Deep Blue vs todays (computer chess programs)

bro_Robert

 

hello everyone, I am just curious, :) just want to ask if the computer chess program called Deep Blue who defeated Kasparov way back 1996 and 1997. Do you think this computer software can match against todays top computer softwares like Fritz and Shredder?

thank you

 

ivan1997

I think it would depend on time setting, and engine parameters, but it would hold good against them i suppose :)

pfren

Deep Blue was not a computer software, you are not well informed.

http://en.wikipedia.org/wiki/Deep_Blue_%28chess_computer%29

Dutchday

Deep Blue was not running on a commercial processor. If I remember it right, the machine weighed more than Kasparov? All things considered, the results will depend on the hardware backing the software. It's not a fair comparison if Fritz is just running on hardware the size of a cigar box. 

bro_Robert

Thanks for the information guys, I thought Deep Blue is a commercial software too :) 

losingmove

What's the go with computers vs humans? Do computers now routinely beat any and all humans?

ChessboardAbs
losingmove wrote:

What's the go with computers vs humans? Do computers now routinely beat any and all humans?

Oh quite. As I understand it the hardest part about designing a chess program to play with humans these days is making it weak enough to beat.

ChessboardAbs

I personally have never found a playing setting that wasn't either trouncing me or making ridiculous errors. Computers do not seem capable of the kind strategic misconceptions that arise in actual games.

sajay

what would have the rating of Deepblue if it existed now with the same features?

waffllemaster

Also, Deep blue lost to Kasparov in 1996 4 to 2.  In 1997 it won 3.5 to 2.5

Kasparov has a plus score agaisnt deep blue, 4 wins, 5 draws, and 3 losses.

bro_Robert

Do you think today's Deep Fritz and Deep Shredder are better than IBM's Deep Blue?

bro_Robert

I agree sir firebrand, even grandmasters are using computers to enhance their skills.

WalterRamiro

can i play vs deep blue

fyy0r

The strength of an engine depends primarily upon it's positional evaluation function.  A good evaluation function means the engine can be effective with much lower CPU strength.  Today's engines are VERY good.  If you watch the Deep Thought (the first match) documentary, you will see some fundamental flaws in the way the programmers think.

Go to 3:50 of this video:

http://www.youtube.com/watch?v=aU1_fE6B2vk

Instead of trying to make the best possible positional evaluation, he's throwing more and more processing power at the problem as if he has all the secrets to chess solved and the computer just isn't searching deep enough.  That ofcourse wasn't true. Today proves that you don't need to see 1 billion positions per second to have truly great engines and that the evaluation function itself is where strength comes from.

Today's engines will defeat Deep Thought and Deep Blue quite handidly on less powerful hardware.

fyy0r
cookiemonster161140 wrote:

The evaluation function is important, but so is horizon effect. Today's supercomputers are several hundred times faster/more capable than the one which ran deep blue.

Modern PC (or MAC) Engines maximize their potential by looking at only the best candidate moves up to that horizon. They are not capable of looking beyond the horizon. For them, the earth is flat. Meanwhile deep blue is capable of looking past the horizon in search of the absolute truth of a position.

So assuming the GM-designed decision making process with deep blue is fairly competent, it's not a big stretch to say modern supercomputer would easily crush a home PC.

Remember supercomputers cost billions of dollars to build. PC's can be had for about $300. Now I KNOW supercomputers are primarily govt contractor lowest bidder with costs that explode once the contract is signed but...they're still more powerful than some silly laptop.

But have thought maybe Deep Blues evaluation function is simply inferior to todays programs?  That is what matters the most.  If modern engines can better evaluate a position then it does not matter if they can't calculate billions of positions per second and that is what im saying.  Todays engines are simply programmed better and no amount of CPU power, then or today will help Deep Blue.

TheGrobe

Deep Blue wouldn't stand a chance.

fyy0r
cookiemonster161140 wrote:

Like I was saying earlier, it's a match that will never take place. Like Fischer and Karpov.

But computer horsepower does matter. If it didn't, government agencies wouldn't need supercomputers to test nuclear weapon designs. They could just run to Wal-Mart for the latest E-machine.

Ofcourse CPU power matters, if ALL OTHER THINGS ARE EQUAL!!! Deep Blue and Rybka are entirely different engines!  If Rybka is simply programmed better to understand chess than Deep Blue, then Deep Blue can have all the processing power in the world, it can't be saved.  Now if Rybka on a home PC is up against Rybka on a supercomputer, ofcourse the supercomputer Rybka would win.

So silly to have to explain

boringidiot

I find it perculiar that people keep mentioning "Fritz" as if it was the top choice. 

Rybka and Houdini have dominated the top list the last 4 years.

fyy0r
boringidiot wrote:

I find it perculiar that people keep mentioning "Fritz" as if it was the top choice. 

Rybka and Houdini have dominated the top list the last 4 years.

Yes maybe, but they are all ultra strong, so who cares in the end?  Rybka, Houdini, Shredder, Fritz, Junior, Stockfish, Critter... the list goes on, they are all giants and will guide their masters

boringidiot
Shadowknight911 wrote:

even middle-of-the-road software such as Chessmaster:Grandmaster Edition is around 2900 or so at the highest levels.  The top dogs such as Houdini and Firebird and Rybka are all running around 3200-3300.

It is certainly true that it doesn't matter much (anything) for a player like myself, but the difference can be hugh in endgames. Even I have seen Houdini solve positions where Fritz and several other engines commit errors. I am not talking about table bases here. Only engine judgment.