with your misinformation.

 

You simply don't see the kind of blind spots that were commonplace 10-15 years ago, where a clever human could still exploit the "horizon effect" to launch a winning attack before the computer could see it coming. And it's not just brute force -- the algorithms have improved massively, and while the scenario of an Atari defeating a modern CPU is unlikely, the Wikipedia article on computer chess points out that a recent program that runs on mobile phones is stronger than the Deep Blue supercomputer, which was able to search 10,000 times as fast (!!).

 

You may be blind then because Tenorman made argument regarding brute force algorithms. And your source is incorrect. Again you are beating around the bush with 10,000X or whatever since the argument is that human doesn't use algorithm of computing millions of moves/second. So let Tenorman answer.

 

By the way, Kasparov hasn't won a match against a computer since 1996. He lost a famous match in 1997 against Deep Blue when he was the reigning world champion; since then he's had a couple of drawn matches, I think, though the computers in question may well have been weaker than Deep Blue. In fact, I don't think any world champion has won a match against a computer since then -- Kramnik was crushed by Deep Fritz a few years ago.

You are misinformed: http://en.wikipedia.org/wiki/Garry_Kasparov

 

The good news is that computers still haven't figured out go. Then again, neither have I!

 

Nor chess like humans. The argument is not whether they can win and lose against a human if you read properly but about the algorithm.

[atariksi]

The good news is that computers still haven't figured out go. Then again, neither have I!

 

Computers don't have it figured out when you go brute force-- basically examining every combination you can think of-- and still screw up every other time or so. Take for example a sorting algorithm that runs on the order of O(n * lg(n)). If an Atari computer w/memory expansion were to sort an array of 1 megabyte using Quicksort, the runtime is 1048576*20*C whereas using brute force you have 1048576*1048576*C so that's 52,000 times slower but given a fast enough parallel processing computers linked up, you could eventually beat the 8-bit but it's still an inferior algorithm and you wouldn't say it's all figured out since the 8-bit would do it using less calculations. And as you increase the data set, the 8-bit would still win.

 

And you not having figured it out has nothing to do with it. People need to be trained for certain tasks.

[Rybags]

The other thing is that you need to weight each loss and gain in a lookahead algorithm.

 

You might have a situation where you lose a third of your pieces in several moves, but can be put in a winning situation, but some algorithms might miss it entirely because they quit the scan if a bad run starts.

 

Additionally you have some chess programs that have a bunch of "classic starts" programmed in - by allowing them to use them would be unfair if against a program that relies only on it's own merits.

[+orpheuswaking]

Damn, there are some serious Chess people in here... This is turning into a Atari Jaguar Thread!

[devwebcl]

I have ran Colossus Chess 3 against 4: http://manillismo.blogspot.com/2009/04/colossus-tournament.html

and Colossus Chess 4 against GNU Chess: http://manillismo.blogspot.com/2009/04/colossus-chess-40-a8-vs-gnu-chess.html

Only with an emulator.

 

Also there are third party attempts for 2600: http://manillismo.blogspot.com/2010/09/competencias-entre-consolas-sobre.html

 

--Devwebcl