Did a bug in Deep Blue lead to Kasparov's defeat?

[hyatt]

Sounds like idle speculation, as I'd think something like that would certainly show up in Hsu's book...

Say again?

First you take the story way too literally, as if the Deep Blue team put in a routine that under circumstances deliberately selects a random move, while a reasonable interpretation of the story would be that Deep Blue had a bug that when triggered resulted in an essentially random move selection (very different).

Then when this is pointed out, you can't simply accept that the story might not be "nonsense", so you backpedal slightly into the position that it is virtually impossible (a coin landing on its edge) that Deep Blue had a serious bug! Oh wow, how many engines are there that have no serious bug that only pops up once every hundred or thoussand games or so? Clearly your new position is absolutely untenable, and I'm sure you would never haven taken this position in the first place, if you had not started by declaring the story "nonsense". But you did start like that, so you have to cement yourself in...

As a psychoanalyst you make a great proctologist. I didn't say a thing about "I can't accept that". I said "it sounds like idle speculation". I did not remember it appearing in Hsu's book and said so. If it did, my mistake. Big deal. I declared the story nonsense because the basic premise of the story was that psyched Kasparov out so badly that he blew the match. THAT I do not buy into. As I said, "speculation". Did a strange move in round 1 cause him to transpose moves into a lost position in game 6? Really a stretch.

I've never seen as much excuse-making as after that match. He simply blew his preparation, was not expecting something of the level of DB's hardware, and he paid for it. I think the only valid excuse he had was "I should never have trusted the ChessBase experts when they told me how to prepare..."

If the DB guys had such a bug, which I presume (now) that they did, given all the evidence cited, I was wrong. I have both his original manuscript that he asked me to proof-read, and the final copy of the book. I read both well over 10 years ago. I forgot something. Sue me... or go complain about it. It is not a big deal. To most, anyway.

[BB+]

The relevant pages in Behind Deep Blue are 223 and 224, where it says "Then it suffered the same bug that caused Deep Blue Jr to toss a pawn against Chrisitiansen in the training match. Instead of playing 44. ... Rf5 and prolonging the game as it intended [what does this mean?!], it played the instantly losing move 44. ... Rd1 [...] That night, Joe worked late and finally fixed the bug that had haunted us for over a month." Then on the next page (224): "Deep Blue played it as a result of a bug", and then there is a section A Bad Move That Wins (more about Ra6 leading to resignation in Game 2), that says "Joe has already fixed the bug responsible for the last move in game one", and then discusses other debated changes (the potential open file being of import). So it definitely indicates there was a bug, but not specifically what it might cause DB to do.

Backing up to the game against Christiansen (page 202), one does indeed read there that: In the first game, Deep Blue Jr set up a promising position, and then suddenly gave up a pawn for no apparent chess reason. It turned out to be a software bug causing it to play a random move. [...] The Deep Blue Jr bug was related to how time control was handled. [...]

[syzygy]

I declared the story nonsense because the basic premise of the story was that psyched Kasparov out so badly that he blew the match. THAT I do not buy into.

Ok, so you change your position for the 2nd time... Just look at what you wrote the first time:

Where does that nonsense come from? Who would program a chess engine to choose a "random" move under some circumstance? Not me. Not Hsu. No idea where that might originate.

Doesn't look like you declared the story nonsense because of "THAT basic premise"... Neither did your other comments in this thread.

[hyatt]

I declared the story nonsense because the basic premise of the story was that psyched Kasparov out so badly that he blew the match. THAT I do not buy into.

Ok, so you change your position for the 2nd time... Just look at what you wrote the first time:

Where does that nonsense come from? Who would program a chess engine to choose a "random" move under some circumstance? Not me. Not Hsu. No idea where that might originate.

Doesn't look like you declared the story nonsense because of "THAT basic premise"... Neither did your other comments in this thread.

Read whatever you want. I read the story linked to...

First quote:

"But did the machine merely psych him out? Statistician Nate Silver's new book "The Signal and The Noise: Why So Many Predictions Fail--But Some Don't" contains an anecdote about how a glitch in Deep Blue may have led Kasparov to overestimate the machine's smarts, according to The Washington Post. "

Second quote:

"Kasparov may have been spooked by that, but his concern turned to panic in the second game, in which Deep Blue started playing much less like a computer and more like a human grandmaster. Kasparov lost, drew the next three games, and collapsed during the sixth, losing the epic battle. "

All idle speculation. Never seen Kasparov make any such statement at all. And the "collapse" in game 6 looked more like a simple transposition of moves type mistake than a panic attack.

I will repeat. I have never seen so much excuse-making for Kasparov's loss. And MOST of it is not coming from Kasparov. Very similar to all the excuse-making for Vas, without his offering a single comment to defend himself against the claims.

You can have the last word. I DO change my opinion when I discover something new or expose a mistake in my reasoning. You act like that is abnormal. You might ask yourself why you feel that way...

[syzygy]

I DO change my opinion when I discover something new or expose a mistake in my reasoning. You act like that is abnormal.

It is fine if you change your position and are honest about it. But you were trying pretty hard to avoid that:

I declared the story nonsense because the basic premise of the story was that psyched Kasparov out so badly that he blew the match. THAT I do not buy into.

No, you did NOT declare the story nonsense because of THAT. You declared it nonsense because of something that evidently you were wrong about. Mistakes happen, but then don't suggest that you were talking about something else all along.

[lmader]

I DO change my opinion when I discover something new or expose a mistake in my reasoning. You act like that is abnormal.

It is fine if you change your position and are honest about it. But you were trying pretty hard to avoid that:

I declared the story nonsense because the basic premise of the story was that psyched Kasparov out so badly that he blew the match. THAT I do not buy into.

No, you did NOT declare the story nonsense because of THAT. You declared it nonsense because of something that evidently you were wrong about. Mistakes happen, but then don't suggest that you were talking about something else all along.

Seems like we're getting awfully literal about casual posting. Pretty soon we'll all have to have an attorney vet our text before posting

For what it's worth I read Bob's first posts as coming off a little too strong against the idea that there was a "random" move in the game, however I read his "idle speculation" quote to be in the context of the idea that this move rattled Kasparov. I have to say, when I read the article linked to in the OP, I saw that too, and thought the same thing - I don't ever remember hearing Kasparov talking about a move in game 1 rattling him or affecting his confidence. That part I am still skeptical about and think is probably "idle speculation", but I could be wrong.

OTOH, I do think game 2 freaked Kasparov out. There were two moves in that game that he thought were not computer-like and/or beyond the computer, and he went so far as to accuse the Deep Blue team of cheating (human intervention). I know that computers today find those moves, but it seems like it still takes them a long time to select those moves... so I kind of wonder about that too. Did the Deep Blue team freak out after getting their hat handed to them in the first game? I know it's a reach...

Cheers

[hyatt]

I DO change my opinion when I discover something new or expose a mistake in my reasoning. You act like that is abnormal.

It is fine if you change your position and are honest about it. But you were trying pretty hard to avoid that:

I declared the story nonsense because the basic premise of the story was that psyched Kasparov out so badly that he blew the match. THAT I do not buy into.

No, you did NOT declare the story nonsense because of THAT. You declared it nonsense because of something that evidently you were wrong about. Mistakes happen, but then don't suggest that you were talking about something else all along.

I was talking about the "Kasparov excuse" all along. I did say I did not think a random move happened. I fixed that when BB+ provided a quote, and after I looked at Hsu's book. The basic premise of the article is still idle speculation / nonsense, however.

I was talking about BOTH. IE the entire article. One point being wrong does not change the other at all...

[syzygy]

I DO change my opinion when I discover something new or expose a mistake in my reasoning. You act like that is abnormal.

It is fine if you change your position and are honest about it. But you were trying pretty hard to avoid that:

I declared the story nonsense because the basic premise of the story was that psyched Kasparov out so badly that he blew the match. THAT I do not buy into.

No, you did NOT declare the story nonsense because of THAT. You declared it nonsense because of something that evidently you were wrong about. Mistakes happen, but then don't suggest that you were talking about something else all along.

Seems like we're getting awfully literal about casual posting. Pretty soon we'll all have to have an attorney vet our text before posting

Anything you say can and will be used against you
I will immediately admit that most of the above is nitpicking on something terribly unimportant. But of course I had to take the last word he gave me, so that he could take it back again

Anyway, I think it is interesting that Bob's initial denial of the possibility of a bug (coin landing on its edge) together with his analysis of Rd1 (Rd1 might have been the best move DB saw, implying that DB searched incredibly deep), does lend some credibility to the idea that Kasparov's team got a false impression of the strength of DB. It is of course silly to blame this bug for Kasparov's loss, but it could be a factor that overall happened to work against Kasparov. It also gives a somewhat plausible explanation for why Kasparov resigned in game 2 in a position he might still have been able to draw. The blame for that loss still goes to Kasparov of course, in the end it was his sole responsibility, but as a partial explanation for his thought process at the time I don't find it completely implausible.

Given that the story explicitly states "But Silver speculates", we know that the bug <-> loss connection is just speculation. But the speculation has a basis in facts: the bug did happen.

[lmader]

Anything you say can and will be used against you
I will immediately admit that most of the above is nitpicking on something terribly unimportant. But of course I had to take the last word he gave me, so that he could take it back again

Anyway, I think it is interesting that Bob's initial denial of the possibility of a bug (coin landing on its edge) together with his analysis of Rd1 (Rd1 might have been the best move DB saw, implying that DB searched incredibly deep), does lend some credibility to the idea that Kasparov's team got a false impression of the strength of DB. It is of course silly to blame this bug for Kasparov's loss, but it could be a factor that overall happened to work against Kasparov. It also gives a somewhat plausible explanation for why Kasparov resigned in game 2 in a position he might still have been able to draw. The blame for that loss still goes to Kasparov of course, in the end it was his sole responsibility, but as a partial explanation for his thought process at the time I don't find it completely implausible.

Given that the story explicitly states "But Silver speculates", we know that the bug <-> loss connection is just speculation. But the speculation has a basis in facts: the bug did happen.

Yeah, I see what you mean. I am definitely under the impression, based on the interviews with Kasparov, that he was indeed rattled by the end of the 2nd game. I was just always under the impression that it was the two extremely positional moves during game 2 that unnerved him, not something from game 1, especially given that he won the 1st game. And I don't recall ever hearing an interview where he mentioned that as a problem for him from game 1, but maybe someone can find it. It is possible that that had already sown some of the seeds that undermined his confidence.

[hyatt]

Anything you say can and will be used against you
I will immediately admit that most of the above is nitpicking on something terribly unimportant. But of course I had to take the last word he gave me, so that he could take it back again

Anyway, I think it is interesting that Bob's initial denial of the possibility of a bug (coin landing on its edge) together with his analysis of Rd1 (Rd1 might have been the best move DB saw, implying that DB searched incredibly deep), does lend some credibility to the idea that Kasparov's team got a false impression of the strength of DB. It is of course silly to blame this bug for Kasparov's loss, but it could be a factor that overall happened to work against Kasparov. It also gives a somewhat plausible explanation for why Kasparov resigned in game 2 in a position he might still have been able to draw. The blame for that loss still goes to Kasparov of course, in the end it was his sole responsibility, but as a partial explanation for his thought process at the time I don't find it completely implausible.

Given that the story explicitly states "But Silver speculates", we know that the bug <-> loss connection is just speculation. But the speculation has a basis in facts: the bug did happen.

Yeah, I see what you mean. I am definitely under the impression, based on the interviews with Kasparov, that he was indeed rattled by the end of the 2nd game. I was just always under the impression that it was the two extremely positional moves during game 2 that unnerved him, not something from game 1, especially given that he won the 1st game. And I don't recall ever hearing an interview where he mentioned that as a problem for him from game 1, but maybe someone can find it. It is possible that that had already sown some of the seeds that undermined his confidence.

Note that there were NO "positional" moves where he questioned the choices. I posted analysis elsewhere (RF) about this issue. He was convinced that Qb6 instead of Be4 was a mistake, that Qb6 won a pawn. It doesn't. Try your favorite program. Here's crafty on old hardware on that position:
21-> 2.78 1.12 37. Qb6 Rxa2 38. Rxa2 Bc7 39. Qe6+ Kf8 40. Kf1 Rd8 41. Ra7 Bd6 42. Rb7 Be7 43. Rb6 Qd7 44. Qxd7 Rxd7 45. Rxb5 Bd6 46. Ke2 Ke7 47. Ra5 Rb7
22 3.74 1.13 37. Qb6 Rxa2 38. Rxa2 Bc7 39. Qe6+ Kf8 40. Kf1 Rd8 41. Ra7 Bd6 42. Rb7 Be7 43. Qc6 Rd6 44. Qxb5 Qxb5 45. Rxb5 Ra6 46. Ra5 Rxa5 47. bxa5 Bd6 48. a6
22-> 4.02 1.13 37. Qb6 Rxa2 38. Rxa2 Bc7 39. Qe6+ Kf8 40. Kf1 Rd8 41. Ra7 Bd6 42. Rb7 Be7 43. Qc6 Rd6 44. Qxb5 Qxb5 45. Rxb5 Ra6 46. Ra5 Rxa5 47. bxa5 Bd6 48. a6
23 6.06 -- 37. Qb6? (<+0.97)
<problem has been found>
23 6.43 -- 37. Qb6? (<+0.81)
23 7.44 -- 37. Qb6? (<+0.49)
23 10.11 0.51 37. Qb6 Rxa2 38. Rxa2 Ra8 39. Ra5 Rxa5 40. bxa5 Qb8 41. Kf1 Qxb6 42. axb6 Kf8 43. Ke2 Ke8 44. Kf3 Kd7 45. Ke4 Bc5 46. b7 Kc7 47. d6+ Kxb7 48. d7
23 29.40 0.70 37. Be4 Rxa2 38. Qxa2 Kf8 39. Qa7 Rb8 40. Ra6 Qe7 41. Kf2 Qxa7+ 42. Rxa7 Rb6 43. g4 Bb8 44. Rd7 Bd6 45. Ke3 Ra6 46. Bf3 h5 47. Ke4 hxg4 48. hxg4 Kg8
23-> 32.48 0.70 37. Be4 Rxa2 38. Qxa2 Kf8 39. Qa7 Rb8 40. Ra6 Qe7 41. Kf2 Qxa7+ 42. Rxa7 Rb6 43. g4 Bb8 44. Rd7 Bd6 45. Ke3 Ra6 46. Bf3 h5 47. Ke4 hxg4 48. hxg4 Kg8


At the 6 second mark, Crafty fails low on Qb6. And then quickly chooses Be4 as best. DB's log shows the SAME behavior, it just took longer back then. But it failed low tactically and then switched. Here's DB's log file for reference, for move 37 in game 2:

9(6) #[Qb6](53)#################################### 53 T=9
qf2b6 Ra8a2r ra1a2R Bd6c7 qb6e6 Kg8h8 bc2e4 Rc8a8 kg1h2 Ra8b8 pg2g3 Qe8f8
10(6) #[Qb6](55)#################################### 55 T=33
qf2b6 Ra8a2r ra1a2R Bd6c7 qb6e6 Kg8h8 bc2e4 Qe8f8 kg1h1 Bc7d6 ra2a6 Rc8d8 ra6a7
11(6)<ch> 'ab'
#[Qb6](32)########[Be4](37)############################ 37 T=182
bc2e4 Rc8b8 pg2g3 Qe8d8 ra2a6 Ra8a6r ra1a6R Bd6c7 ra6f6P
12(6)[TIMEOUT] 37 T=199
bc2e4 Rc8b8 pg2g3 Qe8d8 kg1g2 Ra8a2r ra1a2R Bd6c7 qf2a7 Bc7b6 qa7a6 Qd8d7
---------------------------------------
--> 37. Be4 <-- 3/37:56
---------------------------------------

As you can see, it liked Qb6 until Depth 11 where it failed low and was replaced by Be4. The "#" characters are just things printed as each root move is searched, a sort of "I am alive and thinking" alert...

Nothing odd. But it shows something important. Kasparov claimed Qb6 won a pawn and any computer would play that move. he was TACTICALLY WRONG. It doesn't win a pawn as we can see in seconds on hardware that is 15+ years more current. So it wasn't "positional play" that psyched him. He psyched himself as he was tactically inferior to DB in this position. And rather than accept that, he decided DB had help. It did. It is called alpha/beta-minimax + fast hardware...

After game one, he was all grin and rolling along. After game 2, the grin was gone. Where his "suspicion" came from only he knows.