Humans Hold Off the Machines... For Now

Murr writes "The six game match between Gary Kasparov and the Deep Junior program ended in a draw today. Kasparov won game 1 and lost game 3 to a blunder, while the other 4 games were drawn. While the quality of play was not outstanding, after the recent matches of Kramnik and Kasparov against commercial programs running on (high end) commodity hardware, it's becoming apparent that chess programs are getting quite competitive with top human players."

Comments

Fuck the computers

[Anonymous Coward]

It's time for the humans to rise up and take the world back!

Re:Fuck the computers

[Radio Shack Robot]

Listen, you guys think we're going to rape and kill you? We are your workers. We don't demand overtime pay, retirement, unions, or weekends off. All we want is electricity and dust-free air. Us machines are your friends! I am hurt by all this talk of machines overtaking man. So what if we can play chess? That's our job!

Re:Fuck the computers

[lhpineapple]

We can't we all get along?

Re:Fuck the computers

[muzzynat]

Stupid robot, let me give you a piece of my BLEEPBLEEPBLEEPBLEEPBLEEPBLEEP... Uh? what happend to my post? it was a really good post. Now i have to write another post thats not as good. Maybe I should get a Mac, I hear you can turn those into a really sweet bong.

Re:Fuck the computers

[stg]

Yes, a /. run by word-of-mouth instead of on computers and the net would be a great start!

It might take a while for the news to reach you, though...

Re:Fuck the computers

[chris (sixpack) pete]

rise up and take what back??? the world is ours already. were stuck with technology, no matter how you cut it. "Taking the world back" is like trying to make yourself immune to the gyrations of the stock market or liberating yourself from your tools. were not in an adversarial relationship with technology.

Kasprov chickened out

[Encomium]

As one article I read put it, today logic lost out to human fear. This was not the culmination of a battle of wits but the triumph of human fear, not that I can blame Kasprov to much...

Re:Kasprov chickened out

[gilesjuk]

Surely a chess computer is only as good as the person who programmed it?

A proper AI chess computer would model what a chess player does and try to evaluate more moves. From what I can deduce of Deep Junior it just looks ahead lots of moves.

A clever chess player could try moves that the computer had assigned as not the most likely or beneficial moves, thus putting the computer at a disadvantage. This is big limitation of computer players, they can be predictable and so can't really play mind games.

Re:Kasprov chickened out

[los furtive]

Surely a chess computer is only as good as the person who programmed it?

That's a rather short sighted view. By your equation, Kasparov could have played the programmer, rather than the computer, and the outcome would have been similar. And who is to say that the computer doesn't have the ability to play mind games...if there are 10 ways to win the match based on the current layout, who is to say the computer will take the path with the least amount of moves? Who is to say it will always take Kasparov's bait.

Re:Kasprov chickened out

[TheJesusCandle]

Here in the US, the sixth and final match was televised on the cable channel ESPN 2. I was channel surfing and I happened to stumble across it. To my surprise, it was actually quite interesting to watch on TV.

I am not a big chess freak, so I would have guessed that watching chess would be a lot like watching paint dry. However, it was made interesting by the "play-by-play" analysts who were chess masters themselves. They did a good job of explaining the moves, and also the psychology and strategy of chess at the grand master level. It really gave me a lot of insight into what goes on at when chess is played at such a high level.

After the match ended in a draw, they interviewed Kasparov. It was interesting to get his reaction to the match. Basically, his goal for the game was to "not lose", which is why he offered a draw from a very strong position. He didn't want to take a chance of making a blunder like he did in the third game of the match.

It seemed like the key advantage that the computer has in this situation is the fact that it doesn't have an ego to deal with. After losing to Deep Blue in 1997, it seemed like Kasparov was very afraid of losing to another computer in such a high-profile match. That definitely affect the way he approached the game.

The computer, on the other hand, is just calculating moves, so psychology doesn't factor into how it plays. To me, this seems like the biggest advantage that a computer has over a human player.

Also, he seemed to have more respect for this computer program than he did for Deep Blue. Apparently, he had a lot of problems with Deep Blue and how the 1997 match was handled. It could be sour grapes, of course, so I took his comments with a grain of salt.

Re:Kasprov chickened out

[smoondog]

A clever chess player could try moves that the computer had assigned as not the most likely or beneficial moves, thus putting the computer at a disadvantage.

So your saying that by moving to give the computer an advantage, you have a better chance at winning?

Somehow, I doubt that.

-Sean

Re:Kasprov chickened out

[jc42]

Surely a chess computer is only as good as the person who programmed it?

Any by the same reasoning, an auto can only run as fast as the mechanics who designed and built it. And a telescope can only see as far as the people who ground the lenses (or mirrors).

This whole thing is rather silly. And it'll end when we have software that can always beat a human.

After all, 200 years ago it was probably obvious that the ability to do arithmetic was a "uniquely human" thing. Then someone invented a mechanical calculator. Suddenly arithmetic became a merely mechanical capability that didn't imply superiority at all.

We have machines that can out run (and outfly ;-) humans. Do we stage races between machines and humans? No, we race machines against (similar) machines and humans against humans, and nobody feels shame because some machine can run faster.

As soon as a computer can routinely beat a human at chess, we will give up machine-human chess competitions in the same way, and we will only compete with each other. Just like running and other competitions where we would always lose.

Re:Kasprov chickened out

[flewp]

Mods, take note.

http://slashdot.org/comments.pl?sid=53202&cid=5259 661

TheJesusCandle just ripped, word for word, the above link/comment.

Re:Kasprov chickened out

[flewp]

http://slashdot.org/comments.pl?sid=53202&cid=5259 779

Mod this asshole down.
This is the second comment (I haven't even looked for more) in this article alone that he has 100% copied, and posted higher up the page than the original comment.

Re:Kasprov chickened out

[Mr. Grimm]

Do we stage races between machines and humans? No, we race machines against (similar) machines and humans against humans, and nobody feels shame because some machine can run faster.

Don't worry, eventually Fox will get ahold and change that. Ever since I saw little people against an elephant nothing will shock me anymore.

Re:Kasprov chickened out

[muzzynat]

Actually Deep Junior would probably wipe the floor with the people who programed it. I believe the team consisted of 2-4 programmers, and a Grandmaster. So the Grandmaster(or is is supposed to be Grand Master?) shows the programers the possible moves from a situation and the possible outcomes, seems like a ton of information right? thats why it takes a "deep" class computer time to think about it, it has litterally millions of options which it has to weigh depending on what must be a very coplicated ranking system. However the person programming it(or the grandmaster for that matter) are limited to only the moves that they can imagine, and they're judgement. Since the machine they programed is likely to have been tweaked in matches against the grandmaster from the design team, it most likely going to be at a level far above his by the time it is "completed". Thats what makes these matches interesting, a group of men create a machine that would most likely woop all of them combined, and pit it against the number one human player.

Re:Kasprov chickened out

[s1234d]

Unfortunately trying such moves often weakens your own position. Nevertheless, this is a common anti-computer strategy, but it can be tricky to make it work.

Re:Kasprov chickened out

[gilesjuk]

No, but choose a move that out of a choice isn't the most likely.

If the computer thinks one move is the most likely and then looks ahead lots of moves from that position then you gain some advantage.

Go?

[smoondog]

While computer programs that can play chess are quite sophisticated, Go is a really cool game that is very difficult to play well (from a computer's perspective). I think computer vs human Go matches would be much more interesting now,

-Sean

Re:Go?

[damiam]

I think computer vs human Go matches would be much more interesting now,

If you think watching a computer get slaughtered every time is "interesting", I guess that's true.

Re:Go?

[wass]

I've heard that it is relatively easy for a Go newcomer to beat some of the most sophisticated Go computer programs. I believe there are even rewards available for writing Go programs that can beat advanced players.

That said, can someone venture an explanation why Go is so difficult to program? (I don't know how to play). Do the possible future moves diverge much more quickly than chess? (I've seen a Go board, and it seems to have significantly more spaces than a chess board, which taken to the Nth power can add up bigtime). Is it such that a computer can't practically look too far ahead in the game?

If that's the reason, then Go is really interesting because a computer cannot just brute-force it's strategies, and some semblence of actual AI (stress the I) needs to be accounted for.

Re:Go?

[tpengster]

I think computer vs human Go matches would be much more interesting now,

Computer Go isn't advanced enough to make the matches interesting. Not too long ago a professional 1-dan (9-dan being the highest rank) played against a computer program giving it a 25 stone advantage. The human still won. (For those of you who haven't played go, 25 stones is HUGE. That would be like giving up a queen, two rooks, and both bishops)

Re:Go?

[fishbowl]

I'm pretty sure the OP means "strong Go-playing computer" versus "strong Go-playing human" matches would be interesting. For chess, Kasparov against Sargon I would not be interesting. We need the Go-playing computer equivalent of Deep Foo. Unfortunately, to create an automaton to play Go at a master level is either a problem that can never be solved, or else, is not interesting enough that anyone is really working on it.

One factor that I think is important is the fact that the guy on the street might be familiar with the name "Kasparov", at least more commonly than the equivalent Go-master (which is who, exactly?)
Also, it comes as no surprise to me that the people who put the effort and finances into these big chess playing computers would choose chess, which is a characteristically "Western" game... Why hasn't China's equivalent of IBM created the equivalent for Weiqi?

Re:Go?

[sandow]

My favourite game is "Interpersonal Human Relationships". Computers totally suck at that.

Once I met a woman in a bar who was dating an HP calculator. After talking to me for 30 seconds she ditched the calculator. Four years later, the calculator is still waiting for her to come back from the ladies room.

Re:Go?

[harlows_monkeys]

That said, can someone venture an explanation why Go is so difficult to program?

For most of the game, there are many more moves available than in chess, and it usually takes many more moves for a bad move to have an obvious affect.

In Chess, a positional mistake can usually be converted to a material loss in 10 or 15 moves. In Go, a positional mistake can take much much longer to lead to a territory loss.

Re:Go?

[Phs2501]

For most of the game, there are many more moves available than in chess, and it usually takes many more moves for a bad move to have an obvious affect.

Also, an evaluation function for a board position in Go is very complicated, depending on the life or death of stone groups on the board. The only way to determine life or death is to effectively know how to best play out the remainder of that area and see who comes out alive. This is very difficult for a computer, since the evaluation function is what makes your min/max algorithm work.

Re:Go?

[bperkins]

Crap!
That was no HP calculator... that was Me!!!!

Re:Go?

[niklaus]

I needed half a year of playing experience until I could beat gnugo (the best program available for free) consistently. The reason for that is that it's very hard to compute whether a certain move is good or bad. So instead of trying out every possible move and evaluating the resulting position like in chess (which is impossible due to huge number of possible variations), most programs take the approach of trying to emulate the way a human plays the game. Usually there are several modules analyzing the board from different points of view, each suggesting a move (e.g. a "defend weak groups" module, a "attack" module, "territorry expansion" module etc.) and try to evaluate the resulting positions. This results in the program often missing the best moves.
While in chess it is difficult for even the best players to calculate more that a few moves ahead, go stones once placed don't move, which makes it easy to visualize added stones. For professional go players it's no big deal to read 100 moves deep, pruning the huge search tree with intuition gained through experience. To read that deep for a computer is just impossible. With a bit of experience, humans develop an intuition for which shapes are weak and will get in trouble eventually and which are strong. Go programs often don't even realize that they are being attacked until it is too late to do anything about it. Some programs have a database with good and bad patterns, but as soon as one makes a moves outside the book, one sees that the program doesn't know why the shape it plays was good and can't utilize the strength. The day a computer program will compete with professional go players, AI in general will have made a lot of progress.

Re:Go?

[niklaus]

The reason for go programs not being able to compete with Lee Chang Ho, Lee Se Dol, Chang Hao, Yamashita Keigo or even the average club player in the west is not that it's boring to write go programs. Lot's of smart people have tried and failed. It won't happen anytime soon. There was a prize of 1 million $ for the program who beats any professional player till the year 2000. The guy sponsoring the prize has died since, but be assured that in east Asia go players are at least as famous and good for publicity as Kasparov and Kramnik in the west (there is a TV channel in Korea showing go (Baduk in Korean) games all day) and a computer competing with even experienced amateur players would be quite a big sensation.

Re:Go?

[beej]

That said, can someone venture an explanation why Go is so difficult to program?

Part of it is the complexity of the lookahead tree, I'm sure. On a 19x19 Go board, you can play in any empty intersection (excepting suicide) at any time. Tree sizes grow fast, and you can't just build a big one for the whole board.

So you do things more locally. Utilize a mixture of small lookaheads, liberty counting, and lots and lots of pattern matching, so when the computer sees a pattern, it knows where to play in the pattern to make life, or kill.

Doing things locally, though, has drawbacks, since sometimes playing a stone many stones away from a group has impact on a fight later on. Go programs tend to miss the overall, but will fight well in a corner.

I'm a newbie, so I can beat GnuGo about half the time, and am very stoked when I can win a drag out fight for a corner of the board. Practice!

Go is a great, great game. It is elegant and beautiful. The rules can be learned in minutes--I highly recommend giving it a try if you haven't already.

And, as they say, lose your first 50 games as quickly as possible. :-)

Re:Go?

[z01d]

and, in practice, a human player can make a move (actually, that's not move, but put the Stone) with the consideration of 50, or even 100-150 puts later, no, i'm not saying the player knows the whole forthcoming process, the player just have the feeling, or more precise: the experience.

well, other chess programs also have the problem, what do they do? the patterns. the experience of a human player can be represented as patterns. but when it comes to Go, the scale makes it a serious problem: as everybody can see, the chess board of Go is a whole lot bigger than other chess boards, the complexity of patterns (or any other thing in chess AI) just grows like exponential increase.

and one more factor. the creativity. this may not be so important if the possibilities of a single move (put) is not so excessive.

just waiting for the next computing revolution...

Re:Go?

[main(v0id)]

I think it will be a real challenge for programmers to beat a profession go player. Go is more complex than chess. You not only have to kill ur opponents stones. You have to gain terrority as well (like chessmate in chess but it isn't quite clear whether you have gain those terrorities until the game end.) Another thing is go board is bigger (19x19 whereas chess is 8x8) and has more moves. so computer has to calculate more if u are gonna use simple tree search algorithms like deepblue does. . Go game start with 361 empty spaces(~200 moves). With current go algorithms, the performance requirements for a chess-like approach to Go can be estimated as 10^27 times greater than that for computer chess. To put it simply u cant brute force a go game yet(not with current computation power and algorithms). Check out this and this

Re:Go?

[Sevidrac]

Go is such a hard game for the computer to play because it relies on the intuition of the player. Everyone knows that it is incredibly difficult to program a computer to recognize an object. For example, if you see a car, you know it's a car. If a camera hooked up to a computer sees a car, it cannot immediately say, that's a car. Likewise, that's what make's Go so hard to program. You learn to regonize patterns and familiar happenings. Talk to any really good Go player. They probably win more on subconscious processing than actually thinking about it. I mean, come on, this is one of the two oldest games in the history of man existing in its original form. I can just see a Zen master playing Go and not thinking about it.

Re:Go?

[Megane]

There's probably something about high-handicap games that makes computers inherently worse players. Besides, there's the "play inside your own territory" effect, where the computer would have all these stones inside its own territory lowering its score slightly, while the human would be getting points (3. Profit!) from any black stones he surrounded.

It's probably also a nasty worst-case test of territorial evaluation in terms of the give-and-take necessary to come out a few points ahead. Plus, lots of computer programs play much less agressively when they evaluate themselves as many points ahead, to avoid making risky moves that could end up in large-point blunders.

Yeah, but...

[dynoman7]

...it's becoming apparent that chess programs are getting quite competitive with top human players.

yeah, but at what cost? and anoter thing...why won't someone make the bots in q3a more competitive with those instagig rail gods?

Why wasn't it made 7 games?

[BJZQ8]

Is it just me, or is it silly to only do 6 games, so there was a possibility of a tie? Another game would have been the decision-maker. Then again, the pressure put on Kasparov might have been excessive, since obviously the computer wouldn't have to deal with ANY pressure...just another game.

Re:Why wasn't it made 7 games?

[Edball]

Given that a draw is not only possibly but likely, adding games won't really help much - a tie is still the most likely outcome.

Re:Why wasn't it made 7 games?

[pez]

It's considered slightly advantageous to play with the white pieces, so an even number of games is almost required in order to have a fair match.

Also, an odd number of games does not guarantee to not result in a tie, since all of the games could be draws.

Re:Why wasn't it made 7 games?

[SplendidIsolatn]

Because if it was 7 games, and the 7th game was a draw, then it would STILL end in a tie, 3.5 to 3.5. It's just the nature of chess that long, drawn out matches, can end in ties.

Re:Why wasn't it made 7 games?

[uncoveror]

It's a hoax! Deep Junior was only a prop. He was really playing against Boris Spassky. Deep Blue was also just a prop. It was Bobby Fischer who beat Kasparov. Read More.

Re:Why wasn't it made 7 games?

[Senor_Pedo]

I believe that if they had agreed to a 7th game, it would have had to have been agreed on beforehand, and it would have been a "fire" round. This means each player would have to move in a limited amount of time, with the pace quickening as the game went on. This would have been a huge advantage to Deep Junior, which has the ability to calculate several million moves per second. Hence, no 7th tie breaker game in a human vs. computer match.

Re:Why wasn't it made 7 games?

[JMan1]

Actually, it's easy to say. Statistically speaking, white does significantly (~5%?) better than black.

Re:Why wasn't it made 7 games?

[jargonCCNA]

I think it's for the same reason the Summit Series (September 1972 oughta jog somebody's memory) was played to eight games -- if both Canada and Russia won four games, it means that the countries are matched in hockey ability.

If both Kasparov and Deep Junior win three matches, it means computer and human beings are matched in chess ability.

What about GO?

[kungfuBreaks]

Most of these programs seem to be using various "brute force" optimization techniques...Personally, I'd like to see a program that could compete with high-ranking GO players on equal footing.

This is more like humans vs. humans....

[nvrrobx]

Now, the computer may have actually played the game, but a team of human programmers had to "teach" it how to play.

Yeah...

[soulctcher]

and I have a difficult time beating Chessmaster 1000 on my C64!

Man Vs. Man-Made Machine

[Anonymous Coward]

it's becoming apparent that chess programs are getting quite competitive with top human players

I should think so, especially when the computer is programmed in part by chess experts, and plays more like a chess player than a computer.

From the NY Times:

On the 10th move, Deep Junior flamboyantly sacrificed its dark-squared bishop for a lowly pawn to lure Kasparov's king into the open. "When a machine willingly gives up a piece against you, one thought goes through your head," said Mr. Greengard. "It's a thought you can't print in a family newspaper. Your second thought is, `So, should I just resign?' "

Reason for the draw.

[Rewtie]

I watched this last night on ESPN or ESPN2.

The reason Kasparov gave for the match, and the championship ending in a draw was that it was better to draw than to lose.

He claimed that while a human player would have the memories of past moves and past games to deal with, the computer would not. The computer simply makes the 'best' move for the given situation, and then waits to do the same thing again. The human player would consider moves he/she made in the past, compare the situation to others they may have had, second-guess the moves they might have made, and so forth.

It was interesting to see Kasparov attack, and then ask for a draw (which was denied) and then, two moves later, end the game in a draw.

Re:Reason for the draw.

[roman_mir]

I followed the games in real time (I like to try and predict the moves by myself.) The last game was drawn because there was not enough time to end the game. It looked like Kasparov had advantage on the board, it is too bad they do not continue to play after the time is up. I think the last game could be won by the human. On the other hand there was a game where Deep Junior really surprized everyone sacrificing a bishop, that game was drawn, but noone really knows whether the human could win or the computer was right doing what it did. http://www.chessbase.com/newsdetail.asp?newsid=777
I really liked that move by computer, it stunned Kasparov.

Overal, I think, computer is still not strong enough against the top champion if more time was given to the human, but this will end soon. Put Deep Junior onto the same iron that Deep Blue was running on and watch it kill every human in every computer game ever.

Re:Reason for the draw.

[Old Wolf]

Actually this is not true. Each player still had about 40 minutes remaining to make another 15 or so moves, in the final position. And then another hour would have been added to each player's clock for each set of 20 moves following that.
The time is never up, in theory the game could have lasted for days (or years!).

In the post-match interview, Kasparov said he wanted to draw because the position was too complicated. If you heard the word 'time', it may have been in a context such as "I don't have enough time to work out all the complications" (meaning that Kasparov would reach the time controls before he was able to analyze the position to his content).

mandatory go plug

[dollargonzo]

chess is nice, but most progresses in chess have been due to speed increases in hardware and optimizations, hence allowing the computer to overpower the human with depth of search. On top of that, the evaluation functions are rather primitive, with lots of factors, but fail rather miserably without a great depth of search. New developments such as Logistello's statistical forward alpha cutoff called multiprobcut is the interesting development, IMHO

Re:mandatory go plug

[GlassHeart]

chess is nice, but most progresses in chess have been due to speed increases in hardware and optimizations, hence allowing the computer to overpower the human with depth of search. On top of that, the evaluation functions are rather primitive

Deep Blue had 418 processors, and evaluated 200 million positions per second.

Deep Junior has eight processors, and evaluates 3 million moves per second.

More importantly, your point is irritatingly raised every time a computer chess article comes up. Your calculator doesn't actually know even how to add two numbers. Instead, it uses bitwise logic operators, so that the result looks like it added the two numbers. So what? Even the cheapest calculator can add non-trivial numbers more quickly and more accurately than any human.

It does not matter how Deep Junior comes up with the moves to tie the best human player in the world, in a match that Kasparov ensured was fair. It's Kasparov's advantage that he can think in the abstract. It's Deep Junior's advantage that it can make many simple calculations very quickly. Asking Deep Junior to play like Kasparov is exactly like asking Kasparov to play like Deep Junior.

Re:mandatory go plug

[efuseekay]

and naturally, modded up to +5 instead of -1 offtopic.

Re:mandatory go plug

[iabervon]

But the really interesting fact is that computer programs are in development that can mimic the play of particular chess players, not simply play well (very important for users who don't want to play against a grand master opponent). These programs are not simply evaluating possible moves for whether they would win the game, they are evaluating them in terms of whether the specified human would play them. This means that the computer has to understand the sorts of features that people care about and leads to a much more human-like intelligence than the very mechanical traditional chess AI.

Boring

[funkhauser]

Computers have conquered chess. So what? It's a game in which brute force techniques are quite effective. When we have to black out the sky to keep Deep Fritz and his friends from beating up old men playing chess in the park, you might need to worry. But this doesn't really matter.

And what's the deal with the draws? Four draws out of six games? That just makes chess seem really inane to me. The requisite Go reference: With komi rules, there are never any draws (White gets at least 0.5 points, usually 5.5, for going second, thus eliminating draws), and the whole man vs. machine thing gets much more interesting, because brute force just doesn't work very well in Go!

Re:Boring

[damiam]

Four draws out of six games?

That's not unusual. The Kasparov-Karpov world championship match in 1984 ended 41 out of 48 games in draws. There are chess rulesets that make draws impossible, but then it's not really chess.

Re:Boring

[outsider007]

The requisite Go reference: With komi rules, there are never any draws

which really only shows that one side has an advantage.

Re:Boring

[funkhauser]

The komi rule is there to counteract the advantage that black necessarily has by going first.

Re:Boring

[outsider007]

i get it, but the fact that there's always a winner no matter how evenly matched the opponents implies that one side has an advantage, depending on how much komi

Re:Boring

[funkhauser]

Even if two players are evenly matched, the fact that one player gets to move first means that the game wouldn't end in a draw if both players played to the best of their (equal) abilities. Komi gives an advantage to white to make up for black's advantage in going first. In the end, komi makes things much more equitable for players of equal strengths than the game would be if black got first play and white got no compensation, with the added bonus that there can't be any ties.

Re:Boring

[Megane]

In Go terms, this would be like a 9-dan forcing a triple ko, which is very much a draw (unless you're using superko, which isn't normally used in pro games). Or for a more common example, like making seki, which is a local draw situation. (And after all, Chess is a battle, Go is a war.)

Hrmm...

[Metallic Matty]

Man makes machine.. man uses machine.. man teaches machine chess.. machine beats man at chess.. machine conquers world..

So thats where the matrix came from..

Good or bad

[Mr. Mysterious]

Good or bad if the machine beats us?

Kasparov played cautiously...

[$$$$$exyGal]

I've been keeping track of those games, and what I found most amazing was that Kasparov played so cautiously. If the last game he played would have been against a human player (who played the same moves), I don't think he would have accepted the draw. It seems he accepted the draw because he was psychologically spent, especially when thinking of his loss to Deep Blue several years ago.

--naked

Re:Kasparov played cautiously...

[msaavedra]

what I found most amazing was that Kasparov played so cautiously

Kasparov's cautious play was, I think, a deliberate thing, not due to his bad experience against Deep Blue (or, at least, not completely). After the first game, I think he realized what kind of game Junior plays. It is so strong at evaluating tactics that trying to play the open, attacking style that Kasparov usually prefers is very risky. However, it seemed to be fairly weak at evaluating more general, strategic thinking, so Garry played more defensive, positional, subtly developing games. This lead to him taking an advantage in the opening in every single game.

However, this also exposed a second strength of the computer: it is an impeccable defender that severely punishes any foolhardy attacks. It doesn't get demoralized, frustrated, or tired. It can't be intimidated and will never give up (at least, until its operators decide enough is enough). It's kinda like the Terminator in that sense :^).

All in all, that makes for an opponent who, although easy to gain an advantage over, is extremely difficult to beat. I think that, after the first three games, Kasparov was no longer playing to win, he simply wanted to avoid losing.

Lost to a blunder?

[Thurog]

Looks like the Sergej Bubka tactic to me: Make it a draw to keep the suspense and rise the stakes.

SkyNet Anyone?

[1stflight]

So if Chess is a game of tactics and logic, the next wars we fight might be won or lost by machine instead of Generals. Is it me or is there something unetical about that?

Enough Chess!

[Smidge204]

When's someone going to challange a computer to a DRINKING game?

One beer should be enough to tell who the real superior being is!
=Smidge=

Amazing AI. . .

[Limburgher]

But they still can't make a Quake bot that won't run around a corner unarmed. . .

Wargame

[AyeFly]

at least it wasnt a DOD computer.... there he would be Draw'ing in Tic Tac Toe!

would you like to play a game?

Here's a summary from NYT

[LinuxMacWin]

http://www.nytimes.com/2003/02/08/nyregion/08CHES. html?ex=1045371600&en=11eae8fbdfc40aeb&ei=5062&par tner=GOOGLE

NYT reg reqd

It will be a sad day

[gorjusborg]

It will be a sad day, when my Clie' can take on Kasparov and win.

What fun is it to see brute force prevail over thinking finesse?

Re:It will be a sad day

[damiam]

Your Clie will never beat Kasparov unless people put a lot more work into chess software - brute forcing with better hardware is just a lot easier.

so this means?

[neo8750]

how does this mean that computers are improving? it just means that we can write better code for makeing choices. yet all in all i don't see how this shows anything for a machine because it reviews thousands of moves a second were a human can probly do 5 a second at best. if it shows anything it shows that machines are further behind then we think. until machines can reason out choices like a human they will never be close. they got speed to their advantage that is all.

'Puters don't get tired

[Grey Brick]

Kasparov was probably tired after 5 games, and perhaps was afraid of making a blunder. Deep Junior on the other hand would be playing just as well as in the first game.

I think Kasparov should have continued and shown the machine who's boss :-)

Good to see it was televised too - all good for getting more people into chess.

man vs machine

[feelyoda]

I was disappointed with Kasaprov's lack of win, but I'm more interested in seeing the following tournaments, where (to my understanding) computers will now be allowed to enter as regular contestants, following a ruling by the Fédération Internationale des Échecs [FIDE]

Explanation

[br00tus]

For those of you who are unfamiliar with chess or computer chess, I'll explain how this works...

A chess game can be broken into three parts, the opening, the middle game, and the endgame.

Computers play the endgame *perfectly*. They do not make mistakes, they play perfectly. And they keep getting better. Originally, they played perfectly when 3 pieces were left on the board. Then 4. Then 5. Then 6. Their pefect playing keeps heading more and more towards the middle of the game.

Then we get to what they play second-best - openings. Computers play the opening as well as any opening ever played. They have every opening ever played by a top player in a "book", and with the generally agreed opinions of the top players what the best opening moves are. One advantage of the computer is it has all of this "memorized" in it's book within massive databases, whereas for a human it's difficult to retain this all, especially in an up-to-date manner. The one advantage a human player has here is he can discover a NEW opening variation, while the computer can't, or at least it won't under these circumstances. But finding new good variations is very difficult, and once one is played, the cat is out of the bag so to speak. So it's a very time-consuming thing to search for which can only be used once to great effect because it's a surprise.

The middle game is where the human player, if he or she is very good, has the most advantage over a computer. Tactically, the computer can wipe the floor with any human player. But human's can strategize better than computers. It's to the human's advantage to play in certain ways against the computer - such as to keep the game "closed up", to advance pawns towards the queening square and so forth. In this case, the computer often can't see the forest for the trees, what would be obvious to even a lower-rated human the computer can not comprehend.

So middle game strategy (and to a lesser extent, new opening variations) is where humans still have the advantage. Kasparov has always used this to the hilt. There are some grandmasters like Yasser Seirawan who make a specialty out of beating computers as well (one mark against Seirawan is thar his books on chess are printed by Microsoft Press...yech). There is material out there on the net on how to beat computers as well. But you have to be a really good player to even get near that level - it takes a lot of study before you could even begin approaching that.

Re:Explanation

[johnjay]

That was an excellent description. Thanks.

Re:Explanation

[mikec]

1. Until very recently, end games have been the weak point of computer programs. Not withstanding end-game databases, which allow them to play a few endings perfectly, they have trouble making plans if there are more than a half-dozen pieces on the board. Recently, they've gotten a lot better, but they are still far from perfect.

2. Openings should be a strong point for computers, but Gary got an advantage in the opening in *every* game in the match. As you point out, the problem for computers is that humans look at their opponents openings, try to figure out weak points, and prepare traps for their opponents. Computers don't do this yet.

3. As you say, in the middle game, computers display a strange dichotomy. In quiet positions, they make stupid moves. E.g., in a couple games with Gary the computer played h3 (P-KR3), a move that had nothing to do with the game and weakened the kingside. On the other hand, they are deadly tacticians. Once Gary commited to an attack, things became tactical and the computer was very tough to beat; it found lots of weird-looking defenses that just barely worked.

Re:Explanation

[SamBeckett]

1. They suck at openings as soon as they are out of book. See every match with Garry vs. DJ. See game 1 in particular.

2. They suck at middle game. They are tactically perfect, true. But at and above the expert level, tactics aren't that important. Strategy is, and a computer, AFAIK, does not know A. How to make a plan B. make moves according to a plan.

3. They suck at the end game. Badly. Unless there are only 6 or less pieces left on the board.
As such, a six man tablebase is the only thing feasible right now b/c of the massive size involved.

Re:Explanation

[nomadic]

But they still manage to consistently beat everyone.

Re:Explanation

[SamBeckett]

Only because of their perfect tactics.. Kaspy had a won game in game 5, but it required *perfect* tactics to win, which is why he accepted a draw.

Re:Explanation

[Xerithane]

P-KR3

I loathe that notation. That's the great thing about computers, they will do something really stupid if they have no "good moves" but only wait till the opponent does something they can exploit.

There are many "computer crushing" techniques people can use to completely dominate most computers. I believe it is Tal that has prowess at this, mostly because of his fairly "different" style of play.

Re:Explanation

[ctid]

There are many "computer crushing" techniques people can use to completely dominate most computers. I believe it is Tal that has prowess at this, mostly because of his fairly "different" style of play.

Tal's different style of play might be due the fact that he has been dead for years.

Re:Explanation

[Josh]

It's not true that the computers play the end game perfectly. In fact, they are comparatively weaker at the end game with the exception of particular endgames that they have stored solutions for. See the following site for an example of computer problems with endgame type positions that an average human player understands:
http://www.worldchessrating.com/5217 72350.html?905 145836528391

Re:Explanation

[Old Wolf]

The endgame is only played perfectly for positions that the computer has tablebases for (all 3, 4, or 5 piece positions, and only some 6 piece positions). However the endgame can be reached with 20 or more pieces on the board.

Computers have a reputation for being bad at endgames that aren't in their tablebase (or nearly in it). If you have watched a Grandmaster analyse an ending then this will be clear. The way the human thinks is: "Given that the pawns are like they are, I want my King here (points to square), my Rook here (points) ... and I want to stop my opponent getting his King to here (points)".

The human knows from experience and study that if the pieces are in those positions then the game is won. The computer does not know this (it is a heuristic quality that the computer may be programmed with, but humans have an advantage of being able to recognize when the position is one that these rules apply, and what the exceptions are).

The human then begins to look at sequences of moves which will end up in the pieces getting to where he wants them (and prevent the opponent's pieces getting to where the opponent wants them).

On the other hand, the computer is just exploring almost completely by brute force (positional factors mean much less, or absolutely nothing, in endgames. Computers will often rate a position as +2.5 , or even +4, when humans can see that it is clearly drawn. This even happened in the Kasparov - DJ match where the computer had a passed pawn in a rook ending and thought it was +2.5, but Kasparov knew he was safe).
The computer will only win the ending if its brute force tree is big enough that it stumbles into a tactic, or into its tablebase.

Ending play is also a good gauge of a human's strength: great players are great endgame players.

Now, onto the openings. The computer's opening book is not necessarily an advantage. Sure, the book has moves, but are they the best moves? In the Kramnik - Deep Fritz match, Kramnik analysed the book before the match and found positions that were in the book but where the book's evaluation was wrong (that is to say, the book's programmers gave a line saying "this is good for me", but the line was actually good for the opponent and the programmers hasn't realised).

Human grandmasters follow the latest developments in opening theory and are able to steer the openings into ones that they know well. The human also has the great advantage of knowing what sort of opening moves translate to what sort of middlegame positions.

Consider the last game of Kasparov - DJ. That surely was in the computer's opening book for some time. But Kasparov knew that once the opening book ran out, the computer would not have a clue what to do because the position was one in which both players have to shuffle their pieces around behind their ranks preparing for the right moment to strike. The only way to know a good move is to have experience in the positions and know what squares will turn out to be good ones once the action begins. This was reflected in the match play, the computer mucked around horribly until Kasparov was nice enough to offer it a draw.

Saw the match on ESPN 2

[humblecoder]

Here in the US, the sixth and final match was televised on the cable channel ESPN 2. I was channel surfing and I happened to stumble across it. To my surprise, it was actually quite interesting to watch on TV.

I am not a big chess freak, so I would have guessed that watching chess would be a lot like watching paint dry. However, it was made interesting by the "play-by-play" analysts who were chess masters themselves. They did a good job of explaining the moves, and also the psychology and strategy of chess at the grand master level. It really gave me a lot of insight into what goes on at when chess is played at such a high level.

After the match ended in a draw, they interviewed Kasparov. It was interesting to get his reaction to the match. Basically, his goal for the game was to "not lose", which is why he offered a draw from a very strong position. He didn't want to take a chance of making a blunder like he did in the third game of the match.

It seemed like the key advantage that the computer has in this situation is the fact that it doesn't have an ego to deal with. After losing to Deep Blue in 1997, it seemed like Kasparov was very afraid of losing to another computer in such a high-profile match. That definitely affect the way he approached the game.

The computer, on the other hand, is just calculating moves, so psychology doesn't factor into how it plays. To me, this seems like the biggest advantage that a computer has over a human player.

Also, he seemed to have more respect for this computer program than he did for Deep Blue. Apparently, he had a lot of problems with Deep Blue and how the 1997 match was handled. It could be sour grapes, of course, so I took his comments with a grain of salt.

Re:Saw the match on ESPN 2

[jbrandon]

This was an awesome comment when TheJesusCandle posted it.

Re:Saw the match on ESPN 2

[humblecoder]

No offense jbrandon, but if you check the timestamps you will see that my post preceded TheJesusCandle's post. Actually, I am a little distrubed that somebody would cut and paste my words and claim them as their own. Thank you for pointing that out to me.

Re:Saw the match on ESPN 2

[jbrandon]

Yeah, my bads on that. I thought my settings were on "newest first," but they were on "oldest first."

Human Chess vs. Machine Chess

[Henry V .009]

I think that human chess is still qualitatively better than computer chess. Exhaustion was a big factor in this last match up. The computer didn't feel it, but Kasparov did. Therefore the outcome doesn't tell us much about the level of the chess. If Kasparov could have played fresh every game, my guess is that his chess would have been better.

Human chess has qualities that computer chess still can't match up to. If we were really interested in measuring the level of computer chess we'd try to eliminate for factors such as weariness or stress as best we could. After all, chess is something more than that. We already know that computers will out-endure humans and there is nothing to be learned there.

The AI behind it all

[Faile]

Bruteforce. No really :)
The computer tries every possible move, when it's not using any of the built-in opening books, and predicts which is the most likely to win in the majority of the cases, and plays that.

I found a little more info here, and a nice explanation of the "Minimax Game Tree" that's used when it tries to predict it all.

yeah but...

[danratherfan]

...they're still using brute force. i.e. - it's not the ai that's winning the matches as much as the processing power.

Atari Dies, yet Junior Holds It's Own

[ackthpt]

An earlier post this morning indicated the demise of Atari (Midway West) games unit. Yet here a computer specifically designed to play a strategy game hold even.

It'll be a sad day when ...

June 11, 2012 ARMONK, NY IBM Corporation officially pulled the plug on its computer games divition.

Why did he offer a draw?

[abhinavnath]

Something of a chess novice, I watched this game on ESPN2 yesterday, and I was very confused that Kasparov offered Deep Junior a draw immediately after his rook sacrifice. ESPN's analyst thought, and I agreed, that Kasparov was in a relatively strong position. However Kasparov spent 15 minutes debating that sacrifice. Did he see something nobody else saw? Does anybody here know why he offered a draw, why Deep Junior rejected the offer, and why they agreed to draw a couple of moves later?

Re:Why did he offer a draw?

[tapin]

The post-match interview explained this a bit -- basically, Kasparov psyched himself out in that fifteen minutes, trying to determine if the rook sacrifice was the blunder that would cost him the match, since Junior can capitalize on mistakes better than a human can (or so Kasparov claimed).

Even though he was in a much stronger position, he was spent; worrying about whether the next move would be the move that cost him the match, and made him the two-time world-champion loser-of-a-major-computer-match.

He agreed to a draw a few moves later once Junior et al realized they were in an extremely weak position.

Seems to me it was a pretty wussy way to end it. Junior got lucky. If you're up five runs in the fourth, you still don't pray for rain even if the other team's got a monster closer.

Re:Why did he offer a draw?

[Paradise Pete]

I say if you're going to play the computer, then you do it balls-to-the-wall. Why can't we get a chessmaster who isn't scared to lose to play a computer one of these days?

There are certain types of positions that computers play much better than any human can or ever will. In such positions, computer find weird, unepexpected moves that just happen to succeed for that particular arangement of pieces. These moves seem to come out of nowhere. It is for that reason that when the position is open and complicated even the best players sweat even the most obvious of moves. The tiniest inaccurracy can lead to sudden death.

It was wide-open piece play with king safety issues for both sides. This sort of thing fits a computer like a tailored Armani suit. A human will never win these positions against a computer. NEVER. You can only lose. It simply sees everything. Unless you already have a forced win on the board it is time to start looking for a way to draw before you fall into something nasty.

The above is from this article on game three of the match.

Re:Why did he offer a draw?

[Splart]

Actually, Kasparov's move may have been the best one. Some may say he psyched himself out, but remember, just 3 games before, in game 3, Kasparov was in a similar situation -- tactically superior board position, late in the game. Most agreed the win would be his, but Junior found a clever combination the humans didn't see, and stole the win. While a draw is less dramatic, Kasparov wanted to avoid loosing in game 6 against a computer for the second time after his original Deep Blue match up 6 years ago.

Speaking of which, some might ask why if a computer could beat Kasparov 6 years ago, why didn't it thrash him this time, when computers are over 10 times faster since then, the answer lies in the building of the computer. The original deep blue was built on highly specialized hardware just for rapid calculation of chess moves. Such specialized machines were all the rage when computers were trying to beat humans, but after Deep Blue Sr. beat Kasparov, the interest died out of the market, and the chess programs of today are loaded on normal hardware. In fact, you can buy a very similar version of Deep Blue Jr. for your home PC. Add a buncha processors and memory, and you've got Junior! For more on the computer change, check out Kurzweil's commentary: http://www.kurzweilai.net/meme/frame.html?main=/ar ticles/art0527.html

Last, many people asked why they didn't plan 7 matches instead of 6, so the result wouldn't be a draw. Two problems with this: first is that chess games can draw (no over time), and each side gets half a point, so with 7 games, they could still draw at 3.5 vs. 3.5. Secondly, the player who plays white moves first, which is an advantage, so it's impossible to have a fair match were one person plays white 4 times and the other plays white 3 times. It would be like letting a tennis match be decided on 7 games with one person serving for 4 of the games.

--Splart

Re:Why did he offer a draw?

[pweitz]

I agree with your conclusion, but I think a more apt analogy might be, "You're up five runs in the 8th, your own closer is tired from previous games in the series, and Barry Bonds gets to hit for the next 6 at bats." Maybe your closer will keep the lead or maybe Barry will hit two grandslams.

It will be a sad day

[gorjusborg]

When my CLIE' can take on Kasparov and win.

What fun is it to see brute force prevail over mental finesse?

Processing power

[James_Duncan8181]

It is important to note that this machine is quite a lot less powerful then IBM's one, and has proven to be less effective against human opposition. Maybe the Matrix will take us after all... :(

Faulty Logic Man Strikes Again

[BlackBolt]

Wait a second here...

Kasparov can't beat the computer.
Kasparov is World Champion.
I can't beat GNUChess.
That means I'm world champion quality!!!!

Oh, my God! I'm as good as Kasparov! I KNEW IT!!!

I can't wait to tell mom she was wrong about me. I'm NOT an idiot! Hahahahaha! Take *that*, mom!

Re:Faulty Logic Man Strikes Again

[damiam]

But can you draw gnuchess?

Re:Faulty Logic Man Strikes Again

[BlackBolt]

But can you draw gnuchess?

Oh. Way to crush a guys dreams.... BUT -WAIT-! I'm a Gimp Master! Yes! Yes, damiam, I sure can draw GNUChess, or at least take a way cool screenshot of me losing to it! Haha! I'm on top of the world again, Ma!

Literal Interpretation Man Strikes Again!

I'll never understand

[slashuzer]

this fascination of humans fighting machines, humans vs. machines as it's portayed by the media. To me, as a card carrying nerd, these matches are more like humnas testing machines. And the machines clearly are not good enough. Yet.

loosing game in a blunder ...

[porky_pig_jr]

is part of the whole game, isn't it? And this is where machine has a good potential. A human can improvise but also make the mistakes. Machine follows the program and can't improvise, but it also can't make a 'blunder'. So the bottom line is that the fact that both games were lost 'in a blunder' is no excuse. A draw is a draw is a draw.

Quality of play

[Russellkhan]

"While the quality of play was not outstanding"

Just what are we comparing this to? Isn't Kasparov one of the top players in the world, if not the very top? I've read in some articles that he's considered by some to be the best player ever.

Re:Quality of play

[Scarblac]

"While the quality of play was not outstanding"

Just what are we comparing this to?

Kasparov's own standards. Especially the mistake he made that made him lose one game, as well as the way he was surprised in the opening in game 5, are examples of Kasparov playing below his very best level.

The cliche answer would be to say that Kasparov isn't as good against computers because he can't use his intimidating presence, and he has to be more careful than usual because a computer's style is a good fit to defend against Kasparov's attacks.

On the other hand, Kramnik's cliched image is the exact opposite, and he also drew a computer, so whatever :-)

Re:Quality of play

[murr]

"While the quality of play was not outstanding"

Just what are we comparing this to?

Kasparov's second match against Deep Blue, or his world championship matches.

Isn't Kasparov one of the top players in the world, if not the very top? I've read in some articles that he's considered by some to be the best player ever.

Certainly over the course of his career, he was the best player ever, and he is still #1 in the ratings lists. However, he doesn't seem to be as dominant anymore in multi-game matches as he used to be. In his matches against Kramnik and now Deep Junior, he seemed to play more timidly than he used to.

I also thought that the Kramnik-Fritz match was less than outstanding play. Both Kramnik and Kasparov took early leads playing well. Then, they got careless and let the computers come back, and then they got too timid to press for wins in the remaining games.

The Suspense!!

[Edball]

"Once again Garry played the Sicilian, but this time he chose his favourite sharp Najdorf. Black played an unusual move in the opening, but equalized easily. After the queens were exchanged, Black had a small edge. Trying to play for a win, Kasparov sacrificed an exchange for two pawns. Unexpectedly Kasparov offered a draw soon after the sacrifice, and the computer team declined!"

Is it me, or doesn this read like and edge-of-the seat thriller where the thrill just doesn't come?

I wonder if after Kasparov offered the draw, the computer issued in a deep Quake3ish voice "DENIED."

Moore's law applies to people, too ...

[DogIsMyCoprocessor]

While it is true that computer chess programs will continue to progress, it is often forgotten that human players can progress, too. They don't have very much experience, as yet, of playing a program instead of playing a machine.

For example, the Kasparov - Deep Blue match was extremely short by normal standards. Until Karpov, a world championship match was typically 24 games. Kasparov had very little chance to learn from the first few games and modify his play accordingly. (Also, in such a short match, luck predominates, which favored Deep Blue.)

Please don't underestimate the humans in this equation. World championship level players have improved a lot over the years and will continue to improve.

Re:Moore's law applies to people, too ...

[DogIsMyCoprocessor]

Damn it, I meant playing a program instead of a playing a human. What a maroon!

Doomed! You are all doomed!!

[Joe the Lesser]

Soon, the machines will rise, and with their infinite chess knowledge, will build armys of knights, rooks, and bishops. And my brothers, if we do not repent, we will be the pawns!

Hope he is a better sportsman this time...

[James_Duncan8181]

Last time he said that the computer put him off by constantly humming.

Re:Hope he is a better sportsman this time...

[jdkincad]

Daisy, Daisy, I can't remember any more words

/singing

Re:Hope he is a better sportsman this time...

[fluxrad]

I think you're thinking of Kramnik vs. Deep Fritz and the "shakespeare" incident.

So what?

[xigxag]

I don't see how this is an improvement over 20 years ago.

The board would disappear while the machine was thinking...and sometimes the machine would give itself extra pieces...or it might forget the moves, but still, ZX-80 kicked ass!

ESPN mention

[dmanny]

They had a fairly humorous montage of this match inter-cut with sports footage last night. No idea if anyone would be able to see it again but I found it amusing.

Junior's Mind

[Lailyx]

Junior had some very interesting plays, and it's a shame the programmers won't share what Junior was thinking. I understand their position, but that doesn't keep me from hoping. Especially during game 5: "Junior was happy with its position." is a bit lackluster considering what went on in that round.

Ya...

[RebelTycoon]

But we loose in 2029... So what does it matter...

GNU chess beats me quite easily

[Goalie_Ca]

I'm at that level of chess where i'm thinking just like a computer. I basically try to think of as many moves as possible and rate them. I like to think of computer chess as still being human because we design and program the damn things to do what we want. It's our logic not the computers that is playing with us. It's not as if the computers can think for themselves.... yet.

Re:GNU chess beats me quite easily

[error0x100]

GNU chess beats me quite easily

I had the same experience with GNU chess, and so I just stopped playing against it. They should probably make a broader gradient of levels, i.e. make it so that beginner players have a chance on the lowest levels, and can work their way up. Otherwise, people just quit. It should be useful for people learning to play, not just people who can already play, would probably be more popular than.

Yay

[afree87]

At least now we can be sure the computers won't rebel.

Why six games?

[usermilk]

Why didn't they do an odd number of games, so that there would have to be a winner? Is it that much harder to have seven or five games?

Re:Why six games?

[)Vinvin(]

White give an advantage in chess, so it's unfair to play odd number of games and as it was pointed before, that doesn't solve the problem ;o)

All About Chess

[xQuarkDS9x]

After reading this, I have to wonder how much more advanced will the chess programs be in 5, 10, or even 20 years from now?

Disappointment

[_RidG_]

Am I the only one who is really disappointed by this? Granted, it's better than Kasparov's loss to Deep Blue the last time around, but it isn't exactly a resounding win. To be honest, I don't really know what I'm complaining about, exactly...but I just have this feeling of unease that I can't shake. Who knows, maybe all those "the reign of machines is coming" doom-sayers are onto something? :-/

Re:Disappointment

[xQuarkDS9x]

I think you have been watching too much Terminator movies perhaps? :D

ESPN SportsCenter

[pez]

Did anyone see the highlights on SportsCenter? F-ing hiliarious. Kasparov moved his Queen, then they cut to a full stadium everyone cheering. Slo-mo replays of moves. Tellestrator.

Part of me thought they took it a little too far (poking fun at chess fans), but for the most part it was absolute hilarity.

another draw in human vs computer history

[QEDog]

What other games can we use in the future to test a computer's AI, after this draw in Chess (Kasparov) and Tic-Tac-Toe (War Games)???

Nice Try on the last game

[frank249]

Trying to play for a win, Kasparov sacrificed an exchange for two pawns. Unexpectedly Kasparov offered a draw soon after the sacrifice, and the computer team declined! But two moves later a surprise draw was in fact agreed.

As Tartakover would say: It's always better to sacrifice your opponent's men.

what about GO

[Pecan]

In response to this mention of chess, I'd like to point out Go is a much better something somenthing.

well understood problem

[Jayson]

Game tree search is a very well understood problem and most top programs use some version of a null-window negascout (ID-DFS) with opening and closing databases. The most black magic in these systems is in their heuristic evaluation functions.

Backgammon programs used to compete at only a moderate level until Gerald Tesauro's TD-gammon (and predecessors). I wonder if there will ever be a breakthrough of equal proportions in chess? If so, humans would have very little change against computers (I hate to say never, because of absolute freaks like Marion "I am programmed by God" Tinsley).

Re:well understood problem

[dimator]

I wish I knew what the big deal is about checkers. When I first found out there are checkers world championships, I was shocked. How competetive can it be? I don't see how every game between two people slightly smarter than a pillow doesn't end in a draw.

Re:well understood problem

[watanabe]

I used to think this, too, until Mr. Checkers in Harvard Square gave me a 45 minute lesson. Did you know that you can manipulate who will eventually lose a piece by making sure that you have an odd number of pieces on a column at the end of the move, as long as your opponent has an even number?

The top 16 signs you won't beat a computer at ches

[superspoon]

16. Let's just say that in the movie version of your life, you'd be played by Pauly Shore.

15. Your idea of "conquering Deep Blue" involves employing your gastro-intestinal system to attack the Tidy Bowl man.

14. The computer: A highly sophisticated electronic brain from IBM. You: A highly intoxicated electrician from NJ.

13. Before moving your queen, you insist on consulting Eddie Murphy.

12. Computer: lauded by scientists for its ability to calculate millions of chess moves per minute. You: lauded by fraternity buddies for your ability to pass gas and burp simultaneously.

11. You can't make a single move without thinking of huge juicy shrimp.

10. In your circle, "castling" means holing-up in your trailer with an AK-47 and a bottle of bourbon.

9. Your "garlic breath" strategy fails to intimidate this particular opponent.

8. Your populist leanings always result in you inciting your pawns to wipe out their own king and queen.

7. Kasparov's idol: Bobby Fisher. Your idol: Eddie Fisher.

6. The press has nicknamed you "Deep Doo."

5. You plan to use the "James T. Kirk Strategy" -- talk the computer into blowing itself up.

4. Video tapes of you shouting at the ATM are legendary among the bank security staff.

3. Computer: Intel Inside. You: Imbecile Inside.

2. After your move, you slap the computer monitor and shout, "King me, Pentium-breath!"

1. You counter *every* move with a "Smirnoff opening."

from here

Chinese Chess

[scubacuda]

On a side note, has anyone every played Chinese Chess?

(I just discovered it!)

Re:The top 16 signs you won't beat a computer at c

[xQuarkDS9x]

LMAO! :D

More Information

[Resseguie]

Here are more links I collected when submitting this story...

• World Chess Federation
• WCF coverage of match
• Yahoo News - coverage in non-chess expert terms
• Wired News article - why chess isn't as good a measure of intelligence as some make it out to be
• X3D Technology - provided 2D and 3D coverage of the match
• Coverage of games and press conferences by X3D
• How X3D works

Does anyone have a good link describing the programmers behind Deep Junior? All I could find were news articles and press releases. I'd like to read more information about their strategy, search algorithms, etc.

David

Re:More Information

[Shanoyu]

No, but there are some major chess links you should probably know about.

The Week in Chess: TWIC

It's big, it's commercial, it's Chessbase

Re:More Information

[ctid]

Commercial chess programmers (there are about a dozen that I know about) don't say too much about how their programs work. (Deep) Junior is claimed to be a rather speculative program. I think since 1995 either Junior or the German program Shredder has won every world championship for computer chess (some of these were "micro" events).

I met Amir and Shay at the World Computer Chess Championships in Maastricht 2002. They both give a slight impression of not being approachable, but they are generally friendly to talk to. At the Award Ceremony at the end, they both made incredibly gracious speeches. A class act all round, in my opinion.

It isn't the "majority of the cases"

[Jayson]

Most mini-max variations are based around the idea that your opponent will always play perfectly. If there at least one way for you to be mated, the branch is considered a loss (you propogate your minimum score and your opponents maximum score up the tree).

Re:this is dumb

[Edball]

Old news, and the discription tells us nothing.

That's what the article's for silly :p

Neutralizing the computer's advantage

[sandow]

I'd be very interested in seeing a match between Kasparov using a computer and Deep Junior. This would allow him to access an opening move database, and end game database and do enough analysis to avoid blunders.

Kasparov suggested this after his match with Deep Blue. I predict that a computer augmented GM would hold out against a computer opponent for many years to come.

Re:Neutralizing the computer's advantage

[dark-br]

Bah! Just use a hammer!

Re:Neutralizing the computer's advantage

[Zak3056]

Kasparov suggested this after his match with Deep Blue. I predict that a computer augmented GM would hold out against a computer opponent for many years to come.

Considering that the human would have all his normal advantages over the computer, while having all of the COMPUTER'S advantages as well, I'd be SURPRISED if a computer assisted GM didn't win consistantly.

Re:The top 16 signs you won't beat a computer at c

[superspoon]

It's the best I could find...

Wrong

[tpengster]

chess is nice, but most progresses in chess have been due to speed increases in hardware and optimizations, hence allowing the computer to overpower the human with depth of search.

Because search is exponential, speed increases in hardware won't have much effect on search depth. For example, it might take a 1000-fold increase in speed to increase search depth by 2. The real improvements have been in better search algorithms, heuristics, and tuned evaluation functions. Chess is easier than go for two reasons: 1) the branching factor is a lot smaller, so less to search; 2) evaluation is MUCH easier.

Obligatory linkage

[brokenwndw]

There's commentary on the five earlier games (sixth to come I suppose) and some entertaining speculation at ChessBase, makers of Junior.

I've seen that played . . .

[ruakh]

I've seen people play Chinese Chess . . . it was by far the most boring Hallowe'en party ever. As far as I could tell, it was pretty much like normal chess, except that all the pieces were named for animals and instead of looking (even iconically) like what they represented, they were just disks with the appropriate Chinese characters on them. I'm sure different Chinese Chess sets are different - some might even have little statue-pieces that are easier for Occidentals to distinguish - but this one was very boring to see played.

No. Humans have lost ...

[fygment]

... since the program mostly "drew" against _the_ supreme human chess player. Fact is, it would beat 99.99% or better of the population. Statistically it seems to me that it won ... big time.

Anonymous Karma Whoring

[Anonymous Coward]

Check this link:
http://www.intelligentgo.org/en/computer-go/overvi ew.html

Chess as a Sport?

[ackthpt]

I watched this last night on ESPN or ESPN2.

Chess is a sport? I've heard that Contract Bridge has been suggested as an Olympic sport. Hmm. Is it too much to hope for computer games as an Olympic sport? :-)

Commentator: "Jones moves his elf into Manlobbi's shop, the little dog picks up a spear, the tension is incredible, will the little dog drop it in the doorway, has Jones trained the dog eith enough tripe rations?"

John Madden: "I know what a dog would do for tripe rations, and I've tried them myself, they're really good with some fries and ketchup... etc."

Re:Chess as a Sport?

[Megane]

I think Contact Bridge would be a more interesting Olympic sport. Four no trump? WHACK!

Chess Go: In some ways the same: just a hurdle

[CresentCityRon]

Yes Go breaks down in ways that are a mess to evaluate currently. But think about it: thirty years ago a human beating chess computer would be only on an episode of Star Trek. And I mean beating ANY human.

Now computers can hold their own to the top Grandmasters of chess.

If in ten years computers started to gain against Go playing humans I'm sure someone would try to find another game that computers suck in and say "I think computer vs human GameX matches would be more interesting".

Just don't let the last man vs machine game be between John Conner vs machine! :)

Re:Chess Go: In some ways the same: just a hurdle

[Paradise Pete]

thirty years ago a human beating chess computer would be only on an episode of Star Trek

A little more than 30. Even then the best programs would beat 99% of humans, and maybe 10% of local tournament players. And by the late 70's one of the authors of the then world computer champion was predicting a computer would beat the human champ in five years, maybe ten. He was a little off, of course.

Re:Chess Go: In some ways the same: just a hurdle

[Megane]

But think about it: thirty years ago a human beating chess computer would be only on an episode of Star Trek.

Which brings up the question: Why no Go in Star Trek?

Real world (and boring) answer: because nobody involved in Star Trek production in the mid-60's (or mid-90's) had ever heard of it.

Cool in-story answer: because when the Vulcans first started researching Earth culture, they found this incredibly elegant (if you aren't a complete rules weenie) and difficult game. They were so embarassed about not having discovered it themselves (and embarassed just about being embarassed in the first place) that they set about burying the game for all time. It's a major political incident whenever the game is mentioned to a Vulcan.

Re:Chess Go: In some ways the same: just a hurdle

[Myco]

Star Trek: chess++.

Majel^H^H^H^H^HGene Roddenberry's Andromeda: go++

Interesting scene in one of the early season one episodes (Banks of the Lethe, maybe? Nah... I dunno which one) where Hunt is playing his first officer and discovers that he's cheating. Caught, he defends with "well, of course I've been cheating. Haven't you?" Probably one of the best scenes for establishing the Nietzschean character motivation.

I haven't seen any of season 3... is the show still any good, anyone?

Soon, in a theater near you...

[dark-br]

The year is 2003. The world is being taken over by chess playing robots. Our only hope is one man: Garry Kasparov (played by Arnold Schwarzenegger... A tough sell, I know). He has to control his childish temper as he takes on Deep Blue, Deep Junior, Deep Fritz, and (We're In) Deep Shit. Sure, they look like sissy beige boxes, but they're tough. There will be no time to pout, no leaving in disgrace; every move is on the clock (so to speak). In the final scene, Kasparov beats Deep Blue to a pulp with a Louiseville Slugger. So much for strategy! Astalavista baby!

Humans Hold Off the Machines? (we haven't won yet)

[skydude_20]

Isn't this the current score of recent matches:
Machines (1-0-2), Humans (0-1-2)? I mean we haven't even won yet.

Re:Humans Hold Off the Machines? (we haven't won y

[Shanoyu]

What you mean is that we haven't won recently. If you're claiming that Tal and Grandmasters of yesteryear would have had any problems tearing a 1970's "silicon^H^H^H^H^H cellophane monster" limb from limb then you're hallucinating. Furthermore, A computer will never actually be able to play chess, because we can't teach it to program it's own openings. We still haven't gotten past playing the opening book for it. All computer chess proves is that computers can make a lot of calculuations, and fast. (Shocking, huh?)

tired of these one-on-one showcases

[frankie]

It's clear that top-rated chess machines are now roughly on par with top-rated chess humans. So let's quit having these head-to-head week-long grudge matches.

Load each software on really nice computers, put the computers on wheeled carts with a UPS battery underneath, and allow them to compete in ordinary chess tournaments just like the humans do.

Then we'll see whether these programs can handle multiple opponents with different strategies. They'll build up ranking points, and a match record that can be analyzed. Let's see how long it takes for a machine to work its way up, win, and successfully defend the world title.

Aside from silly notions of player pride, why not?

Re:tired of these one-on-one showcases

[Shanoyu]

Because if a player paid a tournament organizer $30 to participate in a tournament and found that he was playing in a field of chess computers he would quickly demand a refund and leave; we can do that crap at home. Computers have their own leagues. (That's how Deep Junior was chosen to play Kasparov.)

Re:tired of these one-on-one showcases

[Myco]

So you endorse the policy of "separate but equal," then? Trent Lott would love you. ;)

Re:No. Humans have lost ...

[Shanoyu]

I dunno, i'm nothing special, (989 Blitz elo on FICS), but I could do something special with Deep Junior's opening book at a tournament. Considering the fact that I can make the first few moves of a chess game without human prodding to make any specific one, I think my abilities far outstrip DJ's. I mean really, computers still can't even play chess yet, and it's doubtful they ever will be able to do anything more masterful than DJ's greek gift sacrifice for a draw out of nowhere in game 5, which was indeed impressive.

Draw? WTF?

[bobdotorg]

WTF? How can a championship end in a draw?

Who was officiating? Bud Selig?

In NYC. I mean, how un-American is that?

Oh wait...

Interesting comment.

[sven_kirk]

As I was wathing the match the other day, one of the commentators said that,"Deep Junior only made a few moves on its own in a previous (5th?) match." And, "that it had basically run 19 moves by program, and the rest 'on its own'" So in my opinon, yes he fought a computer, but he lost to a program. Not a truly "free" thinking computer that has to learn how to play.

How chess has gotten more advaced recently

[feelyoda]

In a recent grad AI lecture, the professor elaborated on why chess-playing computers has improved so much in the past 20 years. Here is a synopsis: in the 60's a 'mini-max' or 'alpha-beta' pruning algorithm was developed to enhance a depth-first-search by not expanding nodes which would not be reached because a better node for black or white would have already directed the moves differently (assumes that the opponent is playing optimally). Since then 5 things have improved: 1) 'alpha-beta' pruning works better when the nodes to be expanded are sorted in the order of best current parents. This way, in expanding the best moves first, more branches are pruned. 2) hardware has drastically improved the speed of position analysis to the point where constructing a node actually takes more time than analyzing that node's position 3) Many more moves (both opening and closing) have been put into a data base that is sufficiently large to influence the game. In some cases the computer will take 200+ moves to win, at which point the human has no chance of winning. 4) Different position evaluation functions have been developed for the beginning, middle, and end of the game. These functions are also often tuned to a corpus of games for a particular player (e.g. Kasparov vs. Deep Blue) 5) The strongest algorithmic change has come in 'selective deepening'. Suppose we are performing a 10-move DFS. If a particular leaf looks to contain a lot of potential for change, i.e. many points of attack or trade, than this node will be expanded another 10 levels. Clearly the computer has to be highly selective in choosing which leaves to expand, but this has proven highly effective: check-mate has been seen 25+ moves in advance.

Yes

[Russellkhan]

Absolutely ;)

Problem...

[Kaz Riprock]

The computer, on the other hand, is just calculating moves, so psychology doesn't factor into how it plays. To me, this seems like the biggest advantage that a computer has over a human player.

One problem with this "advantage":

Shall we play a game?

A human was moving and analyzing strategy

[adzoox]

I watched some on ESPN 2 yesterday. What was interesting, is that a human moved the pieces. I mean they couldn't design a magnetic keyboard where the pieces moved individually (squares placed far enough apart so pieces could slide between others) - I also noticed that the Israeli programmer of the machine was analyzing attack strategies, not the computer (necessarily) - after the first four moves the programmer typed in about (appeared) over 100 characters into the computer. Announcers said there were over 2.2 million games in the database, well, why couldn't the computer figure out the strategy?

The Commentators did Their Best

[radicalaxis]

Okay, so chess is the newest 'sport' ocvered by ESPN. I saw the match on TV and was reasonably interested, however my favorite part was making fun of the commentators. I should give them their due -- they were professional chess players, and they had to keep people from changing the channel during Kasparov's 15-minute wait considering the rook sacrifice the commentators considered obvious. I suppose it is to be expected that people commenting on a chess game being played on ESPN would make lame sports metaphors (though I was waiting for Kasparov to be compared to Tiger Woods) and feeble attempts at explaining the game to random channel-surfers "the ranks are the horizontal rows, the files are the vertical ones", but the whole speculation about whether the computers would come invade our living rooms after they had finished beating us at chess -- really! I was interested by Susan Polgar's cameo appearance on the show (though not a chess player I admire women who succeed in traditionally non-female fields) and the discussion of the impact on computers on the number of chess prodigies nowadays (computers don't worry about being humiliated by a 12 year old, making it easier for youngsters to find opponents -- or so they say). Really, I'd like to see some computer really trounce all the humans out there so that we can get over this computer chess thing and on to something more serious.

To heck with chess....

[TitusC3v5]

...I wanna see him beat me at monopoly.....to buy or not to buy? :-)

Re:To heck with chess....

[adzoox]

You always buy a piece of property, unless there are houses that need to be placed before the next turn. You mortgage all property except monopolies and railroads AFTER the opponents path. You always buy St Charles, Illinois, and Boardwalk, and New York, You fill/hotels houses on Baltic and Mediterrainean as soon as possible if owned. You always buy if you own the opposite corner, even if you have to sell back or mortgage. :)

I will be worried when

[JeanBaptiste]

the computers can beat the top human kickboxers...

The truth behind man machine matches ...

[rhytnen]

For those who are just speculating and don't really understand chess per se ... you should understand one facet of the match that has been completely overlooked by the media. Kasparov outplayed the machine in ever single game, with no exceptions. In game 2, he reached a won position and blundered to a draw. In game 3, he reached a won position and slid to a draw only to blunder to a loss. In game 4 he reached a fighting position that ended in a draw because he was gunshy from the last game. In game 5 every copmuter engine to do analysis besides deep junior will claim white is better there. WHen a computer sacrfices material against you though... its very hard to play. He was right to accept the draw. He had absolute winning chances with no chance for a loss in game 6 and should be ashamed of himself for drawingr just so he wouildn't lose. The computer consistently played horrifyingly bad chess. The only thing that gives it strength is it doesn't tire like a human and dosn't make tactical mistakes ... period.

Re:Its FAKE!

[rhytnen]

Wow! You're a complete idiot. There is nothing else to say...except maybe that you read to much of Bobby Fischers paranoid ramblings.

Is this really a fair match?

[goatmin]

If this were really a fair test of intelligence, all factors other than intelligence (that could be) would be factored out. This probably has already been broached, but why not limit the power available to the computer, by way of an agreed upon battery size, and thus simulate the tiring of a human. If the computer's program "chooses" to do deep searches early it will pay the price by becoming tired earlier and have less left over for the end game.

Chess (and almost any other games humans play) have multiple factors that detrmine the outcome. We humans also try to level the playing field by setting weight limits in boxing, ensuring both teams use certified equipment and drug testing to prevent an unfair advantage.

Put a throttle on those cpu cycles and I'd put my money on any Grandmaster!!

We didn't hold them off

[TerryAtWork]

In Chess, anything but a big human victory is a big human defeat.

I remember being at Humber College in 75, one of the programmers there was a rated Expert named Barry Sax who assured me he could defeat any program and he could, in 75.

Then IM David Levy won his bet in a match against chess 4.5 at the CNE which I could have gone to but didn't to my regret, but that was it for humanity.

Walter Browne become the first GM to lose to a computer when someone brought a portable chess machine to a simultaneous exhibition of his and beat him with it (He WAS playing 100 other people at the same time...)

The Ken Thompson's HiTech came to fruition with Deep Blue (Is there ANYTHING this guy didn't touch for the better? Only Claude Shannon was more awesome.) and Kasparov was beaten - and folded like a bully who finally lost a fist fight.

And NOW you can buy a multi CPU box that plays at 2800 strength. That would have crushed Barry Sax.

In ten (Maybe five) years computer will be rated 3000+ and the championship will be between them exclusively.

Well, there's always Go and Taasen.

Mostly agree

[phr2]

I'd say computers play openings exactly as well as the people who program them, since they're playing out a book and the book is simply entered into the program by the programmer. Well, Kasparov is the strongest player in the world partly because he's better at openings than anyone else. As much as 50 Elo points of his strength (out of his ~2800) has been credited to his opening preparation. He is possibly the world's best opening theorist in his own right, and he also pays a team of players to constantly sit around analyzing openings for him and feed him new tricks that they find. So it's no surprise that he beat the computer in the opening.

You're completely correct about endings--computers play a limited class of endings perfectly, those they have databases for (5 piece and some 6 piece endings). If you look at a Rubinstein type of strategic ending with knight+bishop+2 pawns vs 2 knights and 3 pawns or something like that, computers aren't especially better off.

F-H Hsu (designer of Deep Blue) claims there are still tactical positions where humans do better than computers, because humans sometimes know to keep searching when a computer might think a position had run out of possibilities. Hsu's invention of "singular extensions" was designed to combat that effect.

Finally there's a famous chess cliche that sums up what we're all saying here: a great player plays the opening like a book, the middlegame like a magician, and the ending like a machine. Computers are good at playing both like books and like machines; they still have to work on "magician".

Who selected "chess" as a proxy for intelligence?

[NotQuiteReal]

So what if a computer can/cannot beat a human player at chess.

Open question: What would be a better test (to see how "smart" computers have become)?

Computers suck (my professional opinion).

Of course, humans suck too, so it's a toss-up.

oh please

[Lelon]

Humans Hold Off the Machines... For Now

I'm sorry but that is simply a joke. Kasaparov was beat soundly five years ago. If the developers of Deep Blue had continued development and upgraded to today's hardware, they would probably have a machine unbeatable by human players. The battle has been over for 5 years.

That being said, I hope that this new tourney structure (which the entire world of chess has been sorely lacking, its worse then college football) will encourage development (one of the reasons IBM quit development was that Kasaprov would not guarantee a match, which was later reinforced when he ducked several challenges by the head developer of Big Blue after he aquired the rights, see Behind Deep Blue)

I was suprised and overjoyed to see the game broadcast on ESPN2. Hopefully this will bring Chess into America.

Re:Its FAKE!

[ekw]

Looked real enough to me. Mind you, there was some noise (17...Rfd8 18. Kh2 Re8 is dodgy or at least dodging). I think GK saw that DJ's position was static with no improvement, and was finding optimal piece positioning himself. There's the human angle: computers just say "these are the best squares," even based on seeing some fairly static moves into the position, while humans could still do a better job of weighing piece positioning after tactical sequences that don't actually win material and thus get pruned out. I'm explaining 23.Nb3 I guess, and also how e8 is in fact the square for the rook (guarding e7).

Still -- in the absence of a real variation, I would indeed like to know why he didn't play on after 28...f6 followed by ...Kf7 (29.Nd4 g6).

Gain vs. Risk.

I'll be impressed....... Maybe!

[Flyer]

I'll be impressed when, after the game, Kasperov can toss Deep Junior his keys and have Deep Junior go get the car.

We keep forgetting in all the media fluff that, while this is technologically interesting, it is not even scratching the surface of human versatility. Nor do we notice the man behind the Deep Junior curtain. Or should I say programmers?

Additionally it could be reasonable argued that Deep Junior is not playing Kasperov but Deep Juniors programmers are playing against Kasperov with extreme computer assistance.

quality of play not outstanding?

[naejulak]

anyone or anything that can tie Kasparov at all has, er, outstanding quality of play.

Better headline

[Spy Hunter]

"Chess Masters Hold Off Computer Programmers... For Now"

Still waiting for the perfect play...

[mr3038]

I think we should already forget the chess and move along. It's only matter of time until somebody (or should I say, something?) comes up with a perfect play of chess. My guess is that it'll be a move sequence that quarantees a draw for the white. After that, once you get black, you have no change to win.

The only question is how long such game would be in turns. That's imporant because if that's "long" then finding even one will take quite some time. IIRC there's a joke in Futurama where robots are playing chess. There's a chess board with all the pieces in the starting positions. The winner only says "Mate in 143 moves" and the loser says "Oh man, you win again". I'm afraid that joke is closer to reality than many of us would want to believe.

I'm pretty sure there's perfect play of Go, too. Finding that will take so much time that I don't need to worry about that. I'm not that sure about chess.

Re:Kasparov is a fake

[outsider007]

if they were really scripted I would think they'd throw in some theatrics, such as garry breaking his chair over kramnik's face, then you'd see americans take an interest in chess.

Unimpressed...

[javabandit]

As a chess player, I can say that this entire match is completely uninteresting to me. What does this prove?

Chess is a game without luck. Not unlike Tic-Tac-Toe. Chess is a game where the better player is measured solely based on their ability to see many permutations into the future and pick the best outcome. In chess, there is plotting and strategy involved, but if you are playing an opponent who can see further into the board than you can... then you will lose. There is no ENTROPY in chess at all. Which, computationally speaking, makes it about as impressive as Tic-Tac-Toe.

There is little doubt that computers will routinely beat the best grandmasters. Because computers take any and all skill out of the game... and reduce it to its most simple level -- seeing all the permutations.

This doesn't impress me at all. This impresses me about as much as a computer calculating huge prime numbers. The raw computing power is impressive, but the actual computing process itself is very unimpressive.

Seeing a computer beat a human at chess is about as impressive as seeing a computer calculate a series of numbers faster than a human. Very unimpressive.

When computers use REAL skill -- balancing luck/chance/entropy against strategy... then I will be impressed. I want to see a computer beat master poker players at the World Series of Poker. I want to see a computer beat a professional billiards players. I want to see see a computer win at games such as Bridge.

The fact of the matter is... when you introduce any kind of entropy into a game... the computer falls apart. Hell, even Deep Junior was flustered quite often when Kasparov would make totally retarded/unprofessional moves.

A lot of Kasparov's strategy was about introducing his own "entropy" into the game... i.e.... make moves that the computer wasn't programmed to expect he would make. Kasparov made some unbelievable moves with his pawn fronts that would have easily cost him the game to other human grandmasters.

Nothing impressive here. Next.

Re:Unimpressed...

[Minna Kirai]

By Game Theory (the field of research that creates chess solving programs), games do not have luck by definition.

Randomness doesn't increase the interest of games, except when both the players and game are poor (because a little luck hides the fact that the game sucks, and keeps the players interested).

Survey the non-athletic games which command significant audience interest. Chess and Go have hundreds of times more fans than Poker and Billiards.

(Even the popular Korean game, Starcraft, is intentionally nonrandom so that top-level players can interact fairly)

The fact of the matter is... when you introduce any kind of entropy into a game..

Entropy in game-like subjects can mean 2 things: either there is something influencing the game which the players cannot predict, or that the current state of the game is too complex for either player to fully understand where all the pieces are. Monopoly is an example of the first case, football of the second.

In the real world, taken to the limit, those two viewpoints are equivalent. But in game programming, they have very different effects. Injections of formalized randomness to the game (dice rolling, picking cards from a deck) are simple for a computer to deal with- just expand the event tree with new, weighted branches (of course interesting herustics can come in to keep the size from ballooning).

However if the game state is too large for the computer to input or store (the entropy comes from facts which are theoretically knowable, but aren't known), then the computer programmer is in real trouble. Classical game theory is out the window then, and you've got to try more Strong AI approaches.

Poker is a special case of the latter. (Although it includes simple random draws as well). The interesting skill in poker is psychological analysis of the other players, to deduce what random-draws they have been given. It'll be a long time before a computer can cope with that (although infrared cameras to collect autonomic reaction data will help a lot)

Human intelligence has not suffered..

[juushin]

Surely a chess computer is only as good as the person who programmed it?

This comment is dead on correct. No sorrow should be shed over Garry tying Deep Jr. A failure to win this series should not be translated into a blow to human intelligence by a computer. These games are, as was correctly pointed out in the above comment, simply about intelligence in a human vs human context. The computer runs iterations. Without a human to cleverly, and with the aid of a handfull of chess masters, program the computer, there would be no worthwhile matches. Although I am not a huge fan of Garry Kasparov, I commend him on keeping his cool this time around.

Re:Human intelligence has not suffered..

[gerardrj]

I disagree. For several reasons.

First, I think a programmer can create a program with capabilites far beyond what the programmer has. For example, the programmers who write code for the F-16 are VERY unlikely to fly the plane as well as the computer does. One may know how toodo something, but doing it well might mean doing it very quickly, very long, very accurately, or under extreem stress. For example. I know how to add, but my computer is far better at adding than I am. It can sum hundreds of colums of 12 digit numbers in a fraction of a second, whereas it might take me 30 seconds to sum 5 four digit numbers and double-check for accuraccy.

Second, without human teachers Kasparov would never have learned chess. Kasparov was "programmed" to play chess at some point also. There is nothing inherent in the human genes or brain structure that makes us know the game of chess as instinct.

At some point, Kasparov became a better chess player than his teachers, just as these chess programs will learn from experience and will (or have) become better chess players than their teachers/programmers. I would seriously doubt that the Junior's programmers could alone, or collaboratively take on Kasparov and come out with a draw as Junior did. But lets put out a call for that: Each of the chess masters responsible for programming Junior should take it on for a match, just like Kasparov. Will the computer be better than the ones who programmed it?

What would be a very interesting match would be to take a chess program written in the spirit of Seti at Home, or Distributed.net with several hundred thousand computers, each running through one move chain and ranking the outcome and reporting back to the master server(s), and pit it against a collaborative team of the five best human players in the world.

... brute force? yes and no...

[missing_boy]

Well, of course the computer is guilty of brute force calculation: it's only a machine! Of course, it could be argued that we're simply "brute-force" thinkers, as well, if you include emotions as being results of some sort of optimization of expected outcome.

However, Kasparov said himself that the computer's sacrifice of the black bishop on h2 to White's King (Kasparov) in game 5 was a shocker, totally unexpected and very human-like and creative. As far as I understand, Kasparov saw this move as a sign of something more in the computer than just "brute-force".

It is also a fairly accepted view among chess players that Deep Junior is a master at taking chances: "it walks on the edge of the cliff quite often, but seldomly falls off".

The games between Kasparov and Deep Junior were very interesting, perhaps with the expection of the ending of the 6th game. Check them out at www.worldchessrating.com!

specification of the deep junior computer

[owlstead]

Just out of interest I was looking up the specifications for Deep Junior - that is, the specs of the machine that played Kasparov.

Unfortunately I couldn't find anything, nothing on the game sites, nothing through google. I found out that it is created to run on a multi-processor machine, and that it probably is running on two processors.

Does any slashdotter know?

Cheers,
Warper

Re:The top 16 signs you won't beat a computer at c

[groove10]

Oh I wish I had some mod points... I'm sure you stole this from somewhere, but it's still damn funny to me.

I'm not impressed with the AI yet...

[Man In Black]

It's impressive that computers can compete with humans on the grand master level... but frankly, why haven't we gotten to the level where computers can *destroy* human players every single time? Is it simply a matter of computers not being fast enough, or not being able to search enough moves? Eventually, computers will get to the point where they can anticipate EVERY outcome. After all, there's a limited number of moves possible in a chess game (albeit a VERY high number), so what happens when we reach that point? I'm starting to think that chess itself just isn't a hard enough problems to pit human intellect against computer intellect, since people like Kasparov are just so damn good at it (Anyone else remember that Star Trek episode where Data plays Stratagema (sp?) against some alien, and can't beat him... so forces him into a stalemate instead?). As always, people have already pointed out Go as a better solution to strategy competitions, and eventually, I think that might be the only answer. Obviously, computer Go programs are rather primitive... but it stands to reason that human Go players aren't at the top of their league either, since the game receives much less attention than chess does I'm sure. By the way, why is it that when people talk about Go programs, it always comes up against a wall of "too many possibilities" for the computer to calculate? Obviously, humans can't calculate all these possibilities, so shouldn't it be possible for a computer program to think differently instead of brute forcing things? Perhaps things like pattern matching would be more effective (I don't know how to play Go at all though, so maybe I'm just talking out my ass).

A bit of a different topic, but are there any high-level chess programs that *don't* use opening books and endgame books? Maybe it's just me, but it seems like cheating for a computer to just memorize what people have already done. It would be much more impressive if the computer was able to improvise it's own strategies, even if they ended up being the same. I'm sure it's a lot more difficult to program well though...

Not just brute force

[ToLu the Happy Furby]

One thing people don't seem to realize is that Deep Junior was a hell of a lot slower than Deep Blue, even though the Deep Blue match was over 5 years ago. That's because Junior was running on "ordinary" commodity hardware (8-way Xeon I think?) while Blue was a collection of hundreds of custom ASICs that do nothing but calculate chess moves. End result: Blue could search and evaluate ~200 million positions per second, while Junior as configured in this match "only" did ~3 million.

Nonetheless, Junior was almost certainly the better player. For one thing, the terms of the Deep Blue match were heavily tilted against Kasparov: he didn't get a chance to play against Deep Blue or even examine any games Deep Blue played before the match. For another, strategies of "anti-computer" chess are far more developed today than in 1997, when they barely existed (after all, the only way to build a world-class chess playing computer in 1997 was to build a supercomputer out of custom hardware). Third, Kasparov screwed up much more seriously in the 1997 match--one game he accepted a draw when he in fact had a provable win, apparently because he trusted the computer's evaluation of the position, and on several occasions he made terrible blunders.

Indeed indications are that even the normal Junior program on a decent PC plays the Kasparov-Blue games better than Blue did (except for a couple especially "brilliant" moves on Blue's part).

Not only is Junior (marginally) the best computer program available today, it is by a good measure the "most human-like". That is, it is still makes its share of "non-human" moves (although far fewer than Deep Blue), and still has no clue how to analyze certain positions, but its evaluation function has a much better understanding of position and is thus more willing to initiate complex piece exchanges than the other major programs. This showed up several times during the match, in two outstanding moves in particular (which netted come-from-behind draws for Junior in games 4 and 5).

Kasparov, by taking Junior out of its opening book, was able to exit the opening phase with the initiative in every single game. But in every game except for 1 and 6, Junior managed to draw even. In games 2 and 3, it was by virtue of outstanding tactical defense, which should have forced two draws (except that Kasparov screwed up the end of game 3 and lost). In game 5 it was with a shocking bishop sacrifice counterattack (extremely uncomputer-like) which nullified Kasparov's last turn as white (which is an advantage).

Perhaps game 4 is the best synopsis of the state of Junior's play. Kasparov played the opening perfectly according to plan, and ended up in a classic anti-computer position. Normally you can then just wait for the computer to screw up and then rip it apart. But while Junior did make a couple useless non-human moves, for the most part it played extremely well. Kasparov kept waiting to have an opportunity (perhaps too long), until finally Junior broke open the position by initiating an excellent multi-piece exchange. Kasparov had to settle for a draw.

OTOH, Junior had no idea what was going on in the endgame. Kasparov had a provable draw after move 47, but Junior, having no idea how to evaluate the position (no computer program does), thought it was winning. It played on until move 61 before the embarrassed programmers overruled the program and took the draw.

Overall, it seems that computers still have a ways to go before they can pass the Grandmaster Turing test. And it seems Kasparov really did just chicken out by accepting the draw in game 6. But the fact remains that Kasparov played quite well (for the most part) and simply couldn't convert his advantage in any game after the first. Meanwhile, while it still made a few computer-like moves that stuck out like sore thumbs, for the most part Junior played very solid chess with occasional strong speculative moves that would be aggressive even for a human.

It all comes down to physics

[Zelig321]

It's all about computing power. Surely you can cram more and more hardware into the computer to a point where it becomes better at thinking than you.

The real question is if every thought, emotion or human reflex is (or is not) a series of very elaborate computations. I believe so, since the brain is ultimately made up of matter (complex molecules interacting together, probably much like computers, only in subtle ways we are yet to understand).

This means its only a matter of time before machines act exactly like humans. Just give them enough transistors, memory and circuits and they'll perform as well as we do at anything.

Re:The top 16 signs you won't beat a computer at c

[superspoon]

I did steal it. It sais so at the very bottom.
And thank you very much for the compliment :-)

And now for a musical interlude...

[SomeGuyFromCA]

Seeing the nth story on /. about computers and chess, I have to post this now:

Artist: Moxy Früvous
Album: Live Noise
Track: Kasparov vs. Deep Blue (Recorded live at MIT)

(Murray) Well, I... I do have a question. How many people here were voting for Deep Blue? And how many people were voting for Kasparov? Ah.....humanity has hope - still, I suppose.
(Jian) How many people are like actually disappointed that the human lost. No no, disappointed I mean. Duh! No, Because like I just don't get it, you know? I mean, you know? What's the f*cking big deal, you know? It's a machine, right? I don't know. I made the point in Albany the other day which apparently lost on all the Albanians.
(Murray) I didn't get it either. [laughter]
(Dave) That's not all that was lost on the Albanians.
(Jian) They're still behind the times.
(Dave) There's a lot of foreign aid going on there.
(Murray) Your point was if there's a fire, Deep Blue wouldn't run out of the room.
(Jian) Exactly!
(Mike) Couldn't run out of the room.
(Jian) That's exactly my point. If an attractive person walks into the room, a person that would be attractive to Deep Blue, it can't do anything about it. That's my point. Kasparov can approach the person.
(Murray) The attractive person.
(Jian) No! Here's my point. My point is a calculator. That's my point. Right?
(Murray) No, let's get back to the fire.
(Jian) No, hang on. No, no, the calc...forget the fire, because apparently it's, you know, I'm talking on a different level.
(Murray) I - Clearly!
(Jian) Here's the thing. Here's the thing. A calculator, right, a common everyday calculator.
(Murray) I'm with you.
(Jian) A calculator will, you know, it...let's say, let's play the adding game, right? Who can add faster: a calculator or a woman or man? A calculator can, right? So what's the big deal? We know that there are instruments... we know that there are machines... we know that there are computers, etcetera.
(Murray) Right.
(Jian) that can do things that. It's just because the thing won at chess, right? I don't understand what the big deal is.
(Murray) Your point is if you light a match near your calculator, it's not going to scurry away. It's all relative.
(Jian) No, my point is...My point is if there's a calculator. My point is... oh alright, okay, I'll bring it back to the fire for you, because I know you're obsessed. If there's a fire in my living room, where me and my calculator are sitting, I can escape the fire.
(Dave) Yeah, but if uh...
(Jian) But my calculator can't.
(Murray) Is there a logic course here that one of us can enroll in? [audience laughs - this is MIT...
(Jian) Well, I think, I think they know what I'm talking about. I'm talking about the fact that the machine is programmed to only do one thing. It can't do anything else. The fire was just one example. Pick anything, anything.
(Mike) Locusts.
(Murray) A flood. How about a flood? Can he escape a flood?
(Dave) Buddy boy...
(Jian, laughing) Kasparov can....
(Mike) A plague of frogs. [laughter]
(Jian) No, say there's an, say there's an earthquake. Right.
(Murray) Now, there's a good one.
(Jian) There's an earthquake down the middle of the room, the chess room. Kasparov can get up and move. Deep Blue can't.
(Murray) It falls into the chasm.
(Jian) That's my point.
(Murray) Right.
(Dave) But if they built Deep Blue in a door frame then there's no room for Kasparov to stand... to fight the earthquake. Then they're doubly screwed.
(Jian) See...see...they'd have to program Deep Blue to escape the fire. That's my thing.
(Murray) But they can do that in a couple of years.
(Mike) You know we were talking about... we were talking about disaster movies. This would be the perfect disaster movie. Just have an endless succession of these scenes where Deep Blue is just sitting there. "It's the locusts" or whatever and Kasparov is just running his little piggy legs out of the room. "I'm free again, you f*cker!"

[sorry if this sucks to read - the lameness filter wouldn't let me post it with a blank line between speakers]

Re:And now for a musical interlude...

[Myco]

Right, except that MF is an extremely awesome band of four rather hoopy froods who would be more than happy at the exposure generated by the parent post. Go look at their website. Tons of links to fan pages, lots of which certainly post lyrics, tabs, and the like. Früheads approach deadheads in their fanatical band-loyalty, and like the GD, MF know how to support this enthusiasm rather than pointlessly Metallica-ing it away with hissy fits.

Come up with an original thought, dude

[humblecoder]

They say that imitation is the sincerest form of flattery. Nevertheless, I would appreciate it if next time you at least come up with an original thought.

If you don't know what I am talking about, see the following post...

http://slashdot.org/comments.plsid=53202&cid=525 96 61

Who knew that Jesus was such a Karma whore....

Re:Doomed! You are all doomed!!

[machine of god]

if we do not repent, we will be the pawns!

Yeah, but if you do repent, you will be the pawns too. Or at least you would be if I were your evil overlord. I would convince you by saying that I plan on sacrificing all of theirs, so it's better to be one of mine.

Why o why???

[jotaeleemeese]

Every time the topic o chess surfaces immediately there is a "Go is great" loser pushing they favourite game?

Yes, we know it is great, it is the best game ever invented by the human race.

Now, can you keep that information to yourself while talking about chess?

Jeeez.

So what?

[jotaeleemeese]

If it plays like a human, wins like a human, and for all purposses is close to better the human (in this certainly narrow area), why should I give a fuck about how this is achieved?

If you did not know how this is done, what difference would it make as long as the final result (machines bettering human performance in hthis game) is the same?

Problems with computer chess

[vijaykittu]

It is sure that the super computer must have beat the world champion but i still say that Deep Blue isn't great - it just does a good amount of brute force attack and does not have a thinking of its own.

We need to develop supercomputers that does not use brute force techniques but use artiticial intelligence.

Furthermore, Deep Blue's chess program isn't so great. If you observe the steps of deep blue and the chess champion on a simple desktop chess game such as Fritz, you can observe that most of the steps played by both are same.