An intriguing question....

General Discussion about the game of Checkers.

An intriguing question....

Postby CheckersStrongplayer on Mon Dec 19, 2005 9:44 pm

I was wondering who won between Alex Moi--- The russian guy and Nemesis? It says in the Nemesis site that Alex reconmends it... But i've wondered, how did he do against Nemesis on those matches? Have any of you ever thought of this?
Checkers is like being in heavens.
User avatar
CheckersStrongplayer
 
Posts: 92
Joined: Mon Nov 14, 2005 8:06 pm
Location: Houston,tx

Re: An intriguing question....

Postby matthewkooshad on Mon Dec 19, 2005 10:26 pm

From 2002, Alex vs Wyllie program http://nemesis.info/2002%20Moiseyev-Wyllie.htm
From 2002, Nemesis vs Wyllie program http://nemesis.info/2002%20Nemesis-Wyllie.htm
A viewing of these games might be enjoyable to you.
User avatar
matthewkooshad
 
Posts: 289
Joined: Tue Nov 15, 2005 3:08 pm
Location: Mississippi, USA

Re: An intriguing question....

Postby Alex_Moiseyev on Tue Dec 20, 2005 12:02 am

Prime time for human is over :roll: Marion Tinsley was lucky to play his historical match with Chinook just right on time, in 1992.

One time I asked professor Jonathan Schaeffer, Chinook author, and he told me, that he has a full complete copy of Chinook software "versus 1992" - when program played match with Marion Tinsley, including opening book and ending database (little bit less than 8 pieces). He said - it will not be a serious problem to reproduce program, databases and hardware exactly to the level which it had in 1992.

I asked him - if I can play match with this version of program some day, and he said: "Why not, if there will be a good sponsors there" ? :lol:

Alex
I am playing checkers, not chess.
User avatar
Alex_Moiseyev
 
Posts: 4082
Joined: Sat Nov 12, 2005 5:03 pm

Re: An intriguing question....

Postby Ingo_Zachos on Tue Dec 20, 2005 11:39 am

Well, a famous chess grandmaster once said, that no human can run as fast as a ferrari, and he is indeed surprised that it is still possilbe for slow thinkers that humans, to compete with an incredible fast coimputer.
A Ferrari formular one car has 350 km/h speed, a human 36 km/h (an athlete), so the ferrari has an advantage-factor of 10.
A computer calculates 2 million positions a second, a human maybe 9 or 10, provided she is trained and fast. That's an advantage factor of 200.000! in checkers or chess.
And it has books and databases bigger then our memory, not to mention that the computer does not commit easy blunders because of nerve strain.
So any game that is not a win for the computer is a surprise.

Given that brute facts, do you still think that a comparison between human and computers makes sense?
Does any human being have the chance to compete with a ferrari?
Do athletes compete in car racing?
Would you bet on them?
It seems like a silly quaestion, but the thing is , that a "mechanical device" can always outperform a human organ, no matter which one, including the brain in an isolated, specialised task like running or calculating.
In mind sports its advantage in numbers is even higher then in athletic sports!
Things r not always look as "logic" as you thought, CheckersStronmgplayer, if you consider the fact that building devices to outperform humans has a long tradition, think of ballons, cars, dishwashers, mills, guns, amplifiers, etc...

Greetinx from Dortmund, Germany

Ingo Zachos
Ingo_Zachos
 
Posts: 1272
Joined: Sun Nov 13, 2005 7:41 am
Location: Dortmund, Germany

Re: An intriguing question....

Postby Pedro Saavedra on Tue Dec 20, 2005 3:33 pm

> Given that brute facts, do you still think that a comparison between
> human and computers makes sense?

Not if the purpose is to determine who is best. Yes if the purpose is to determine which human can do better against the same program. Alex suggested that he wants to reproduce the Tinsley feat. He is not trying to beat Chinook. He is trying to beat or at least match Tinsley.

Suppose we had a program that could set its own level. Whenever you beat it, in increases the level. When you lose, it decreases the level. Eventually it can estimate the level at which he an you are even. Now bring 100 players, and have each play twenty games against the computer. You can have a tournament and declare a winner, and no player would have to face another.

There are many variants, but they all make sense as long as it is understood that the real competition is other humans.

Pedro Saavedra
Pedro Saavedra
 
Posts: 205
Joined: Fri Nov 18, 2005 10:54 am

Re: An intriguing question....

Postby Ingo_Zachos on Tue Dec 20, 2005 4:39 pm

This wouldn't work, because playing strenghts is "not transitive" as the mathematicians say.

Suppose Long beats Ryan, Ryan beats Hellmann, but Hellmann beats Long.
You see? Long ans Hellmann did quite different against Ryan than in their own encounter?
Why?
Because styles r different, someone has a bete noir or he the players were in different shape in their different encounters...

Do you know that there was the computer program called Hydra, who beat Grandmaster Adams, appox. number 10 in the World (Rating 27100) by 7-1.
The same programm lost 3,5 to 4,5 against a computer programmer with a rating of 2200.
Why?
The programmer played some "scrappy" lines that r positionally bad, but offered no tactical possibilities.
The program did not convert his advantages in the games, won one on a tactical blunder and lost two by bad manouvering.
I tell you, if Adams would have had those positions against the same opponent, he would win 7-1 or even 8-0 , knowing how to exploit his advantages.
Checkers is even more based upon calculating, so I suppose the program here faces an easier task to convert positional advantages, but some players might play better against the program because they prefer sound lines with a lot of drawn games against the program, others would loose because their style is more a "calculating" style, but they still could win against the player that performed better against the program.

So your idea is bound to fail, as well as those comparisons like,
"Tinsley beat Lowder by 15 games to nil, Moiseyev beat him by 12 games to nil, so Tinsley was better".

The fact is you cannot compare it!

Only recent results can be compared based opon statestical caslculation as the Elo-System offers, but even then you cannot compare 2800 Tinsley had with 2700 Moiseyev has.
The quality of the players and the resistance the opponents offer r different, and can only be compared with a "comparatively" sound standard deviation in results of the recent past.

This means that your method would be futile and in vain, even if you produce a computer playing always with the same strenghs.
And as a bad programmer I know that the best program may perfom badly on a different hardware, as the greatest advantage of the program is its brute force, but to change the hardware, means always to change the calculation speed of the program.
So even a program without changing its code or datases or options may have a "bad" shape on another hardware.

Greetinx from Dortmund, Gertmany

Ingo Zachos


[/quote]
Ingo_Zachos
 
Posts: 1272
Joined: Sun Nov 13, 2005 7:41 am
Location: Dortmund, Germany

Re: An intriguing question....

Postby CheckersStrongplayer on Tue Dec 20, 2005 5:02 pm

Ok, thank you for the Info. I guess that Nemesis can beat Alex then.
Checkers is like being in heavens.
User avatar
CheckersStrongplayer
 
Posts: 92
Joined: Mon Nov 14, 2005 8:06 pm
Location: Houston,tx


Return to General Discussion

Who is online

Users browsing this forum: Bing [Bot] and 0 guests

class=