This is the
talk page for discussing improvements to the
Deep Blue (chess computer) article. This is not a forum for general discussion of the article's subject. |
Article policies
|
Find sources: Google ( books · news · scholar · free images · WP refs) · FENS · JSTOR · TWL |
Archives: 1, 2 |
![]() | Deep Blue (chess computer) has been listed as one of the Engineering and technology good articles under the good article criteria. If you can improve it further, please do so. If it no longer meets these criteria, you can reassess it. | ||||||||||||||||||||||||
|
![]() | This ![]() It is of interest to the following WikiProjects: | |||||||||||||||||||||||||||||||||||||||||||
|
![]() | On 11 June 2023, it was proposed that this article be moved to Deep Blue. The result of the discussion was not moved. |
|
||
This article was the subject of a Wiki Education Foundation-supported course assignment, between 15 January 2020 and 6 May 2020. Further details are available
on the course page. Student editor(s):
The Tanner B. Peer reviewers:
Bobaylobor.
Above undated message substituted from Template:Dashboard.wikiedu.org assignment by PrimeBOT ( talk) 19:11, 16 January 2022 (UTC)
Is the fact that a youtube video was made about this really important to understand this event? Like, good for the guy who made it, but it seems somewhat out of place. Unless it leads to some larger discussion the match with Kasperov, I, at the moment, think it should be removed from the article. Argis113 ( talk) 02:53, 15 November 2020 (UTC)
Agreed. This seems more like advertisement. It should be enough if the video author's own wiki page links to the Deep Blue article, should he have one. Ludens123 ( talk) 21:22, 23 February 2021 (UTC)
The claim that Deep Blue contained 700,000 grandmaster games (back in the day!) sounded surprising to me, and does not have an explicit citation. Initially I was sceptical, but having looked through a bunch of the references to the article, I found a claim in the Hsu--Campbell "Deep Blue System Overview" (1995) paper (note date: this was presumably before the upgrade which got it beating Kasparov) that it had "an opening book created from a chess game database with 300,000 games". The paper does not seem to claim that they're grandmaster games, however if you're going to digitise games in 1995 then you may as well digitise grandmaster games, and note also that they are not claiming that all the games themselves are in the system, just that the 300,000 games were used to generate the opening book. Ultimately then, I am not suggesting that the claim is false, but I'd like to use this claim in a talk and I am a bit wary of citing it right now (I'll use the quote in the system overview pdf). Kevin Buzzard.
Was the Prozessor too simple or even in Power Saving mode 2A00:20:3001:91E4:E8C3:6BCC:5765:A6C0 ( talk) 23:07, 25 December 2021 (UTC)
GA toolbox |
---|
Reviewing |
Reviewer: Bwoodcock ( talk · contribs) 03:26, 4 January 2022 (UTC)
The article is not GA quality yet. It fails to clearly and unambiguously address or answer a fundamental question up-front: Is this an expert system, implemented as software on general-purpose hardware, or implemented partially or wholly in hardware, or is it a machine-learning system? Clues are buried way down in the Aftermath/Chess section and in the Design section, but this is a fundamental deficiency in the article. In general, citations are well provided and structured.
The two "Origins" paragraphs need copyediting by a native English speaker.
Use of the word "upgraded" is ambiguous and would seem to require further explication. Does it refer to more or faster hardware? Further training of a machine-learning system? Development of additional constraining rules?
The word "mistake" is likewise ambiguous. Is this being used in its commonly-understood sense, as in, Kasparov meant to move a piece from one square to another, but mistakenly moved it to a different square? Or Kasparov meant to execute one strategy, but got confused and made a move from a different strategy instead? Or is this some chess term-of-art with some other meaning? After watching the documentary, it appears that the "mistake" was instead a defensive one, in falling for a relatively common ploy. Yet such ploys would not exist if people didn't fall for them. So the word "mistake" seems to mischaracterize the situation here.
The attribution of the information about the software bug to Nate Silver is incorrect. Silver's extensive Wikipedia article makes no mention of his having any expertise in chess, nor do search engines find anything of substance at the intersection of "Nate Silver" and "chess." I've just watched the short Silver documentary, and it's clear in context that he's merely reiterating what the interviewees have said, and which they in turn attribute to the Deep Blue team, or are in fact members of the Deep Blue team. So, that needs to be fixed.
The passive voice needs to be fixed. The repetition ("was the first computer to face a world chess champion in a formal match") needs to be fixed.
Fritz is referred to earlier out-of-context but is internally linked to an explanatory article. In this section, however, "Deep Fritz" is referred to without any explanation. Why is this relevant to an article about Deep Blue? Are they related somehow?
What's a "plie?"
This section seems to be being used as a dumping-ground for trivia which hasn't been woven into a narrative. Which isn't a characteristic of a "good" article.
This finally gets to the meat of the article, in a burying-the-lede sort of way. It feels like a discussion of IBM's many cited refusals to disclose information would be worthwhile relative to the much different context of open-source development that obtains today. From a 21st-century perspective, that kind of caginess gives a "they must have had something to hide" sense, lending credibility to Kasparov's assertion of mechanical-turk intervention; but in the context of IBM as a slow-moving company in the mid-1990s, their behavior really wasn't that unusual. Proprietary closed-source software was still the norm, and trade secrecy was commonly employed and fought out in the courts.
The article seems close to GA. A thorough copyedit, some trimming of trivial factoids, and general promotion of core content up to the lede will get it there. A fun read, and the Nate Silver mini-documentary provided good context. Bill Woodcock ( talk) 03:26, 4 January 2022 (UTC)
Rate | Attribute | Review Comment |
---|---|---|
1. Well-written: | ||
![]() |
1a. the prose is clear, concise, and understandable to an appropriately broad audience; spelling and grammar are correct. | The writing is now clear, the organization is straight-forward, and it reads easily. |
![]() |
1b. it complies with the Manual of Style guidelines for lead sections, layout, words to watch, fiction, and list incorporation. | "Aftermath" is a little dramatic, a more neutral section name might be appropriate. It might also be useful to separate the two Kasparov matches each under their own subheads. |
2. Verifiable with no original research: | ||
![]() |
2a. it contains a list of all references (sources of information), presented in accordance with the layout style guideline. | The references seem comprehensive, and they're well and clearly formatted. And relatively uniform in their formatting. |
![]() |
2b. reliable sources are cited inline. All content that could reasonably be challenged, except for plot summaries and that which summarizes cited content elsewhere in the article, must be cited no later than the end of the paragraph (or line if the content is not in prose). | The references cited to support the hardware architecture of the RS/6000 were in conflict. (fixed) Per this discussion Mental Floss is not a reliable source. (fixed) Apparently the
New York Daily News isn't always a reliable source, but I don't see any problem with it here. |
![]() |
2c. it contains no original research. | I don't spot any claims that aren't backed by a citation. |
![]() |
2d. it contains no copyright violations or plagiarism. | Appears to me to properly summarize its sources. |
3. Broad in its coverage: | ||
![]() |
3a. it addresses the main aspects of the topic. | The article doesn't appear to miss anything I think a casual reader would want to know. My first reading of it was frustrating because of vagueness on what exactly Deep Blue was, and I think some of that has been addressed with more detail in the Design section. For a modern reader, who'd be very used to something like this being implemented as a machine learning system trained on winning and losing sides of historical games, I think it would be very useful to point out that this is an expert system, and that its success depended upon evaluating the state of the board and comparing it to a lot of stored rules about how the game could subsequently play out, devised by experts. |
![]() |
3b. it stays focused on the topic without going into unnecessary detail (see summary style). | The article spends quite a lot of space on name-dropping, which won't be of interest to most readers, though may be useful to people researching and trying to follow up in more detail. |
![]() |
4. Neutral: it represents viewpoints fairly and without editorial bias, giving due weight to each. | I don't see any examples of bias, and the controversial issues are handled fairly and well. |
![]() |
5. Stable: it does not change significantly from day to day because of an ongoing edit war or content dispute. | No evidence of any edit-warring here. |
6. Illustrated, if possible, by media such as images, video, or audio: | ||
![]() |
6a. media are tagged with their copyright statuses, and valid non-free use rationales are provided for non-free content. | Both seem plausible. |
![]() |
6b. media are relevant to the topic, and have suitable captions. | The illustration situation is ok, under the circumstances. The lead photo is actually of half of the cabinet, in the Computer History Museum, so that's appropriate. The photo of Kasparov is from an unrelated and dissimilar match a decade earlier. |
![]() |
7. Overall assessment. |
- agreed. Changed. Best Wishes, Lee Vilenski ( talk • contribs) 15:34, 11 January 2022 (UTC)
It looks like there's some confusion regarding the hardware architecture, which is understandable given that it went through at least three major generations, but I think some clarification is warranted. This source describes the 1995 state, without going into much detail about which model of PowerStation was used, since that wasn't really the operative part. Then you've got this which seems quite clear, and was undoubtedly taken directly from an IBM press release that would have been checked for technical accuracy before it hit the wires. That says that they went from a PowerPC 604 High 1 to a PowerPC 604e High 2 and the dates match up. Contradicting that, you've got this Summers and Winters of Artificial Intelligence reference, which you quote as saying it was a POWER2 Thin model. Which could also be possible, given the dates, but I can't check the reference because I don't have the paper book. Then you've got a discrepancy between whether there were 30 processor nodes or 32 (I assumed 32 was the correct number and changed the single reference to 30 before I realized how deep the discrepancies were running), and whether there were 512 VLSIs, or 480 VLSIs, or 512x32=16,384, or 480x30=14,400. So, I'd like to see that cleared up with some references that I can check without buying a $600 book. And, yes, I get that all of this is incidental to chess. But not to the subject of the article, which is the first major intersection of chess and computing. Bill Woodcock ( talk) 15:14, 11 January 2022 (UTC)
I just noticed this article... It doesn't have direct bearing, but might be worth a mention regarding what direction things have taken since the time of Deep Blue... What direction that work has sent us. Bill Woodcock ( talk) 12:37, 10 February 2022 (UTC)
I restored the longstanding lede description of Deep Blue as a supercomputer instead of the unhelpful relatively recent change to characterize it as an expert system. Chess playing computers have occasionally been described as expert systems, but that isn't especially common. The expert system article does not mention Deep Blue or chess playing computers or even any games playing computers at all. The link to the expert systems page didn't help the reader learn anything about Deep Blue at all. By contrast, the supercomputer article does mention Deep Blue specifically. The lede is supposed to summarize the important points in the body of the article, but "expert system" was mentioned nowhere except the lede. Quale ( talk) 06:48, 8 May 2022 (UTC)
In the article it says on the 44th move in the first game of the second match there was a bug. The source is also talking about 1997, first game: "[...]referred to an incident that had occurred toward the end of the first game in their 1997 match with Kasparov". This is the first game of the 1997 match: https://www.chessgames.com/perl/chessgame?gid=1070915
44. Rb1 is the only valid move in this position. There is no other move. Where is the bug? — Preceding unsigned comment added by 109.250.206.130 ( talk) 08:04, 13 September 2022 (UTC)
It makes no sense to me that Kasparov would be surprised when Deep Blue played the only possible move in this poition. Like, There are no other legal moves, so why should Kasparov be surprised. This makes no sense at all, something is not right here. — Preceding unsigned comment added by 109.250.206.130 ( talk) 08:07, 13 September 2022 (UTC)
the remark " The opening book encapsulated more than 4,000 positions and 700,000 grandmaster games " sets foot on a big compare that is impossible for a human being. is there any REAL information about the REAL software used? how many moved where result of a giant COMPARE function and how many maove were realy CALCUlated... instead of "holding up to the light and look for that one open pinchhole" but then very, very, very(etc) quick and are references or links to those computer routines ? — Preceding unsigned comment added by 85.149.83.125 ( talk) 17:13, 2 October 2022 (UTC)
I don't get the trend of classifying everything that has Alpha Beta search as brute force. Sure there is a large search and the speed (averaged) of 200 M positions per seconds is impressive but it is negligible compared to the search space. There is a great deal of smart pruning, I wouldn't call it mainly brute force.
Sources:
[1]IBM’S DEEP BLUE CHESS GRANDMASTER CHIPS
Behind Deep Blue, Feng-hsiung Hsu
And related articles to 1.
Pier4r (
talk)
22:15, 3 April 2023 (UTC)
The result of the move request was: not moved. Deep Blue (supercomputer) can be discussed separately. ( closed by non-admin page mover) CLYDE TALK TO ME/ STUFF DONE (please mention me on reply) 01:57, 19 June 2023 (UTC)
– Clear primary topic, by both page views and long term significance. Reviews of page views suggests the computer receives ten times the page views of all other Deep Blue's put together, and the long term significance of the others is negligible, while the long term significance of the computer - which revitalized optimism in the AI field - is significant. BilledMammal ( talk) 10:57, 11 June 2023 (UTC)