Anand-Topalov

[gwnn]

Why do they play the same opening over and over and over? d4 d5. really weird...... And just Slavic and Catalan (I think).

[Jlall]

Seriously epic game, and topalov evens it up.

 

han: 11:43 am

(11:43:50 AM): why doesn't he surrender like a man?

jlall :11:48 am

(11:48:43 AM): what

(11:48:46 AM): he resigned

(11:50:14 AM): lol

 

lol, hanp rules!

[PassedOut]

This is a great match. And being able to follow it on the web makes me stop to appreciate modern technology once again.

 

When I was a kid, I had to wait for a newspaper to print the moves, and sometimes they even messed that up. Now that I'm thinking about it, I also realize once again how much I appreciate the vugraph presentations on BBO...

[Aberlour10]

My interest in "Big Chess" is rather marginally...but I would like to know...Do todays champs have still big stuff of own analysts ( like WC's in 70's 80's etc.) or were they replaced by computer simulations in last decades?

[hanp]

Seriously epic game, and topalov evens it up.

 

han: 11:43 am

(11:43:50 AM): why doesn't he surrender like a man?

jlall :11:48 am

(11:48:43 AM): what

(11:48:46 AM): he resigned

(11:50:14 AM): lol

 

lol, hanp rules!

Weren't there some FU shut ups in between?

[Jlall]

My interest in "Big Chess" is rather marginally...but I would like to know...Do todays champs have still big stuff of own analysts ( like WC's in 70's 80's etc.) or were they replaced by computer simulations in last decades?

They use computers AND humans for preperations, and a lot of both!

 

Everyone knows that computers are better players than humans now, but not everyone seems to realize that a computer with human assistance will always beat the top computer.

 

But as far as preparation obviously you need human guidance there.

[Jlall]

Seriously epic game, and topalov evens it up.

 

    han: 11:43 am

    (11:43:50 AM):    why doesn't he surrender like a man?

    jlall :11:48 am

    (11:48:43 AM):    what

    (11:48:46 AM):    he resigned

    (11:50:14 AM):    lol

 

lol, hanp rules!

Weren't there some FU shut ups in between?

FU STFU

[jdonn]

Everyone knows that computers are better players than humans now, but not everyone seems to realize that a computer with human assistance will always beat the top computer.

In general terms why is this?

[gwnn]

well it was opposite bishops endgame which is supposed to be auto draw almost always, it was great manoeuvreing by Topalov to break the position.

[cherdanno]

Everyone knows that computers are better players than humans now, but not everyone seems to realize that a computer with human assistance will always beat the top computer.

In general terms why is this?

Because humans and computers have different strengths. A human might see a long-term plan that is very promising ("bring the bishop to e7, the rock on f6 and i don't see how black could save this position") and then use a computer to check it tactically ("Oops, if I first bring the bishop to e7 I get mated in 7 moves, so lets try starting with the rock - YEAH it works!")

[jdonn]

That makes no sense to me, which I'm sure is a result of my own shortcomings as I'm even worse at chess than I am at poker. But why shouldn't a computer, analyzing all long term plans, be able to figure out which is most promising?

[awm]

That makes no sense to me, which I'm sure is a result of my own shortcomings as I'm even worse at chess than I am at poker. But why shouldn't a computer, analyzing all long term plans, be able to figure out which is most promising?

Basically the problem is that the computer would have to analyze all long term plans.

 

Human experts are good at narrowing the set of reasonable moves. They can quickly reject certain actions as being bad "from experience" or based on the appearance of a result position, with a high degree of accuracy. Computers are bad at this, and typically deal with the problem by just trying all the moves and progressing down the game tree until the position becomes obviously bad even to an intermediate-level player. This obviously takes a long time, but computers are fast...

 

In any case, combining the human expert ability to quickly eliminate bad moves and positions with the computer ability to simulate down the game tree for a large number of moves is quite potent. Basically, the computer can consider millions (but perhaps not billions or trillions) of positions in a reasonable amount of time (which humans can't do) but the human input helps the computer avoid wasting its time on "obviously bad" positions from which any human expert would quickly resign.

[jdonn]

That makes sense. So in theory computers by themselves will be better than human + computer currently is (or at least as good) if/when they get more computing ability. It would be kind of like if it took a gps system a long time to scan all places and find where I want to go, but I could start off by pointing it in the right direction then let it find the exact location for me.

[gwnn]

BTW looks like Anand threw away a perfectly simple draw in move 54.

 

http://www.chessbase.com/newsdetail.asp?newsid=6307

 

54...Bc6?? And here comes his miracle: Black absolutely had to be able to protect the h7 pawn with his bishop. By being forced to retreat with Kg8, the game is now lost. Tragically, Anand played the only losing move. 55.Kh6 Kg8 56.g4

 

and Anand resigned

[Jlall]

Josh, I would say humans are still better strategically and positionally at least early on. For instance computers often don't like lines involving sacrificing a pawn to gain some initiative/possible attack. Computers like sacrifices that gain material in the (nearish) future, or lead to a mate.

 

Humans can force computers to evaluate things like the pawn sacrifice more deeply etc (or at all).

[jdonn]

That just goes to lack of computing power. If computers don't like a sacrifice without payoff in the near future it's only because there are too many permutations for it to handle in the distant future.

[rogerclee]

Also sometimes even very good humans are just careless.

 

See: http://en.wikipedia.org/wiki/Blunder_%28ch...ladimir_Kramnik

[Jlall]

That just goes to lack of computing power. If computers don't like a sacrifice without payoff in the near future it's only because there are too many permutations for it to handle in the distant future.

Obviously, I don't really get your point?

 

The fact that chess and go are not solved is because of lack of processing power. They are games of perfect information. With infinite processing power, we would know everything about perfect information games.

[jdonn]

My point was it seemed like you were just repeating awm's point. Whatever sorry anyway I obviously see it now, yay me!

 

Did you know I once came in 3rd in my county in high school chess? Frightening. I know of at least one (at the time) 11 year old girl and one 7 year old boy who were both clearly better than I was, not to mention 4 or 5 people who played in the actual event. But that me, who may have been as high as a 1300 player, would slaughter the current me who I doubt is even a 1000 player.

[jjbrr]

Also sometimes even very good humans are just careless.

 

See: http://en.wikipedia.org/wiki/Blunder_%28ch...ladimir_Kramnik

i hadn't seen this before. pretty epic.

[Jlall]

My point was it seemed like you were just repeating awm's point. Whatever sorry anyway I obviously see it now, yay me!

 

Did you know I once came in 3rd in my county in high school chess? Frightening. I know of at least one (at the time) 11 year old girl and one 7 year old boy who were both clearly better than I was, not to mention 4 or 5 people who played in the actual event. But that me, who may have been as high as a 1300 player, would slaughter the current me who I doubt is even a 1000 player.

I think I'm hopeless at chess again (not that I was ever beyond a novice). I gave it a go for like a month but I lost my skillz in the mental hospital :blink:

 

I might do some kind of year long game challenge where I have to reach a certain rating on yahoo in backgammon, hearts, spades, and chess, as well as maintain a certain long term robo dupe average, and also maintain a certain winrate in both limit hold em and no limit hold em or something. Been thinking of the details, I would probably need action on it, but if I did it chess would definitely be the hardest (and I would make it the lowest necessary rating, like I'm thinking 3000 for hearts and spades, and 2000 for chess [2000 yahoo rating not FIDE lol!]).

[jdonn]

I was between a 1400 and 1700 player on yahoo for years. Played over 5k games there, amazingly. Would never have happened if bbo existed. I definitely padded the stats by beating up weaklings, I rarely played anyone even as high as 100 below me lol.

[Jlall]

I was between a 1400 and 1700 player on yahoo for years. Played over 5k games there, amazingly. Would never have happened if bbo existed. I definitely padded the stats by beating up weaklings, I rarely played anyone even as high as 100 below me lol.

I was 1500 and finally breaking out of the beginner room into the intermediate room lol. But I got standard jlall serious about it and did tactics problems on a site non stop, studied from waitzkins software, watched videos on chesstv, bought a database, learned how to convert yahoo games to pgn and saved all my games and analyzed it with rybka/annotated them/etc. Basically got obsessed.

 

Fournier used to watch my games and it was standard for me to be down a pawn out of the opening, then up a piece after the midgame, then lose the endgame. Basically I was only good at tactics/midgame, but I was learning some endgame stuff so that was helping. Fournier= god at endgames btw.

 

Chess is a great game, much respect to people who are good at it.

[Mbodell]

That just goes to lack of computing power. If computers don't like a sacrifice without payoff in the near future it's only because there are too many permutations for it to handle in the distant future.

Obviously, I don't really get your point?

 

The fact that chess and go are not solved is because of lack of processing power. They are games of perfect information. With infinite processing power, we would know everything about perfect information games.

But in both cases it is doubtful the will be solved ever, certainly not with the current approaches.

 

The game tree complexity of chess is somewhere in the order of 10^123. The number of possible positions is around 10^50. To put that in perspective the number of atoms in the observable universe is around 10^80.

 

Even if you were doing more than a million positions each microseconds it would take roughly 10^80 years of processing. That is one year for each atom in the observable universe.

 

So just "faster computers" are unlikely to solve chess.

 

Go is even worse (I.e., more complex) than chess in that there are about 10^172 board positions.

 

Certainly if we came up with a different computing model (I.e., quantum computers or some other fundamentally different approach than exhaustive search and minimax evaluation) then maybe we would solve it. So we can't really say never, although it seems quite doubtful.

[Jlall]

I dunno, people thought all sorts of things were impossible. We fly now. We fly to the freaking moon. I would be surprised if chess never gets solved. I'm sure in 100 years some amazing things will have happened with technology that we cannot even imagine yet.