Total Posts:42|Showing Posts:1-30|Last Page
Jump to topic:

Deep Blue and "intelligence"

dylancatlow
Posts: 12,255
Add as Friend
Challenge to a Debate
Send a Message
11/25/2014 1:55:50 PM
Posted: 2 years ago
In 1997, the world chess champion Garry Kasparov was defeated in a chess tournament by Deep Blue, a supercomputer developed by IBM. Many people came away with the idea that Deep Blue had outsmarted Kasparov. However, if you know anything about how Deep Blue made its moves, this seems rather absurd. Rather than deciding its moves "intelligently", Deep Blue simply generated millions upon millions of possible moves, assigned a rating to each move using an algorithm, and chose the move with the highest rating. Of course, assigning a useful rating to each move is no easy task. Each move in chess becomes a condition of subsequent moves, so determining a "good" move requires an astronomically high number of computations and an algorithm capable of recognizing good moves, as well as some mechanism for considering how moves might influence your opponent's moves, and in turn how that sets the stage for your moves, and so on. So although Deep Blue deserves credit for its computational power, it does not deserve credit for "outsmarting" Kasparov (it merely deserves credit for outplaying him). For is intelligence not precisely that which allows you to avoid blunt strategies when solving a problem? Shakespeare is considered a brilliant writer, not because he produced brilliant works per se, but because he produced brilliant works without needing to produce a trillion bad ones. The same goes for Deep Blue. It did not come up with good moves because it was intelligent, it came up with good moves because it came up with lots and lots of moves, some of which were good, and was able to recognize the good moves as "good" only because it had the computational power necessary to "force" the answer. The fact that Deep Blue came out barely ahead even though it had the advantage of choosing its moves from a pool trillions of times larger than Kasparov's just goes to show how much more intelligent a player Kasparov really was. Kasparov had to make his moves on the basis of far less data, and was thus only able to compete because he was able to make better use of that data i.e., because was able to more intelligently interpret the data. Deep Blue made up for its profound stupidity by repeatedly throwing itself against a wall until finally it broke through, while Kasparov simply climbed over. That is the difference.
bladerunner060
Posts: 7,126
Add as Friend
Challenge to a Debate
Send a Message
11/25/2014 2:18:03 PM
Posted: 2 years ago
At 11/25/2014 1:55:50 PM, dylancatlow wrote:
Deep Blue simply generated millions upon millions of possible moves, assigned a rating to each move using an algorithm, and chose the move with the highest rating.

And how exactly does Kasparov pick his moves, in your opinion? Do you have proof that your opinion as to how he selected his moves is correct?
Assistant moderator to airmax1227. PM me with any questions or concerns!
dylancatlow
Posts: 12,255
Add as Friend
Challenge to a Debate
Send a Message
11/25/2014 2:25:14 PM
Posted: 2 years ago
At 11/25/2014 2:18:03 PM, bladerunner060 wrote:
At 11/25/2014 1:55:50 PM, dylancatlow wrote:
Deep Blue simply generated millions upon millions of possible moves, assigned a rating to each move using an algorithm, and chose the move with the highest rating.

And how exactly does Kasparov pick his moves, in your opinion? Do you have proof that your opinion as to how he selected his moves is correct?

I don't claim to know exactly how Kasparov picks his moves. However, the human brain is simply not capable of generating millions of moves per second and tracking each move along all possible paths, so I can be sure that Kasparov was not making his moves like Deep Blue was, for there's simply no way he could compete against Deep Blue if that were the case.
bladerunner060
Posts: 7,126
Add as Friend
Challenge to a Debate
Send a Message
11/25/2014 2:28:54 PM
Posted: 2 years ago
At 11/25/2014 2:25:14 PM, dylancatlow wrote:
At 11/25/2014 2:18:03 PM, bladerunner060 wrote:
At 11/25/2014 1:55:50 PM, dylancatlow wrote:
Deep Blue simply generated millions upon millions of possible moves, assigned a rating to each move using an algorithm, and chose the move with the highest rating.

And how exactly does Kasparov pick his moves, in your opinion? Do you have proof that your opinion as to how he selected his moves is correct?

I don't claim to know exactly how Kasparov picks his moves. However, the human brain is simply not capable of generating millions of moves per second and tracking each move along all possible paths, so I can be sure that Kasparov was not making his moves like Deep Blue was, for there's simply no way he could compete against Deep Blue if that were the case.

So, to be clear, you don't know how he does it, but you know he does it "more intelligently" than Deep Blue?
Assistant moderator to airmax1227. PM me with any questions or concerns!
dylancatlow
Posts: 12,255
Add as Friend
Challenge to a Debate
Send a Message
11/25/2014 2:36:42 PM
Posted: 2 years ago
At 11/25/2014 2:28:54 PM, bladerunner060 wrote:
At 11/25/2014 2:25:14 PM, dylancatlow wrote:
At 11/25/2014 2:18:03 PM, bladerunner060 wrote:
At 11/25/2014 1:55:50 PM, dylancatlow wrote:
Deep Blue simply generated millions upon millions of possible moves, assigned a rating to each move using an algorithm, and chose the move with the highest rating.

And how exactly does Kasparov pick his moves, in your opinion? Do you have proof that your opinion as to how he selected his moves is correct?

I don't claim to know exactly how Kasparov picks his moves. However, the human brain is simply not capable of generating millions of moves per second and tracking each move along all possible paths, so I can be sure that Kasparov was not making his moves like Deep Blue was, for there's simply no way he could compete against Deep Blue if that were the case.

So, to be clear, you don't know how he does it, but you know he does it "more intelligently" than Deep Blue?

I don't know how he does it, but I know how he doesn't do it (brute force). Given the sheer computational advantage of Deep Blue - the fact that it could simulate trillions of moves within the given timeframe, while Kasparov could simulate only a fraction of that - implies that Kasparov was a smarter player, given that he won 2.5 games out of six. What other explanation is there?
bladerunner060
Posts: 7,126
Add as Friend
Challenge to a Debate
Send a Message
11/25/2014 2:38:49 PM
Posted: 2 years ago
At 11/25/2014 2:36:42 PM, dylancatlow wrote:
At 11/25/2014 2:28:54 PM, bladerunner060 wrote:
At 11/25/2014 2:25:14 PM, dylancatlow wrote:
At 11/25/2014 2:18:03 PM, bladerunner060 wrote:
At 11/25/2014 1:55:50 PM, dylancatlow wrote:
Deep Blue simply generated millions upon millions of possible moves, assigned a rating to each move using an algorithm, and chose the move with the highest rating.

And how exactly does Kasparov pick his moves, in your opinion? Do you have proof that your opinion as to how he selected his moves is correct?

I don't claim to know exactly how Kasparov picks his moves. However, the human brain is simply not capable of generating millions of moves per second and tracking each move along all possible paths, so I can be sure that Kasparov was not making his moves like Deep Blue was, for there's simply no way he could compete against Deep Blue if that were the case.

So, to be clear, you don't know how he does it, but you know he does it "more intelligently" than Deep Blue?

I don't know how he does it, but I know how he doesn't do it (brute force). Given the sheer computational advantage of Deep Blue - the fact that it could simulate trillions of moves within the given timeframe, while Kasparov could simulate only a fraction of that - implies that Kasparov was a smarter player, given that he won 2.5 games out of six. What other explanation is there?

That the algorithm that DB used was better than the algorithm that GK used.
Assistant moderator to airmax1227. PM me with any questions or concerns!
dylancatlow
Posts: 12,255
Add as Friend
Challenge to a Debate
Send a Message
11/25/2014 2:43:10 PM
Posted: 2 years ago
At 11/25/2014 2:38:49 PM, bladerunner060 wrote:
At 11/25/2014 2:36:42 PM, dylancatlow wrote:
At 11/25/2014 2:28:54 PM, bladerunner060 wrote:
At 11/25/2014 2:25:14 PM, dylancatlow wrote:
At 11/25/2014 2:18:03 PM, bladerunner060 wrote:
At 11/25/2014 1:55:50 PM, dylancatlow wrote:
Deep Blue simply generated millions upon millions of possible moves, assigned a rating to each move using an algorithm, and chose the move with the highest rating.

And how exactly does Kasparov pick his moves, in your opinion? Do you have proof that your opinion as to how he selected his moves is correct?

I don't claim to know exactly how Kasparov picks his moves. However, the human brain is simply not capable of generating millions of moves per second and tracking each move along all possible paths, so I can be sure that Kasparov was not making his moves like Deep Blue was, for there's simply no way he could compete against Deep Blue if that were the case.

So, to be clear, you don't know how he does it, but you know he does it "more intelligently" than Deep Blue?

I don't know how he does it, but I know how he doesn't do it (brute force). Given the sheer computational advantage of Deep Blue - the fact that it could simulate trillions of moves within the given timeframe, while Kasparov could simulate only a fraction of that - implies that Kasparov was a smarter player, given that he won 2.5 games out of six. What other explanation is there?

That the algorithm that DB used was better than the algorithm that GK used.

Except GK wasn't using an "algorithm", because he's not a computer. If Deep Blue's strategy were truly more intelligent than Kasparov's, then it should have utterly dominated him, given that it was applied to trillions of times more data.
fazz
Posts: 1,617
Add as Friend
Challenge to a Debate
Send a Message
11/25/2014 2:46:39 PM
Posted: 2 years ago
Well no. Deep Blue (computer program) essentially did not agree to the rules that were set before the match. They kept changing the program while the game was on-going which sort of makes it cheating. Earlier, Kasparov had beaten the computer. But when asked about "why he had lost" he said, the computer was "doing thing it could not have been computationally capable of doing" like artificial intelligence! After the game, the developers admitted they had changed the programs mid-game which is cheating.
dylancatlow
Posts: 12,255
Add as Friend
Challenge to a Debate
Send a Message
11/25/2014 2:51:12 PM
Posted: 2 years ago
At 11/25/2014 2:46:39 PM, fazz wrote:
Well no. Deep Blue (computer program) essentially did not agree to the rules that were set before the match. They kept changing the program while the game was on-going which sort of makes it cheating. Earlier, Kasparov had beaten the computer. But when asked about "why he had lost" he said, the computer was "doing thing it could not have been computationally capable of doing" like artificial intelligence! After the game, the developers admitted they had changed the programs mid-game which is cheating.

That's just a conspiracy theory peddled by Kasparov lol. To my knowledge, there is no proof of this.
bladerunner060
Posts: 7,126
Add as Friend
Challenge to a Debate
Send a Message
11/25/2014 2:51:23 PM
Posted: 2 years ago
At 11/25/2014 2:43:10 PM, dylancatlow wrote:
At 11/25/2014 2:38:49 PM, bladerunner060 wrote:
At 11/25/2014 2:36:42 PM, dylancatlow wrote:
At 11/25/2014 2:28:54 PM, bladerunner060 wrote:
At 11/25/2014 2:25:14 PM, dylancatlow wrote:
At 11/25/2014 2:18:03 PM, bladerunner060 wrote:
At 11/25/2014 1:55:50 PM, dylancatlow wrote:
Deep Blue simply generated millions upon millions of possible moves, assigned a rating to each move using an algorithm, and chose the move with the highest rating.

And how exactly does Kasparov pick his moves, in your opinion? Do you have proof that your opinion as to how he selected his moves is correct?

I don't claim to know exactly how Kasparov picks his moves. However, the human brain is simply not capable of generating millions of moves per second and tracking each move along all possible paths, so I can be sure that Kasparov was not making his moves like Deep Blue was, for there's simply no way he could compete against Deep Blue if that were the case.

So, to be clear, you don't know how he does it, but you know he does it "more intelligently" than Deep Blue?

I don't know how he does it, but I know how he doesn't do it (brute force). Given the sheer computational advantage of Deep Blue - the fact that it could simulate trillions of moves within the given timeframe, while Kasparov could simulate only a fraction of that - implies that Kasparov was a smarter player, given that he won 2.5 games out of six. What other explanation is there?

That the algorithm that DB used was better than the algorithm that GK used.

Except GK wasn't using an "algorithm", because he's not a computer.

It's not impossible for a non-computer to use an algorithm. An algorithm is a step-by-step process of solving a calculation. Moves are calculated. He most definitely DID have an algorithm--unless you're claiming that he "felt" his way to the correct answer, in which case I find it very hard to attribute that to "intelligence". His strategies had been named before he came along.

If Deep Blue's strategy were truly more intelligent than Kasparov's, then it should have utterly dominated him, given that it was applied to trillions of times more data.

It was better sufficient to win. It's worth noting that Kasparov's own frustration led him to resign a game that could have been a draw, too.

The notion that Deep Blue could only have had a better strategy if its algorithm was perfect is absurd.
Assistant moderator to airmax1227. PM me with any questions or concerns!
bladerunner060
Posts: 7,126
Add as Friend
Challenge to a Debate
Send a Message
11/25/2014 2:52:27 PM
Posted: 2 years ago
At 11/25/2014 2:46:39 PM, fazz wrote:
Well no. Deep Blue (computer program) essentially did not agree to the rules that were set before the match. They kept changing the program while the game was on-going which sort of makes it cheating. Earlier, Kasparov had beaten the computer. But when asked about "why he had lost" he said, the computer was "doing thing it could not have been computationally capable of doing" like artificial intelligence! After the game, the developers admitted they had changed the programs mid-game which is cheating.

No. The rules allowed for changes to the program between games--it's no different than Kasparov learning from his mistakes. What Kasparov accused them of was human intervention DURING the game. A claim for which he has no evidence at all.
Assistant moderator to airmax1227. PM me with any questions or concerns!
dylancatlow
Posts: 12,255
Add as Friend
Challenge to a Debate
Send a Message
11/25/2014 3:08:22 PM
Posted: 2 years ago
At 11/25/2014 2:51:23 PM, bladerunner060 wrote:
At 11/25/2014 2:43:10 PM, dylancatlow wrote:
At 11/25/2014 2:38:49 PM, bladerunner060 wrote:
At 11/25/2014 2:36:42 PM, dylancatlow wrote:
At 11/25/2014 2:28:54 PM, bladerunner060 wrote:
At 11/25/2014 2:25:14 PM, dylancatlow wrote:
At 11/25/2014 2:18:03 PM, bladerunner060 wrote:
At 11/25/2014 1:55:50 PM, dylancatlow wrote:
Deep Blue simply generated millions upon millions of possible moves, assigned a rating to each move using an algorithm, and chose the move with the highest rating.

And how exactly does Kasparov pick his moves, in your opinion? Do you have proof that your opinion as to how he selected his moves is correct?

I don't claim to know exactly how Kasparov picks his moves. However, the human brain is simply not capable of generating millions of moves per second and tracking each move along all possible paths, so I can be sure that Kasparov was not making his moves like Deep Blue was, for there's simply no way he could compete against Deep Blue if that were the case.

So, to be clear, you don't know how he does it, but you know he does it "more intelligently" than Deep Blue?

I don't know how he does it, but I know how he doesn't do it (brute force). Given the sheer computational advantage of Deep Blue - the fact that it could simulate trillions of moves within the given timeframe, while Kasparov could simulate only a fraction of that - implies that Kasparov was a smarter player, given that he won 2.5 games out of six. What other explanation is there?

That the algorithm that DB used was better than the algorithm that GK used.

Except GK wasn't using an "algorithm", because he's not a computer.

It's not impossible for a non-computer to use an algorithm. An algorithm is a step-by-step process of solving a calculation. Moves are calculated. He most definitely DID have an algorithm--unless you're claiming that he "felt" his way to the correct answer, in which case I find it very hard to attribute that to "intelligence". His strategies had been named before he came along.

In order to claim that Kasparov was using an "algorithm", you'd have to accept the premise that all of his moves were made according to an invariant set of rules, i.e., made exactly as he always would have made them, which is of course nonsense. The time he spent on his moves was him figuring out what to do in the first place, not how to apply an algorithm.

If Deep Blue's strategy were truly more intelligent than Kasparov's, then it should have utterly dominated him, given that it was applied to trillions of times more data.

It was better sufficient to win. It's worth noting that Kasparov's own frustration led him to resign a game that could have been a draw, too.

Are you trying to imply that this makes him less intelligent than Deep Blue?


The notion that Deep Blue could only have had a better strategy if its algorithm was perfect is absurd.

What on earth are you talking about? I never suggested anything close to this.
fazz
Posts: 1,617
Add as Friend
Challenge to a Debate
Send a Message
11/25/2014 3:09:31 PM
Posted: 2 years ago
At 11/25/2014 2:52:27 PM, bladerunner060 wrote:
At 11/25/2014 2:46:39 PM, fazz wrote:
Well no. Deep Blue (computer program) essentially did not agree to the rules that were set before the match. They kept changing the program while the game was on-going which sort of makes it cheating. Earlier, Kasparov had beaten the computer. But when asked about "why he had lost" he said, the computer was "doing thing it could not have been computationally capable of doing" like artificial intelligence! After the game, the developers admitted they had changed the programs mid-game which is cheating.

Dylan wrote:
That's just a conspiracy theory peddled by Kasparov lol. To my knowledge, there is no proof of this.

Bladerunner wrote:
No. The rules allowed for changes to the program between games--it's no different than Kasparov learning from his mistakes. What Kasparov accused them of was human intervention DURING the game. A claim for which he has no evidence at all.

http://www.research.ibm.com.... 8.html [ibm.com] "13. At any time during play, IBM may replace any or all of the computer hardware and/or software being used to play the games" That sounds pretty shady to me.

^the above does not indicate "between games" as Blade claims. It basically says we can re-program it to do a move that is "not in the book" as it were. Shady.

Again, Kasparov thought IBM the move deep blue made was so uncharacteristic for a computer and it was according an anamoly for deep blue's play so far. The problem is IBM refused to publish the move-list for the game. Now, that does not make any sense.

If you won you won. But why would you seal it up like the CIA or NSA, lol. Shady.
bladerunner060
Posts: 7,126
Add as Friend
Challenge to a Debate
Send a Message
11/25/2014 3:12:29 PM
Posted: 2 years ago
At 11/25/2014 3:08:22 PM, dylancatlow wrote:
At 11/25/2014 2:51:23 PM, bladerunner060 wrote:
At 11/25/2014 2:43:10 PM, dylancatlow wrote:
At 11/25/2014 2:38:49 PM, bladerunner060 wrote:
At 11/25/2014 2:36:42 PM, dylancatlow wrote:
At 11/25/2014 2:28:54 PM, bladerunner060 wrote:
At 11/25/2014 2:25:14 PM, dylancatlow wrote:
At 11/25/2014 2:18:03 PM, bladerunner060 wrote:
At 11/25/2014 1:55:50 PM, dylancatlow wrote:
Deep Blue simply generated millions upon millions of possible moves, assigned a rating to each move using an algorithm, and chose the move with the highest rating.

And how exactly does Kasparov pick his moves, in your opinion? Do you have proof that your opinion as to how he selected his moves is correct?

I don't claim to know exactly how Kasparov picks his moves. However, the human brain is simply not capable of generating millions of moves per second and tracking each move along all possible paths, so I can be sure that Kasparov was not making his moves like Deep Blue was, for there's simply no way he could compete against Deep Blue if that were the case.

So, to be clear, you don't know how he does it, but you know he does it "more intelligently" than Deep Blue?

I don't know how he does it, but I know how he doesn't do it (brute force). Given the sheer computational advantage of Deep Blue - the fact that it could simulate trillions of moves within the given timeframe, while Kasparov could simulate only a fraction of that - implies that Kasparov was a smarter player, given that he won 2.5 games out of six. What other explanation is there?

That the algorithm that DB used was better than the algorithm that GK used.

Except GK wasn't using an "algorithm", because he's not a computer.

It's not impossible for a non-computer to use an algorithm. An algorithm is a step-by-step process of solving a calculation. Moves are calculated. He most definitely DID have an algorithm--unless you're claiming that he "felt" his way to the correct answer, in which case I find it very hard to attribute that to "intelligence". His strategies had been named before he came along.

In order to claim that Kasparov was using an "algorithm", you'd have to accept the premise that all of his moves were made according to an invariant set of rules, i.e., made exactly as he always would have made them, which is of course nonsense. The time he spent on his moves was him figuring out what to do in the first place, not how to apply an algorithm.

It's nonsense because he would have to factor in previous play, and what he thinks his opponent will do. Chess computers don't always make the exact same moves, either--depending on their algorithm. The rules (generally speaking) don't vary, but the notion that the game would be identical is absurd.

If Deep Blue's strategy were truly more intelligent than Kasparov's, then it should have utterly dominated him, given that it was applied to trillions of times more data.

It was better sufficient to win. It's worth noting that Kasparov's own frustration led him to resign a game that could have been a draw, too.

Are you trying to imply that this makes him less intelligent than Deep Blue?

You've been explicitly saying that he's MORE intelligent. I think the fact that he conceded a game he could have won belies that notion. I'm not implying, necessarily, that Deep Blue was "more" intelligent, but rather saying I don't think YOUR position is supported.

The notion that Deep Blue could only have had a better strategy if its algorithm was perfect is absurd.

What on earth are you talking about? I never suggested anything close to this.

Well, that's false, you outright SAID It:

"If Deep Blue's strategy were truly more intelligent than Kasparov's, then it should have utterly dominated him, given that it was applied to trillions of times more data."

Perhaps "Perfect" is a bit of hyperbole, but feel free to substitute your actual language.
Assistant moderator to airmax1227. PM me with any questions or concerns!
bladerunner060
Posts: 7,126
Add as Friend
Challenge to a Debate
Send a Message
11/25/2014 3:19:35 PM
Posted: 2 years ago
At 11/25/2014 3:09:31 PM, fazz wrote:
At 11/25/2014 2:52:27 PM, bladerunner060 wrote:
At 11/25/2014 2:46:39 PM, fazz wrote:
Well no. Deep Blue (computer program) essentially did not agree to the rules that were set before the match. They kept changing the program while the game was on-going which sort of makes it cheating. Earlier, Kasparov had beaten the computer. But when asked about "why he had lost" he said, the computer was "doing thing it could not have been computationally capable of doing" like artificial intelligence! After the game, the developers admitted they had changed the programs mid-game which is cheating.

Dylan wrote:
That's just a conspiracy theory peddled by Kasparov lol. To my knowledge, there is no proof of this.

Bladerunner wrote:
No. The rules allowed for changes to the program between games--it's no different than Kasparov learning from his mistakes. What Kasparov accused them of was human intervention DURING the game. A claim for which he has no evidence at all.

http://www.research.ibm.com.... 8.html [ibm.com] "13. At any time during play, IBM may replace any or all of the computer hardware and/or software being used to play the games" That sounds pretty shady to me.

^the above does not indicate "between games" as Blade claims. It basically says we can re-program it to do a move that is "not in the book" as it were. Shady.

Again, Kasparov thought IBM the move deep blue made was so uncharacteristic for a computer and it was according an anamoly for deep blue's play so far. The problem is IBM refused to publish the move-list for the game. Now, that does not make any sense.

If you won you won. But why would you seal it up like the CIA or NSA, lol. Shady.

It wasn't the "move-list" it was the logs. The move list for the game was apparent to all--because that's how chess works. The logs showed the computers "thought process". They DID release the logs eventually. You can fault them for not doing so immediately if you want, but nonetheless, Kasparov has no evidence of cheating.

The rules were agreed in advance.

The "cheat" Kasparov claimed was a human player--not a change to the program.

http://www.chess.com...

"After the loss Kasparov said that he sometimes saw deep intelligence and creativity in the machine's moves, suggesting that during the second game, human chess players, in contravention of the rules, intervened. IBM denied that it cheated, saying the only human intervention occurred between games. The rules provided for the developers to modify the program between games, an opportunity they said they used to shore up weaknesses in the computer's play revealed during the course of the match."
Assistant moderator to airmax1227. PM me with any questions or concerns!
fazz
Posts: 1,617
Add as Friend
Challenge to a Debate
Send a Message
11/25/2014 3:29:55 PM
Posted: 2 years ago
At 11/25/2014 3:19:35 PM, bladerunner060 wrote:
At 11/25/2014 3:09:31 PM, fazz wrote:
At 11/25/2014 2:52:27 PM, bladerunner060 wrote:
At 11/25/2014 2:46:39 PM, fazz wrote:
Well no. Deep Blue (computer program) essentially did not agree to the rules that were set before the match. They kept changing the program while the game was on-going which sort of makes it cheating. Earlier, Kasparov had beaten the computer. But when asked about "why he had lost" he said, the computer was "doing thing it could not have been computationally capable of doing" like artificial intelligence! After the game, the developers admitted they had changed the programs mid-game which is cheating.

Dylan wrote:
That's just a conspiracy theory peddled by Kasparov lol. To my knowledge, there is no proof of this.

Bladerunner wrote:
No. The rules allowed for changes to the program between games--it's no different than Kasparov learning from his mistakes. What Kasparov accused them of was human intervention DURING the game. A claim for which he has no evidence at all.

http://www.research.ibm.com.... 8.html [ibm.com] "13. At any time during play, IBM may replace any or all of the computer hardware and/or software being used to play the games" That sounds pretty shady to me.

^the above does not indicate "between games" as Blade claims. It basically says we can re-program it to do a move that is "not in the book" as it were. Shady.

Again, Kasparov thought IBM the move deep blue made was so uncharacteristic for a computer and it was according an anamoly for deep blue's play so far. The problem is IBM refused to publish the move-list for the game. Now, that does not make any sense.

If you won you won. But why would you seal it up like the CIA or NSA, lol. Shady.

It wasn't the "move-list" it was the logs. The move list for the game was apparent to all--because that's how chess works. The logs showed the computers "thought process". They DID release the logs eventually. You can fault them for not doing so immediately if you want, but nonetheless, Kasparov has no evidence of cheating.

The rules were agreed in advance.

The "cheat" Kasparov claimed was a human player--not a change to the program.

http://www.chess.com...

"After the loss Kasparov said that he sometimes saw deep intelligence and creativity in the machine's moves, suggesting that during the second game, human chess players, in contravention of the rules, intervened. IBM denied that it cheated, saying the only human intervention occurred between games. The rules provided for the developers to modify the program between games, an opportunity they said they used to shore up weaknesses in the computer's play revealed during the course of the match."

The article also said: The video "Kasparov vs. The Machine" explains the theory that IBM"s chess team illegally aided Deep Blue in defeating Kasparov to impress the computer world. The video states that IBM"s stock rose 20% after Deep Blue"s victory! Kasparov requested printouts of the machine's log files but IBM refused, although the company later published the logs on the Internet.[58] Kasparov demanded a rematch, but IBM declined and retired Deep Blue, which has been viewed by Kasparov as covering up evidence of tampering during the game." And isn"t it suspicious that, once accused of cheating, IBM did not initially allow the computer"s log files to be released? Even though IBM eventually released the log files, those could have been fabricated. In addition to that, IBM retired Deep Blue after achieving one of the biggest computer accomplishments ever. IBM shut down the program and never opened it again. Shouldn"t this success be modeled, improved, and researched further?

In any Question of Science, or Technology, transparency is the name of the game. Here, IBM threw out any claim to a scientific breakthrough, by refusing to reveal their own datasets, thereby revealing that their so-called "victory" was merely a clever ploy for corporate gain.
dylancatlow
Posts: 12,255
Add as Friend
Challenge to a Debate
Send a Message
11/25/2014 3:34:10 PM
Posted: 2 years ago
At 11/25/2014 3:12:29 PM, bladerunner060 wrote:
At 11/25/2014 3:08:22 PM, dylancatlow wrote:
At 11/25/2014 2:51:23 PM, bladerunner060 wrote:
At 11/25/2014 2:43:10 PM, dylancatlow wrote:
At 11/25/2014 2:38:49 PM, bladerunner060 wrote:
At 11/25/2014 2:36:42 PM, dylancatlow wrote:
At 11/25/2014 2:28:54 PM, bladerunner060 wrote:
At 11/25/2014 2:25:14 PM, dylancatlow wrote:
At 11/25/2014 2:18:03 PM, bladerunner060 wrote:
At 11/25/2014 1:55:50 PM, dylancatlow wrote:
Deep Blue simply generated millions upon millions of possible moves, assigned a rating to each move using an algorithm, and chose the move with the highest rating.

And how exactly does Kasparov pick his moves, in your opinion? Do you have proof that your opinion as to how he selected his moves is correct?

I don't claim to know exactly how Kasparov picks his moves. However, the human brain is simply not capable of generating millions of moves per second and tracking each move along all possible paths, so I can be sure that Kasparov was not making his moves like Deep Blue was, for there's simply no way he could compete against Deep Blue if that were the case.

So, to be clear, you don't know how he does it, but you know he does it "more intelligently" than Deep Blue?

I don't know how he does it, but I know how he doesn't do it (brute force). Given the sheer computational advantage of Deep Blue - the fact that it could simulate trillions of moves within the given timeframe, while Kasparov could simulate only a fraction of that - implies that Kasparov was a smarter player, given that he won 2.5 games out of six. What other explanation is there?

That the algorithm that DB used was better than the algorithm that GK used.

Except GK wasn't using an "algorithm", because he's not a computer.

It's not impossible for a non-computer to use an algorithm. An algorithm is a step-by-step process of solving a calculation. Moves are calculated. He most definitely DID have an algorithm--unless you're claiming that he "felt" his way to the correct answer, in which case I find it very hard to attribute that to "intelligence". His strategies had been named before he came along.

In order to claim that Kasparov was using an "algorithm", you'd have to accept the premise that all of his moves were made according to an invariant set of rules, i.e., made exactly as he always would have made them, which is of course nonsense. The time he spent on his moves was him figuring out what to do in the first place, not how to apply an algorithm.

It's nonsense because he would have to factor in previous play, and what he thinks his opponent will do. Chess computers don't always make the exact same moves, either--depending on their algorithm. The rules (generally speaking) don't vary, but the notion that the game would be identical is absurd.

Just because Kasparov employs general parameters for deciding his moves does not mean he was using an algorithm. An algorithm is an invariant set of rules according to which all moves are made, even nondeterministic ones. Obviously, he was not using an algorithm. He was determining his moves as we went, not calling upon a formula in every situation.


If Deep Blue's strategy were truly more intelligent than Kasparov's, then it should have utterly dominated him, given that it was applied to trillions of times more data.

It was better sufficient to win. It's worth noting that Kasparov's own frustration led him to resign a game that could have been a draw, too.

Are you trying to imply that this makes him less intelligent than Deep Blue?

You've been explicitly saying that he's MORE intelligent. I think the fact that he conceded a game he could have won belies that notion. I'm not implying, necessarily, that Deep Blue was "more" intelligent, but rather saying I don't think YOUR position is supported.

...Except my reasoning does not presume that humans can't make mistakes, even very big ones.


The notion that Deep Blue could only have had a better strategy if its algorithm was perfect is absurd.

What on earth are you talking about? I never suggested anything close to this.

Well, that's false, you outright SAID It:

"If Deep Blue's strategy were truly more intelligent than Kasparov's, then it should have utterly dominated him, given that it was applied to trillions of times more data."

My point was that if DP had a better strategy than Kasparov, then it shouldn't have lost any games, given its computational advantage. Your reading comprehension skills really need some work.
fazz
Posts: 1,617
Add as Friend
Challenge to a Debate
Send a Message
11/25/2014 3:40:19 PM
Posted: 2 years ago
I would like to see the evidence. I still fail to find any place or website on the internet where IBM has published the logs. Since, publishing the 'real' logs could make their stocks plummet 20% (as mentioned) it would be safe to assume publishing at a later date means they doctored the evidence. No offense to medical practicers.
fazz
Posts: 1,617
Add as Friend
Challenge to a Debate
Send a Message
11/25/2014 4:02:40 PM
Posted: 2 years ago
IBM said it published the logs on this website. The link however is blank. No logs. No mention of claims. Nothing: http://www-03.ibm.com...
bladerunner060
Posts: 7,126
Add as Friend
Challenge to a Debate
Send a Message
11/25/2014 4:20:11 PM
Posted: 2 years ago
At 11/25/2014 3:34:10 PM, dylancatlow wrote:
At 11/25/2014 3:12:29 PM, bladerunner060 wrote:
At 11/25/2014 3:08:22 PM, dylancatlow wrote:
At 11/25/2014 2:51:23 PM, bladerunner060 wrote:
At 11/25/2014 2:43:10 PM, dylancatlow wrote:
At 11/25/2014 2:38:49 PM, bladerunner060 wrote:
At 11/25/2014 2:36:42 PM, dylancatlow wrote:
At 11/25/2014 2:28:54 PM, bladerunner060 wrote:
At 11/25/2014 2:25:14 PM, dylancatlow wrote:
At 11/25/2014 2:18:03 PM, bladerunner060 wrote:
At 11/25/2014 1:55:50 PM, dylancatlow wrote:
Deep Blue simply generated millions upon millions of possible moves, assigned a rating to each move using an algorithm, and chose the move with the highest rating.

And how exactly does Kasparov pick his moves, in your opinion? Do you have proof that your opinion as to how he selected his moves is correct?

I don't claim to know exactly how Kasparov picks his moves. However, the human brain is simply not capable of generating millions of moves per second and tracking each move along all possible paths, so I can be sure that Kasparov was not making his moves like Deep Blue was, for there's simply no way he could compete against Deep Blue if that were the case.

So, to be clear, you don't know how he does it, but you know he does it "more intelligently" than Deep Blue?

I don't know how he does it, but I know how he doesn't do it (brute force). Given the sheer computational advantage of Deep Blue - the fact that it could simulate trillions of moves within the given timeframe, while Kasparov could simulate only a fraction of that - implies that Kasparov was a smarter player, given that he won 2.5 games out of six. What other explanation is there?

That the algorithm that DB used was better than the algorithm that GK used.

Except GK wasn't using an "algorithm", because he's not a computer.

It's not impossible for a non-computer to use an algorithm. An algorithm is a step-by-step process of solving a calculation. Moves are calculated. He most definitely DID have an algorithm--unless you're claiming that he "felt" his way to the correct answer, in which case I find it very hard to attribute that to "intelligence". His strategies had been named before he came along.

In order to claim that Kasparov was using an "algorithm", you'd have to accept the premise that all of his moves were made according to an invariant set of rules, i.e., made exactly as he always would have made them, which is of course nonsense. The time he spent on his moves was him figuring out what to do in the first place, not how to apply an algorithm.

It's nonsense because he would have to factor in previous play, and what he thinks his opponent will do. Chess computers don't always make the exact same moves, either--depending on their algorithm. The rules (generally speaking) don't vary, but the notion that the game would be identical is absurd.

Just because Kasparov employs general parameters for deciding his moves does not mean he was using an algorithm. An algorithm is an invariant set of rules according to which all moves are made, even nondeterministic ones. Obviously, he was not using an algorithm. He was determining his moves as we went, not calling upon a formula in every situation.

How did he determine these moves, if not via criteria? Are you saying his moves were arbitrary?

If Deep Blue's strategy were truly more intelligent than Kasparov's, then it should have utterly dominated him, given that it was applied to trillions of times more data.

It was better sufficient to win. It's worth noting that Kasparov's own frustration led him to resign a game that could have been a draw, too.

Are you trying to imply that this makes him less intelligent than Deep Blue?

You've been explicitly saying that he's MORE intelligent. I think the fact that he conceded a game he could have won belies that notion. I'm not implying, necessarily, that Deep Blue was "more" intelligent, but rather saying I don't think YOUR position is supported.

...Except my reasoning does not presume that humans can't make mistakes, even very big ones.

I never said it did. I said that your claim is unsupported. Becasue it is. You keep asserting Kasparov is more intelligent, but I've yet to see a reason for that assertion besides your assertion that it's so.

The notion that Deep Blue could only have had a better strategy if its algorithm was perfect is absurd.

What on earth are you talking about? I never suggested anything close to this.

Well, that's false, you outright SAID It:

"If Deep Blue's strategy were truly more intelligent than Kasparov's, then it should have utterly dominated him, given that it was applied to trillions of times more data."

My point was that if DP had a better strategy than Kasparov, then it shouldn't have lost any games, given its computational advantage. Your reading comprehension skills really need some work.

While your blustery arrogance is, as always, as charming as it is wholly unfounded: No, they don't. Because now you really are saying that it's algorithm had to be perfect--at least in relation to Kasparov. So you're saying that it has to win ALL games for its strategy to be better. That's not how chess is assessed, nor is it how strategy is assessed. You have given no cause to think that the greater processing powers necessitate that it should win every game--you've just asserted it. You really need to learn that merely asserting things does not make them so.
Assistant moderator to airmax1227. PM me with any questions or concerns!
dylancatlow
Posts: 12,255
Add as Friend
Challenge to a Debate
Send a Message
11/25/2014 6:12:47 PM
Posted: 2 years ago
At 11/25/2014 4:20:11 PM, bladerunner060 wrote:
At 11/25/2014 3:34:10 PM, dylancatlow wrote:
At 11/25/2014 3:12:29 PM, bladerunner060 wrote:
At 11/25/2014 3:08:22 PM, dylancatlow wrote:
At 11/25/2014 2:51:23 PM, bladerunner060 wrote:
At 11/25/2014 2:43:10 PM, dylancatlow wrote:
At 11/25/2014 2:38:49 PM, bladerunner060 wrote:
At 11/25/2014 2:36:42 PM, dylancatlow wrote:
At 11/25/2014 2:28:54 PM, bladerunner060 wrote:
At 11/25/2014 2:25:14 PM, dylancatlow wrote:
At 11/25/2014 2:18:03 PM, bladerunner060 wrote:
At 11/25/2014 1:55:50 PM, dylancatlow wrote:
Deep Blue simply generated millions upon millions of possible moves, assigned a rating to each move using an algorithm, and chose the move with the highest rating.

And how exactly does Kasparov pick his moves, in your opinion? Do you have proof that your opinion as to how he selected his moves is correct?

I don't claim to know exactly how Kasparov picks his moves. However, the human brain is simply not capable of generating millions of moves per second and tracking each move along all possible paths, so I can be sure that Kasparov was not making his moves like Deep Blue was, for there's simply no way he could compete against Deep Blue if that were the case.

So, to be clear, you don't know how he does it, but you know he does it "more intelligently" than Deep Blue?

I don't know how he does it, but I know how he doesn't do it (brute force). Given the sheer computational advantage of Deep Blue - the fact that it could simulate trillions of moves within the given timeframe, while Kasparov could simulate only a fraction of that - implies that Kasparov was a smarter player, given that he won 2.5 games out of six. What other explanation is there?

That the algorithm that DB used was better than the algorithm that GK used.

Except GK wasn't using an "algorithm", because he's not a computer.

It's not impossible for a non-computer to use an algorithm. An algorithm is a step-by-step process of solving a calculation. Moves are calculated. He most definitely DID have an algorithm--unless you're claiming that he "felt" his way to the correct answer, in which case I find it very hard to attribute that to "intelligence". His strategies had been named before he came along.

In order to claim that Kasparov was using an "algorithm", you'd have to accept the premise that all of his moves were made according to an invariant set of rules, i.e., made exactly as he always would have made them, which is of course nonsense. The time he spent on his moves was him figuring out what to do in the first place, not how to apply an algorithm.

It's nonsense because he would have to factor in previous play, and what he thinks his opponent will do. Chess computers don't always make the exact same moves, either--depending on their algorithm. The rules (generally speaking) don't vary, but the notion that the game would be identical is absurd.

Just because Kasparov employs general parameters for deciding his moves does not mean he was using an algorithm. An algorithm is an invariant set of rules according to which all moves are made, even nondeterministic ones. Obviously, he was not using an algorithm. He was determining his moves as we went, not calling upon a formula in every situation.

How did he determine these moves, if not via criteria? Are you saying his moves were arbitrary?


The choices aren't "arbitrary" and "algorithmic".

If Deep Blue's strategy were truly more intelligent than Kasparov's, then it should have utterly dominated him, given that it was applied to trillions of times more data.

It was better sufficient to win. It's worth noting that Kasparov's own frustration led him to resign a game that could have been a draw, too.

Are you trying to imply that this makes him less intelligent than Deep Blue?

You've been explicitly saying that he's MORE intelligent. I think the fact that he conceded a game he could have won belies that notion. I'm not implying, necessarily, that Deep Blue was "more" intelligent, but rather saying I don't think YOUR position is supported.

...Except my reasoning does not presume that humans can't make mistakes, even very big ones.

I never said it did. I said that your claim is unsupported. Becasue it is. You keep asserting Kasparov is more intelligent, but I've yet to see a reason for that assertion besides your assertion that it's so.

Then why would the fact that he made a mistake "belie" that notion, as you said? How is that even relevant to that notion?


The notion that Deep Blue could only have had a better strategy if its algorithm was perfect is absurd.

What on earth are you talking about? I never suggested anything close to this.

Well, that's false, you outright SAID It:

"If Deep Blue's strategy were truly more intelligent than Kasparov's, then it should have utterly dominated him, given that it was applied to trillions of times more data."

My point was that if DP had a better strategy than Kasparov, then it shouldn't have lost any games, given its computational advantage. Your reading comprehension skills really need some work.

While your blustery arrogance is, as always, as charming as it is wholly unfounded: No, they don't. Because now you really are saying that it's algorithm had to be perfect--at least in relation to Kasparov.

Please explain how "if Deep Blue had a better strategy than Kasparov, then it shouldn't have lost any games, given its computational advantage" implies that I was suggesting DB would need a perfect algorithm to beat Kasparov in all 6 games.

So you're saying that it has to win ALL games for its strategy to be better. That's not how chess is assessed, nor is it how strategy is assessed. You have given no cause to think that the greater processing powers necessitate that it should win every game--you've just asserted it. You really need to learn that merely asserting things does not make them so.

If Deep Blue were as intelligent as Kasparov at interpreting data, there's no reason it should lose any game, since it can simulate millions of times more moves than Kasparov. But it's not as intelligent, which is why it requires mass-simulation since it's only smart enough to "check" whether a move is good by exhaustively simulating it, not smart enough to arrive at the move deliberately and for intelligent reasons. If Deep Blue weren't allowed to mass-produce potential moves, it would definitely lose. It's not smart enough to arrive at good moves without going through massive calculations that an intelligent being would have the sense to avoid.
bladerunner060
Posts: 7,126
Add as Friend
Challenge to a Debate
Send a Message
11/25/2014 6:18:52 PM
Posted: 2 years ago
At 11/25/2014 6:12:47 PM, dylancatlow wrote:

How did he determine these moves, if not via criteria? Are you saying his moves were arbitrary?

The choices aren't "arbitrary" and "algorithmic".

How about you explain what the other choice in this context is.

I never said it did. I said that your claim is unsupported. Becasue it is. You keep asserting Kasparov is more intelligent, but I've yet to see a reason for that assertion besides your assertion that it's so.


Then why would the fact that he made a mistake "belie" that notion, as you said? How is that even relevant to that notion?

Are you saying that making mistakes is IRRELEVANT to intelligence? Bearing in mind that he had to be TOLD about the mistakes, according to Wikipedia his friends told him. He didn't see the solution.

You're arguing he's more intelligent. You've offered only assertion as to this.

In contrast: we have the overall loss, and at least one VERY big mistake (that would have turned the game from a loss to a draw and mad ethe tournament a draw overall).

Please explain how "if Deep Blue had a better strategy than Kasparov, then it shouldn't have lost any games, given its computational advantage" implies that I was suggesting DB would need a perfect algorithm to beat Kasparov in all 6 games.

6/6 is a perfect score. I'm not sure how you can not understand.

So you're saying that it has to win ALL games for its strategy to be better. That's not how chess is assessed, nor is it how strategy is assessed. You have given no cause to think that the greater processing powers necessitate that it should win every game--you've just asserted it. You really need to learn that merely asserting things does not make them so.

If Deep Blue were as intelligent as Kasparov at interpreting data, there's no reason it should lose any game, since it can simulate millions of times more moves than Kasparov. But it's not as intelligent, which is why it requires mass-simulation since it's only smart enough to "check" whether a move is good by exhaustively simulating it, not smart enough to arrive at the move deliberately and for intelligent reasons. If Deep Blue weren't allowed to mass-produce potential moves, it would definitely lose. It's not smart enough to arrive at good moves without going through massive calculations that an intelligent being would have the sense to avoid.

If Kasparov weren't allowe dto think about the potential outcomes of moves, he probably wouldn't be as good a player either, I would think.
Assistant moderator to airmax1227. PM me with any questions or concerns!
Envisage
Posts: 3,646
Add as Friend
Challenge to a Debate
Send a Message
11/25/2014 6:43:29 PM
Posted: 2 years ago
At 11/25/2014 1:55:50 PM, dylancatlow wrote:

Cool, you have just picked a subject I have massive interest in.
In 1997, the world chess champion Garry Kasparov was defeated in a chess tournament by Deep Blue, a supercomputer developed by IBM. Many people came away with the idea that Deep Blue had outsmarted Kasparov.

Did they" I would only argue that Deep Blue just performed better than Kasparov. What does it even mean to say one was "outsmarted", it"s rather meaningless as far as I see.

However, if you know anything about how Deep Blue made its moves, this seems rather absurd.

Min max algorithms, tablebases, opening books. So some extent, yes. I have serious doubts about the heuristics (i.e. the way a computer gives a quantitative estimate of how "good" each position is by the qualitative positions of it"s pieces) that went into evaluating each position however, which is much akin to how humans evaluate chess positions.

Rather than deciding its moves "intelligently", Deep Blue simply generated millions upon millions of possible moves, assigned a rating to each move using an algorithm, and chose the move with the highest rating.

Define "intelligently". Then we can continue. If "intelligently" is just to mean "human-like" then it just becomes a useless or question begging definition. Why should what deep blue did be any less "intelligent" than the way human"s do it.

Of course, assigning a useful rating to each move is no easy task. Each move in chess becomes a condition of subsequent moves, so determining a "good" move requires an astronomically high number of computations and an algorithm capable of recognizing good moves, as well as some mechanism for considering how moves might influence your opponent's moves, and in turn how that sets the stage for your moves, and so on.

Yaya, minmax tree search programs, pruning, etc.

So although Deep Blue deserves credit for its computational power, it does not deserve credit for "outsmarting" Kasparov (it merely deserves credit for outplaying him). For is intelligence not precisely that which allows you to avoid blunt strategies when solving a problem?
Not necessarily. You have yet to define "intelligent" so you are moving the goalposts at will. But I can tlel you Humans and computers simply just take 2 different approaches to chess:

Computer:
Deep Search-Space
Application of simple, generalised rules (heurestics)
Selection of the move which leads to the best objective evaluation (given it"s ruleset)

Human
Shallow search space
Application of experience in pattern-recognition, themes, long term plans
Selection of the move which leads to the best objective evaluation

The stronger computer programs (computer chess has come a LONG way since 1997) need nowhere near as deep a search space, owing to adopting a more experience-based playstyle, with better objective evaluations of each position, and less need to go as deep to achieve the results. More "human-like" even. However it is always the case that the better the hardware, the better the chess engine will run.

Shakespeare is considered a brilliant writer, not because he produced brilliant works per se, but because he produced brilliant works without needing to produce a trillion bad ones.

" Entirely subjective. Chess is an objective game.

The same goes for Deep Blue. It did not come up with good moves because it was intelligent, it came up with good moves because it came up with lots and lots of moves, some of which were good, and was able to recognize the good moves as "good" only because it had the computational power necessary to "force" the answer.

Still awaiting a definition of "intelligent" so I can"t argue against this, but only say it"s meaningless. But it doesn"t matter which way you spin it, computers always have a level of understanding of the game, otherwise it would not play! It is well-known that the greater a computer understands the game, the better the machine performs with the same computational hardware. An amusing thing is that the only way humans were able to compete with computers at the time was to play "anti-computer" chess. I.e. maximise the positions which are slow, benefit from long-term plans, with fewest tactics.

The fact that Deep Blue came out barely ahead even though it had the advantage of choosing its moves from a pool trillions of times larger than Kasparov's just goes to show how much more intelligent a player Kasparov really was.

Have you ever heard of the law of diminishing returns. Because you are grossly exaggerating the consequences here.
Kasparov had to make his moves on the basis of far less data, and was thus only able to compete because he was able to make better use of that data i.e., because was able to more intelligently interpret the data.

Note necessarily. Anybody who understands anti-computer chess tactics will know that much of it is to basically "find a position that the computer"s heuristic formulas sucks in". A chess program operating on a mobile phone searching only 500 positions per second would play on a grandmaster level today. A far cry from the 200 million positions per second that deep blue crunched.

Deep Blue made up for its profound stupidity by repeatedly throwing itself against a wall until finally it broke through, while Kasparov simply climbed over. That is the difference.
Wtf.
Envisage
Posts: 3,646
Add as Friend
Challenge to a Debate
Send a Message
11/25/2014 6:51:12 PM
Posted: 2 years ago
At 11/25/2014 3:40:19 PM, fazz wrote:
I would like to see the evidence. I still fail to find any place or website on the internet where IBM has published the logs. Since, publishing the 'real' logs could make their stocks plummet 20% (as mentioned) it would be safe to assume publishing at a later date means they doctored the evidence. No offense to medical practicers.

Deep Blue was only meant to be short term project. It barely competed in computer chess tournaments, and was built explicitly to beat Kaparov. So it really isn't a surprise they didn't care for the logs.

To be honest, it seems very unlikely that there was intervention on Deep Blue, since today's programs find the 'Key Move' (which kasparov alluded to) rather quickly (in only a few seconds 'thought'), so it is not exclusivly a 'human move' as Kasparov thought.

True or not, it most certainly did impact the rest of the exhibition for Kasparov, since he was clearly frustrated, played sub-standard and in the last game made a rather trivial mistake (for a Super-GM, what he was) and lost.

On the flip side, Kasparov would probably lose 6-0, or at best 5.5-0.5 to a comparable computer today.
HououinKyouma
Posts: 1,030
Add as Friend
Challenge to a Debate
Send a Message
11/25/2014 6:54:21 PM
Posted: 2 years ago
At 11/25/2014 1:55:50 PM, dylancatlow wrote:
In 1997, the world chess champion Garry Kasparov was defeated in a chess tournament by Deep Blue, a supercomputer developed by IBM. Many people came away with the idea that Deep Blue had outsmarted Kasparov. However, if you know anything about how Deep Blue made its moves, this seems rather absurd. Rather than deciding its moves "intelligently", Deep Blue simply generated millions upon millions of possible moves, assigned a rating to each move using an algorithm, and chose the move with the highest rating. Of course, assigning a useful rating to each move is no easy task. Each move in chess becomes a condition of subsequent moves, so determining a "good" move requires an astronomically high number of computations and an algorithm capable of recognizing good moves, as well as some mechanism for considering how moves might influence your opponent's moves, and in turn how that sets the stage for your moves, and so on. So although Deep Blue deserves credit for its computational power, it does not deserve credit for "outsmarting" Kasparov (it merely deserves credit for outplaying him). For is intelligence not precisely that which allows you to avoid blunt strategies when solving a problem? Shakespeare is considered a brilliant writer, not because he produced brilliant works per se, but because he produced brilliant works without needing to produce a trillion bad ones. The same goes for Deep Blue. It did not come up with good moves because it was intelligent, it came up with good moves because it came up with lots and lots of moves, some of which were good, and was able to recognize the good moves as "good" only because it had the computational power necessary to "force" the answer. The fact that Deep Blue came out barely ahead even though it had the advantage of choosing its moves from a pool trillions of times larger than Kasparov's just goes to show how much more intelligent a player Kasparov really was. Kasparov had to make his moves on the basis of far less data, and was thus only able to compete because he was able to make better use of that data i.e., because was able to more intelligently interpret the data. Deep Blue made up for its profound stupidity by repeatedly throwing itself against a wall until finally it broke through, while Kasparov simply climbed over. That is the difference.

There are some problems with your OP, the most important being that you're not comparing like with like. You're judging the "intelligence" of Deep Blue based not only on its moves but on its "thinking" process--to which we have access; whereas you judge Mr. K's "intelligence" based only on his actions--the moves in the game--without knowing, only intimating, what his thought process in the game actually was.

Therefore, we cannot make a judgment about whether or not the computer was more intelligent than the player, since we know very little about the workings of the player's mind. Unless of course we assume that Mr. K also thinks in terms of algorithms and just lacks the speed and data available to the computer, in which case the computer would be more "intelligent".
"Here the ways of men part: if you wish to strive for peace of soul and pleasure, then believe; if you wish to be a devotee of truth, then inquire." F. Nietzsche.

"Freedom is always freedom for the one who thinks differently." R. Luxemburg.

"The principle of the masochistic left is that, in general, two blacks make a white, half a loaf is the same as no bread." G. Orwell, paraphrase.

"Islamophobia is a word created by fascists, used by cowards, to manipulate morons". Andrew Cummins.
dylancatlow
Posts: 12,255
Add as Friend
Challenge to a Debate
Send a Message
11/25/2014 7:13:50 PM
Posted: 2 years ago
At 11/25/2014 6:54:21 PM, HououinKyouma wrote:
At 11/25/2014 1:55:50 PM, dylancatlow wrote:
In 1997, the world chess champion Garry Kasparov was defeated in a chess tournament by Deep Blue, a supercomputer developed by IBM. Many people came away with the idea that Deep Blue had outsmarted Kasparov. However, if you know anything about how Deep Blue made its moves, this seems rather absurd. Rather than deciding its moves "intelligently", Deep Blue simply generated millions upon millions of possible moves, assigned a rating to each move using an algorithm, and chose the move with the highest rating. Of course, assigning a useful rating to each move is no easy task. Each move in chess becomes a condition of subsequent moves, so determining a "good" move requires an astronomically high number of computations and an algorithm capable of recognizing good moves, as well as some mechanism for considering how moves might influence your opponent's moves, and in turn how that sets the stage for your moves, and so on. So although Deep Blue deserves credit for its computational power, it does not deserve credit for "outsmarting" Kasparov (it merely deserves credit for outplaying him). For is intelligence not precisely that which allows you to avoid blunt strategies when solving a problem? Shakespeare is considered a brilliant writer, not because he produced brilliant works per se, but because he produced brilliant works without needing to produce a trillion bad ones. The same goes for Deep Blue. It did not come up with good moves because it was intelligent, it came up with good moves because it came up with lots and lots of moves, some of which were good, and was able to recognize the good moves as "good" only because it had the computational power necessary to "force" the answer. The fact that Deep Blue came out barely ahead even though it had the advantage of choosing its moves from a pool trillions of times larger than Kasparov's just goes to show how much more intelligent a player Kasparov really was. Kasparov had to make his moves on the basis of far less data, and was thus only able to compete because he was able to make better use of that data i.e., because was able to more intelligently interpret the data. Deep Blue made up for its profound stupidity by repeatedly throwing itself against a wall until finally it broke through, while Kasparov simply climbed over. That is the difference.

There are some problems with your OP, the most important being that you're not comparing like with like. You're judging the "intelligence" of Deep Blue based not only on its moves but on its "thinking" process--to which we have access; whereas you judge Mr. K's "intelligence" based only on his actions--the moves in the game--without knowing, only intimating, what his thought process in the game actually was.

I'm saying that DB's moves do not reflect what is normally meant by intelligence, but rather reflects the sheer number of computations it can perform in a second. It's no different than any computational device, except that it's a lot faster. I say that Kasparov is more intelligent, not because I claim to know what his thought process is, but because he was able to overcome the computational gap. His thought process is irrelevant.

Therefore, we cannot make a judgment about whether or not the computer was more intelligent than the player, since we know very little about the workings of the player's mind. Unless of course we assume that Mr. K also thinks in terms of algorithms and just lacks the speed and data available to the computer, in which case the computer would be more "intelligent".

If Kasparov can keep up with DB, then clearly he's coming up with his moves in a more intelligent way, otherwise he would simply be outcomputed.
HououinKyouma
Posts: 1,030
Add as Friend
Challenge to a Debate
Send a Message
11/25/2014 7:19:13 PM
Posted: 2 years ago
At 11/25/2014 7:13:50 PM, dylancatlow wrote:
At 11/25/2014 6:54:21 PM, HououinKyouma wrote:
At 11/25/2014 1:55:50 PM, dylancatlow wrote:
In 1997, the world chess champion Garry Kasparov was defeated in a chess tournament by Deep Blue, a supercomputer developed by IBM. Many people came away with the idea that Deep Blue had outsmarted Kasparov. However, if you know anything about how Deep Blue made its moves, this seems rather absurd. Rather than deciding its moves "intelligently", Deep Blue simply generated millions upon millions of possible moves, assigned a rating to each move using an algorithm, and chose the move with the highest rating. Of course, assigning a useful rating to each move is no easy task. Each move in chess becomes a condition of subsequent moves, so determining a "good" move requires an astronomically high number of computations and an algorithm capable of recognizing good moves, as well as some mechanism for considering how moves might influence your opponent's moves, and in turn how that sets the stage for your moves, and so on. So although Deep Blue deserves credit for its computational power, it does not deserve credit for "outsmarting" Kasparov (it merely deserves credit for outplaying him). For is intelligence not precisely that which allows you to avoid blunt strategies when solving a problem? Shakespeare is considered a brilliant writer, not because he produced brilliant works per se, but because he produced brilliant works without needing to produce a trillion bad ones. The same goes for Deep Blue. It did not come up with good moves because it was intelligent, it came up with good moves because it came up with lots and lots of moves, some of which were good, and was able to recognize the good moves as "good" only because it had the computational power necessary to "force" the answer. The fact that Deep Blue came out barely ahead even though it had the advantage of choosing its moves from a pool trillions of times larger than Kasparov's just goes to show how much more intelligent a player Kasparov really was. Kasparov had to make his moves on the basis of far less data, and was thus only able to compete because he was able to make better use of that data i.e., because was able to more intelligently interpret the data. Deep Blue made up for its profound stupidity by repeatedly throwing itself against a wall until finally it broke through, while Kasparov simply climbed over. That is the difference.

There are some problems with your OP, the most important being that you're not comparing like with like. You're judging the "intelligence" of Deep Blue based not only on its moves but on its "thinking" process--to which we have access; whereas you judge Mr. K's "intelligence" based only on his actions--the moves in the game--without knowing, only intimating, what his thought process in the game actually was.


I'm saying that DB's moves do not reflect what is normally meant by intelligence, but rather reflects the sheer number of computations it can perform in a second. It's no different than any computational device, except that it's a lot faster. I say that Kasparov is more intelligent, not because I claim to know what his thought process is, but because he was able to overcome the computational gap. His thought process is irrelevant.

The question of Mr. K's thought process is relevant because otherwise all we have is a form of behavior that appears to be intelligent--Mr. K's--being compared with a "thinking" process--the computer's--that we do not accept as "intelligent"--in spite of behaving as if it were intelligent--because it does not satisfy our a priori notions of what intelligence is supposed to be.


Therefore, we cannot make a judgment about whether or not the computer was more intelligent than the player, since we know very little about the workings of the player's mind. Unless of course we assume that Mr. K also thinks in terms of algorithms and just lacks the speed and data available to the computer, in which case the computer would be more "intelligent".

If Kasparov can keep up with DB, then clearly he's coming up with his moves in a more intelligent way, otherwise he would simply be outcomputed.

But didn't he lose the game? Wouldn't that mean that the DB did in fact outcompute him?
"Here the ways of men part: if you wish to strive for peace of soul and pleasure, then believe; if you wish to be a devotee of truth, then inquire." F. Nietzsche.

"Freedom is always freedom for the one who thinks differently." R. Luxemburg.

"The principle of the masochistic left is that, in general, two blacks make a white, half a loaf is the same as no bread." G. Orwell, paraphrase.

"Islamophobia is a word created by fascists, used by cowards, to manipulate morons". Andrew Cummins.
dylancatlow
Posts: 12,255
Add as Friend
Challenge to a Debate
Send a Message
11/25/2014 7:29:09 PM
Posted: 2 years ago
At 11/25/2014 7:19:13 PM, HououinKyouma wrote:
At 11/25/2014 7:13:50 PM, dylancatlow wrote:
At 11/25/2014 6:54:21 PM, HououinKyouma wrote:
At 11/25/2014 1:55:50 PM, dylancatlow wrote:
In 1997, the world chess champion Garry Kasparov was defeated in a chess tournament by Deep Blue, a supercomputer developed by IBM. Many people came away with the idea that Deep Blue had outsmarted Kasparov. However, if you know anything about how Deep Blue made its moves, this seems rather absurd. Rather than deciding its moves "intelligently", Deep Blue simply generated millions upon millions of possible moves, assigned a rating to each move using an algorithm, and chose the move with the highest rating. Of course, assigning a useful rating to each move is no easy task. Each move in chess becomes a condition of subsequent moves, so determining a "good" move requires an astronomically high number of computations and an algorithm capable of recognizing good moves, as well as some mechanism for considering how moves might influence your opponent's moves, and in turn how that sets the stage for your moves, and so on. So although Deep Blue deserves credit for its computational power, it does not deserve credit for "outsmarting" Kasparov (it merely deserves credit for outplaying him). For is intelligence not precisely that which allows you to avoid blunt strategies when solving a problem? Shakespeare is considered a brilliant writer, not because he produced brilliant works per se, but because he produced brilliant works without needing to produce a trillion bad ones. The same goes for Deep Blue. It did not come up with good moves because it was intelligent, it came up with good moves because it came up with lots and lots of moves, some of which were good, and was able to recognize the good moves as "good" only because it had the computational power necessary to "force" the answer. The fact that Deep Blue came out barely ahead even though it had the advantage of choosing its moves from a pool trillions of times larger than Kasparov's just goes to show how much more intelligent a player Kasparov really was. Kasparov had to make his moves on the basis of far less data, and was thus only able to compete because he was able to make better use of that data i.e., because was able to more intelligently interpret the data. Deep Blue made up for its profound stupidity by repeatedly throwing itself against a wall until finally it broke through, while Kasparov simply climbed over. That is the difference.

There are some problems with your OP, the most important being that you're not comparing like with like. You're judging the "intelligence" of Deep Blue based not only on its moves but on its "thinking" process--to which we have access; whereas you judge Mr. K's "intelligence" based only on his actions--the moves in the game--without knowing, only intimating, what his thought process in the game actually was.


I'm saying that DB's moves do not reflect what is normally meant by intelligence, but rather reflects the sheer number of computations it can perform in a second. It's no different than any computational device, except that it's a lot faster. I say that Kasparov is more intelligent, not because I claim to know what his thought process is, but because he was able to overcome the computational gap. His thought process is irrelevant.

The question of Mr. K's thought process is relevant because otherwise all we have is a form of behavior that appears to be intelligent--Mr. K's--being compared with a "thinking" process--the computer's--that we do not accept as "intelligent"--in spite of behaving as if it were intelligent--because it does not satisfy our a priori notions of what intelligence is supposed to be.



His thought process is relevant but only to the extent that "his thought process be more sophisticated than DB's", which it quite obviously is. Otherwise, there's no way he could have won 2.5 games out of 6 given DB's sheer computational speed.

Therefore, we cannot make a judgment about whether or not the computer was more intelligent than the player, since we know very little about the workings of the player's mind. Unless of course we assume that Mr. K also thinks in terms of algorithms and just lacks the speed and data available to the computer, in which case the computer would be more "intelligent".

If Kasparov can keep up with DB, then clearly he's coming up with his moves in a more intelligent way, otherwise he would simply be outcomputed.

But didn't he lose the game? Wouldn't that mean that the DB did in fact outcompute him?

Of course it outcomputed him. The point is that if Kasparov used the same blunt-force strategy as DB, he would lose because DB was able to compute millions of times faster (outcomputes him). Therefore, his strategy must have been far more sophisticated, given that he won 2.5 games out of 6. He didn't have to waste his time computing everything because of his intelligence. DB, on the other hand, had no idea whether a move was good until it ran millions of simulations because it was stupid.
FaustianJustice
Posts: 6,240
Add as Friend
Challenge to a Debate
Send a Message
11/26/2014 2:12:10 AM
Posted: 2 years ago
DB didn't actually 'play' anything. It assigned odds to board position, calculated out odds based on the next board position, and opted for whatever board position would most frequently give it the win by playing thousands of games to their conclusion.

In this respect, DB lost many many MANY times in the course of the game, having played down that ultimately what yielded the highest odds might not net you the most gain at all times. There wasn't a 'game'. The opponent was just providing a new board position with which to give DB new calculations from.

Would DB have known if it's oppenent played an invalid move? That is to say, how knowing of the game was it to enforce rules, not just play by them?
Here we have an advocate for Islamic arranged marriages demonstrating that children can consent to sex.
http://www.debate.org...
fazz
Posts: 1,617
Add as Friend
Challenge to a Debate
Send a Message
11/26/2014 3:20:41 AM
Posted: 2 years ago
At 11/25/2014 6:51:12 PM, Envisage wrote:

Deep Blue was only meant to be short term project. It barely competed in computer chess tournaments, and was built explicitly to beat Kaparov. So it really isn't a surprise they didn't care for the logs.

To be honest, it seems very unlikely that there was intervention on Deep Blue, since today's programs find the 'Key Move' (which kasparov alluded to) rather quickly (in only a few seconds 'thought'), so it is not exclusivly a 'human move' as Kasparov thought.

Do you have a source for this claim? The source that IBM provides on its official website was a dead link.

True or not, it most certainly did impact the rest of the exhibition for Kasparov, since he was clearly frustrated, played sub-standard and in the last game made a rather trivial mistake (for a Super-GM, what he was) and lost.

On the flip side, Kasparov would probably lose 6-0, or at best 5.5-0.5 to a comparable computer today.

Maybe this is true but its is upto new generation to outwit the computers of today. Kasparovs legacy is sound.