Jump to content

Talk:Deep Blue (chess computer)/GA2

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

GA Reassessment

[edit]

Article (edit | visual edit | history) · Article talk (edit | history) · Watch

Hi everyone. I was perusing the GAN list last week and was considering reviewing this article because it caught my interest. In real life, I am interested in computer science topics, so this is a topic that is particularly fascinating to me. After further review, I think there are a few additional issues with the article that need attention before the article can meet the good article criteria. Mz7 (talk) 20:43, 23 May 2020 (UTC)[reply]

Origins

[edit]
  • Overall, this section is written pretty well! The Hsu book looks like a really great source.

Green tickY

  • The third paragraph of the "Origins" section (the one beginning with "In 1995...") exclusively cites this primary source containing a tabulation of the results of the "8th World Computer Chess Championship". It seems like a number of the statements in this paragraph are not immediately verifiable through this link, such as the claim that "Fritz was running on an Intel Pentium 90 MHz" and the claim that Wchess and Junior were running on "personal computers". (It's also not clear what "personal computer" means in this context; we know what kind of computer specifically?)

Red XN not sure what to do there Signed,The4lines |||| (You Asked?) (What I have Done.) 15:47, 24 May 2020 (UTC)[reply]

I think this is problematic because it means that some of the information is not verifiable (i.e. fails WP:V). I haven't looked too deeply, but it looks like if you click the name of each chess computer at that link, the website has some more information about them: see [2] for Fritz. Mz7 (talk) 19:17, 4 June 2020 (UTC)[reply]

Design

[edit]
  • Part of my interest in this article comes from my interest in computer science, and personally I think the biggest area where this article could be improved is its coverage of the science and technical design behind Deep Blue. There seems to be no scarcity of coverage out there regarding the technical design of this computer—surely more than can be summarized in two sentences. Given the importance of this computer expressed in the lead, I feel that an effort should be made to expand this section of the article in order to satisfy the "broadness" criterion of the good article criteria.

Green tickY Done Signed,The4lines |||| (You Asked?) (What I have Done.) 00:05, 26 May 2020 (UTC)[reply]

  • I acknowledge that the "Deep Blue versus Kasparov" section below contains a bit more detail about the design of the computer (i.e. discussing the microprocessor and the program's evaluation function)—moving those details into the "Design" subsection may be a good starting point.

Green tickY Done, Signed,The4lines |||| (You Asked?) (What I have Done.) 23:12, 25 May 2020 (UTC)[reply]

  • Some other ideas for expansion: How was the design of this chess computer different from others? What are the specific features of the search algorithm's evaluation function? What techniques did the designers employ to optimize the algorithm for real-life application?

Green tickY Done, Signed,The4lines |||| (You Asked?) (What I have Done.) 00:05, 26 May 2020 (UTC)[reply]

  • I believe this claim's sourcing require an edit: "one of its developers even denied that it was AI." Forbes [12] cites and links to <https://www.washingtonexaminer.com/weekly-standard/be-afraid-9802>, which argues that DB's brute force calculation is not AI. Therefore BE AFRAID should be cited instead of Forbes. Recommend removing [12] and replacing with link to BE AFRAID.

Please forgive if I am not contributing correctly -- I'm not a frequent Wiki editor. — Preceding unsigned comment added by Jalster2 (talkcontribs) 23:27, 19 June 2020 (UTC)[reply]

  • Furthermore, I question the precision of the claim "one of its developers even denied that it was AI." C. Krauthammer, author of BE AFRAID develops this claim mostly. He interviewed the D.B. developer, Joe Hoane, who stated "No effort was devoted to [artificial intelligence in emulating human thought]. It is not an artificial intelligence project in any way. It is a project in—we play chess through sheer speed of calculation and we just shift through the possibilities and we just pick one line." It feel Hoane's statement may be more accurately cited if the 'emulation of human thought' aspect is included.

Please forgive if I am not contributing correctly -- I'm not a frequent Wiki editor. — Preceding unsigned comment added by Jalster2 (talkcontribs) 23:48, 19 June 2020 (UTC)[reply]

Deep Blue versus Kasparov

[edit]
  • The first paragraph of this section is unsourced. The second paragraph is also mostly unsourced, except for a single citation to this primary source. (I'm a little concerned with the reliance on primary sources in this article, but I suppose it's all right if there isn't any original research involved.)

Green tickY

Unless I'm missing something, the first paragraph seems to still be unsourced? Mz7 (talk) 19:14, 4 June 2020 (UTC)[reply]
  • As I stated in my "Design" section notes above, I think this section contains a number of details about the design of the computer that would fit better in the "Design" subsection, such as the paragraphs beginning with "The system derived..." and "Deep Blue's evaluation function..."

Green tickY Done, Signed,The4lines |||| (You Asked?) (What I have Done.) 18:05, 28 May 2020 (UTC)[reply]

  • This section could potentially include some additional commentary on the chess games themselves. There seems to be some of this at the Deep Blue versus Garry Kasparov article, but admittedly that article is mostly just a copy-paste of the move list.

Red XN ehh, dont think I should do that Signed,The4lines |||| (You Asked?) (What I have Done.) 21:22, 28 May 2020 (UTC)[reply]

@The4lines: Out of curiosity, why? Just glancing through, it looks like you've done good work so far; the article looks improved from when I looked at it a few weeks ago. Mz7 (talk) 19:14, 4 June 2020 (UTC)[reply]
@Mz7: Ah, because I thought that it may be boring for some users. Signed,The4lines |||| (You Asked?) (What I have Done.) 19:26, 4 June 2020 (UTC)[reply]
    • For example, Deep Blue won the deciding game after Kasparov made a mistake in the opening – what was the opening and what was the mistake? Any significant moves in the other games (I see there is that one "random" move mentioned later on)?

Red XN Dont know what to do there.

  • The last paragraph, which discusses the cheating allegations, solely cites this archived page from IBM's website, which contains links to commentaries on all six games. However, it is not immediately obvious to me (and to readers) where in the commentaries the cheating allegations are verified.

Green tickY Done, Signed,The4lines |||| (You Asked?) (What I have Done.) 21:22, 28 May 2020 (UTC)[reply]

Lead

[edit]
  • The lead cites multiple articles, including ones from the NY Daily News, Slate, and Mental Floss (admittedly I've not heard of Mental Floss, but I assume it's reliable). However, I noticed that these articles are not cited again in the body of the article. These seem like pretty good sources, so I recommend perhaps incorporating some of their content into the body of the article, then moving the citation from the lead to the body. Per MOS:LEAD, the lead section should be a summary of the rest of the article, and because the lead usually just repeats information that is cited later on in the article, it does not usually need to contain citations (MOS:LEADCITE).

Green tickY Done, Signed,The4lines |||| (You Asked?) (What I have Done.) 03:49, 1 June 2020 (UTC)[reply]

Citations

[edit]
  • Overall, I think the citation style of the article needs some tidying.
    • Cite note 6 is missing some important bibliographic information such as author, publisher, and publication date. It is also stated to be another another Mental Floss article, but clicking on the link it seems to be a book?
    • Cite notes 14, 17, 18, 22, 26, 28, and 29 appear to also be missing missing some important bibliographic information and should probably be reformatted with {{cite web}} to include bibliographic information like website name, publisher, publication date, etc.

Green tickY Done with both Signed,The4lines |||| (You Asked?) (What I have Done.) 15:45, 4 June 2020 (UTC)[reply]

  • @The4lines: Ah, it looks like some of the numbering has shifted. The main issue is with citations formatted like the last paragraph of the "Origins" section at the moment, where it looks like this:
<ref>[http://www.grappa.univ-lille3.fr/icga/tournament.php?id=29 Deep blue had white and lost to Fritz in 39 moves] {{webarchive|url=https://web.archive.org/web/20081007035001/http://www.grappa.univ-lille3.fr/icga/tournament.php?id=29 |date=7 October 2008 }}</ref>
Ideally, for consistency in citation style, we should reformat the citation with {{cite web}} and also include at minimum: the title of the source, the website name, and the accessdate. If available, we should also include the publisher name (if different from the website name), the author name, and the publication date. The archive URL is important because the original URL appears to be dead, and it can be added by filling in the |archive-url=, |archive-date=, and |url-status= parameters of cite web. So in total, something like this:
<ref>{{cite web |title=8th World Computer Chess Championship |url=http://www.grappa.univ-lille3.fr/icga/tournament.php?id=29 |website=ICGA Tournaments |accessdate=4 June 2020 |archiveurl=https://web.archive.org/web/20081007035001/http://www.grappa.univ-lille3.fr/icga/tournament.php?id=29 |archivedate=7 October 2008 |url-status=dead}}</ref>
There are a number of citations that need similar tidying scattered throughout the article—all of the references that use the {{webarchive}} template should probably be replaced by {{cite web}}. There is even one citation (cite note 15 currently, the New York Times article) that is just a bare URL. Mz7 (talk) 19:02, 4 June 2020 (UTC)[reply]

Preliminary review

[edit]
Preliminary review
GA review (see here for what the criteria are, and here for what they are not)
  1. It is reasonably well written.
    a (prose, spelling, and grammar): b (MoS for lead, layout, word choice, fiction, and lists):
  2. It is factually accurate and verifiable.
    a (reference section): b (citations to reliable sources): c (OR): d (copyvio and plagiarism):
  3. It is broad in its coverage.
    a (major aspects): b (focused):
  4. It follows the neutral point of view policy.
    Fair representation without bias:
  5. It is stable.
    No edit wars, etc.:
  6. It is illustrated by images and other media, where possible and appropriate.
    a (images are tagged and non-free content have fair use rationales): b (appropriate use with suitable captions):
  7. Overall:
    Pass/Fail:
Overall, I think the main issue with this article is the "broadness" criterion. There are some minor issues with citation style and verifiability, but it seems like there is a great deal of coverage about this computer out there in reliable sources (e.g. Google Scholar) that is ignored by this article, especially with respect to the computer's technical design and overall impact on the computer science academic field. Mz7 (talk) 20:43, 23 May 2020 (UTC)[reply]
(original reviewer comment) Hi, thanks for starting this. I admit that the citations are a bit lacking - I should have looked at them more in-depth. About the broadness though, I thought that for a layman like me, it covered the topic sufficiently - I personally was not left with any questions about Deep Blue after reading the article. Of course, I don't have the same amount of interest in the topic that you (and probably many other readers) have, so that may taint my view a little bit. Overall, though, I think the points you bring up are completely valid and I agree with the necessity of this re-review - I should have noticed them in my review. All the best, -- puddleglum2.0 21:56, 23 May 2020 (UTC)[reply]
Mm, I agree that the article doesn't need to get too technical, but nevertheless I think the "Design" subsection specifically should be expanded. After looking for a bit, I found this article published by IEEE which looks like it goes into detail in the computer's design (starting on page 72). It looks like it's targeted to people with some computer science/electrical engineering background, but I want to clarify that I think we don't necessarily have to limit ourselves to academic sources. For example, this NYT article looks like it mentions some of the hardware design (e.g. # of processors) and the Forbes article already cited in the section describes some of the software design (i.e. the "brute force" approach). Mz7 (talk) 22:18, 23 May 2020 (UTC)[reply]
Right, that makes sense, now that you've clarified a bit, I completely agree, that's a problem that should be fixed. Not coming from a computer-y background at all, that demographic didn't come to mind during my review, but now that you point all that out, I definetely think that's a problem that needs to be adressed. Thanks again for the reassesment! -- puddleglum2.0 00:54, 24 May 2020 (UTC)[reply]

Mz7 Done, Thanks, Signed,The4lines |||| (You Asked?) (What I have Done.) 02:44, 3 June 2020 (UTC)[reply]

Mz7 I think I got most of them. Signed,The4lines |||| (You Asked?) (What I have Done.) 19:00, 8 June 2020 (UTC)[reply]


Update: 10 June 2020

[edit]
Extended content
GA review (see here for what the criteria are, and here for what they are not)
  1. It is reasonably well written.
    a (prose, spelling, and grammar): b (MoS for lead, layout, word choice, fiction, and lists):
  2. It is factually accurate and verifiable.
    a (reference section): b (citations to reliable sources): c (OR): d (copyvio and plagiarism):
  3. It is broad in its coverage.
    a (major aspects): b (focused):
  4. It follows the neutral point of view policy.
    Fair representation without bias:
  5. It is stable.
    No edit wars, etc.:
  6. It is illustrated by images and other media, where possible and appropriate.
    a (images are tagged and non-free content have fair use rationales): b (appropriate use with suitable captions):
  7. Overall:
    Pass/Fail:

I apologize for my delay following up here. Following the expansion of the "Design" section and the general reorganization of the article to improve the coverage of the technical background of the computer, I'm willing to say that "broadness" is likely covered now. There remain a few issues, however, with referencing.

  • Cite note 1: it looks like you filled in the wrong author. Chris Higgins wrote the article, not Kevin Warwick.

Green tickY

  • Cite note 13: another bare URL citation that should be replaced by {{cite web}}

Green tickY

  • Cite note 28: Web Archive is not the name of the website here; IBM or IBM Research is. Instead, you should use the |archive-url= and |archive-date= parameters to list the archived URL, i.e. <ref>{{cite web |title=Deep Blue - Replay the Games |url=http://www.research.ibm.com/deepblue/watch/html/c.shtml |website=IBM Research |accessdate=10 June 2020 |archiveurl=https://web.archive.org/web/20080701232743/http://www.research.ibm.com/deepblue/watch/html/c.shtml |archivedate=1 July 2008 |url-status=dead}}</ref>

Green tickY

  • Cite note 33 cites a website called Copmuter history museum—I suspect this is a typo. Make sure to capitalize it as Computer History Museum since it's a proper noun.

Green tickY Ding Ding Ding! You got it :)! Thanks for seeing that. Signed,The4lines |||| (You Asked?) (What I have Done.) 16:21, 11 June 2020 (UTC)[reply]

  • The last paragraph of the "Origins" section doesn't seem to be immediately verified by its citation. As I mentioned above, it looks like if you click the name of each chess computer at that link, the website has some more information about them: see [3] for Fritz.

Green tickY

  • The second paragraph of the "Design" section cites the "TOP500 List - June 1997 (201-300)" list, which appears to verify the last sentence, but it doesn't seem to provide support for anything else in the paragraph.

Green tickY

  • The first paragraph of the "Deep Blue versus Kasparov" section still appears to be unsourced.

Green tickY Please me know if you have any questions. I really appreciate that you've taken the time to address my concerns so far! I do think the referencing concerns here are important, however, and I would like to see them fixed before the article can pass this GA reassessment. Mz7 (talk) 21:47, 10 June 2020 (UTC)[reply]

@Mz7: Thanks for you rewivew! I think I'm done. Best, Signed,The4lines |||| (You Asked?) (What I have Done.) 02:27, 13 June 2020 (UTC)[reply]
@The4lines: For the last paragraph of the "Origins" section, I mentioned the Fritz link as an example, but I didn't look too deeply to check whether it verifies everything in that paragraph. Looking now, the article claims things like Fritz was running on an Intel Pentium 90 MHz personal computer, Wchess was running on a personal computer, and Junior was running on a personal computer—where, if at all, are these statements verified in the source? Mz7 (talk) 00:10, 15 June 2020 (UTC)[reply]
@Mz7: I could not fix it so I removed it. Best, Signed,The4lines |||| (You Asked?) (What I have Done.) 03:10, 15 June 2020 (UTC)[reply]
The4lines, sounds good to me. Do you have a specific page number for the book by Newborn that you added? We should ideally make the citations as specific as possible. Similarly, for the JSTOR article that you cited, do you think you could refer me to where in the article it verifies the information in the article? I think for the computer's technical design, this is a really good source (I linked it earlier in my conversation with Puddleglum). Mz7 (talk) 21:27, 17 June 2020 (UTC)[reply]
@Mz7: Great, thanks! Done with the updates. Signed,The4lines |||| (You Asked?) (What I have Done.) 22:45, 17 June 2020 (UTC)[reply]
The4lines, I think you missed my question about the JSTOR article. The article describes Deep Blue as a massively parallel, RS/6000 SP Thin P2SC-based system with 30 nodes, with each node containing a 120 MHz P2SC microprocessor, but it wasn't immediately clear to me where this and other details in the paragraph are verified in the article. Mz7 (talk) 01:13, 18 June 2020 (UTC)[reply]
Mz7 Yeah, I replaced with the PDF you said earlier. It works just as fine. Best, Signed,The4lines |||| (You Asked?) (What I have Done.) 02:52, 18 June 2020 (UTC)[reply]
@The4lines: Here's the thing though. Have you actually checked whether or not the source itself verifies the information in the paragraph? This is now the third source that has been offered to support the information, and you haven't provided any clarification on this question. Mz7 (talk) 05:20, 18 June 2020 (UTC)[reply]
Mz7 Yes I don't know what you mean by and you haven't provided any clarification on this question. Signed,The4lines |||| (You Asked?) (What I have Done.) 14:47, 18 June 2020 (UTC)[reply]
The4lines, the whole point of this exercise is to ensure that the content that is written in the article can be verified by our readers through the citation. Above, I pointed out to you that the information about the computer's design in that paragraph is not supported by the Top 500 list source, so you added the JSTOR source instead. However, I pointed out once again that this source doesn't seem to verify the information in the article, so you replaced the source once again with the source I provided. This is now the second time you have replaced the citation without changing anything at all in the paragraph, and this back-and-forth seems to indicate to me that you are not carefully reading the source you are adding in order to check whether the content in the article is verified by the source. The paragraph in question makes multiple interesting claims, but they all need to be factually verifiable in order for the article to pass as a good article. Please tell me exactly where in the sources each sentence in that paragraph can be verified. Mz7 (talk) 17:49, 18 June 2020 (UTC)[reply]
Mz7 Ok, sorry. I added a source that does say what the article says, and added the pages. Best, Signed,The4lines |||| (You Asked?) (What I have Done.) 01:30, 19 June 2020 (UTC)[reply]

Delisted

[edit]

Unfortunately, I believe the outcome of this reassessment should be to delist the article. This was not an easy decision for me to make. As you can see, it took a month of mulling things over, and at various points in the process, I was leaning towards keeping the article as a GA. Additionally, I respect the judgment of the GA reviewer before me, so I did not want to overrule her decision capriciously.[a]

I acknowledge that many of the issues that I identified at the start of the GAR have been mitigated over the course of the last month, and because of this, I thank The4lines for his diligent work. I definitely believe that the article today is a substantial improvement over when it was first reviewed. The "broadness" issue has been partially mitigated by some restructuring of the article, moving technical details about the design of the computer from the section about the Kasparov game to the "Design" section, which was upgraded from heading level 3 to level 2. Many of the most glaring issues with the article's citation style have also been mitigated by using standard citation templates where possible and filling in dead links with archived links.

However, the article still has issues with clarity of the prose and verifiability, especially in the "Design" section, enough that I feel that it would be most prudent to delist the article while these issues are addressed through general editing.

  • In the "Design" section, Jalster2 above correctly points out that the claim that one of its developers even denied that it was artificial intelligence does not appear to be 100% accurate given the source.
  • The claim that the computer employs GOFAI (Good Old-Fashioned Artificial Intelligence) rather than of deep learning which would come a decade later is also unsourced.
  • The second paragraph is one of the most problematic. The paragraph was copied directly from a paragraph that was originally in the "Deep Blue versus Kasparov" section before The4lines nominated this article [4]. Originally, the only source that it cited was this Top 500 list source, which fails to verify all of the content in the paragraph.
    After I pointed this out, there were at least three attempts to provide an alternative source for the information, but with each source, it wasn't obvious if all of the paragraph was verifiable (see #Update: 10 June 2020 above). The most recent source, this "Encyclopedia of Information Science" is quite interesting because it appears to be a direct copy-paste of this Wikipedia article! Because the paragraph as written contains longstanding text that was present on Wikipedia since at least 2004, this leads me to think the source may have either plagiarized Wikipedia or used Wikipedia circularly. This paragraph contains the meat of the Design section—it explains some of the fundamental design choices made by IBM, so it's important that we get it right. I suspect the section may need to be rewritten to be verifiable.
  • The third paragraph cites this NYT article for much of its content, but it seems that it only verifies the sentence that cites it (and maybe the sentence before). There are multiple claims that appear to be unsourced, e.g. The evaluation function had been split into 8,000 parts and In the opening book there were over 4,000 positions and 700,000 grandmaster games. The endgame database contained many six-piece endgames and five or fewer piece positions..

In my view, this article would substantially benefit from attention from editors who are more familiar with the topic areas encompassed by this article.[b] This would especially be useful to help improve the article's discussion of the technical design of the computer, particularly to allow it to convey a broad amount of information while still being understandable by non-technical readers. Additionally, it would also be useful to improve the coverage of the chess behind the subject. Earlier, when I suggested this, the GA nominator rejected the idea on the basis that it may be boring for some users [5]. I respectfully disagree. The chess theory behind the matches—e.g. commentary on the mistakes that Kasparov made, the moves that Deep Blue made, etc.—would surely be fascinating to readers interested in this subject and arguably expected for a good article, especially considering the Kasparov matches essentially define the notability of this computer. There is certainly no scarcity of coverage containing this commentary out there, based on my cursory searches.

For these reasons, I believe that this article fails to meet the good article criteria at this time, and I hope that my recommendations throughout this page are helpful for improvement of the article. Mz7 (talk) 06:36, 21 June 2020 (UTC)[reply]

Footnotes

  1. ^ The initial GA review of this article missed a number of significant issues, many of which we have discussed here at the GAR over the past month. The most glaring issue at the start of this reassessment was the relative lack of coverage about the computer science aspects of Deep Blue, e.g. its technical design, challenges faced by the team that created it, and relative importance within its field in computer science. At the start of the GAR, I noted my belief that these omissions caused the article to fail the "broadness" GA criterion, and after a brief discussion, the original GA reviewer also agreed with this assessment. There were also numerous unsourced statements throughout the article, and many of the citations that were present were disorganized. Some citations had bare URLs, others lacked important bibliographic details, and there were even some citations that either had incorrect bibliographic details or did not verify the content that they were supposed to verify.
  2. ^ I would also like to note that the GA nominator, The4lines, did not contribute to this article at all before nominating the article for GA status [1]. This fact has no bearing on the actual quality review of the article; however, it does provide context and an explanation for why the article presented with so many issues. It is clear to me that this article was still very much a "work-in-progress" at the time it was nominated (see permalink), and as such, it seems apparent that the original authors of the article did not intend for the article to be nominated in the state it was in. Avoiding premature nominations such as this is precisely why WP:GAN/I states that it is preferable that nominators have contributed significantly to the article and are familiar with its subject and its cited sources. Nominators who are not significant contributors to the article should consult regular editors of the article on the article talk page prior to a nomination.