The following discussion is an archived debate of the proposed deletion of the article below. Please do not modify it. Subsequent comments should be made on the appropriate discussion page (such as the article's talk page or in a deletion review). No further edits should be made to this page.

The result was keep. Nominator has withdrawn but IMHO this discussion has been open long enough and has enough participation for a "keep" close. (non-admin closure) Ron Ritzman (talk) 00:19, 8 October 2009 (UTC)[reply]

Fowler-Noll-Vo hash function (edit | talk | history | protect | delete | links | watch | logs | views) – (View log)

(Find sources: Google (books · news · scholar · free images · WP refs) · FENS · JSTOR · TWL)

Article is unsupported by any reliable sources. The article topic is not notable enough for reliable sources right now. It does not seem that this topic should be in the wikipedia at present- prime-based hash functions are two a penny. Moreover, this article links only to sources controlled by User:Landon_Curt_Noll and is largely written by him. Phil Spectre (talk) 23:21, 28 September 2009 (UTC)[reply]

The new, shorter version is acceptable, so I'd like to withdraw this AfD. I'm not sure how to do it officially, though. Is it ok for me to just remove the notice on the page? Phil Spectre (talk) 01:25, 7 October 2009 (UTC)[reply]

• I don't support deletion of this article. The article has 24 different editors (not counting bots) so I don't think it is fair to say that it "is largely written by [Landon_Curt_Noll]", indeed the article was not even originally created by Landon_Curt_Noll. Good non-cryptographic hash functions are not "two a penny", they are hard to design to get good results. Most hash function designers concentrate upon cryptographically strong hashes that are a lot slower. So there is definitely a place for the non-cryptographic alternatives that are faster. The FNV happens to fulfil a requirement in non-secure situations. The fasm project is well known and uses FNV for internal table lookups. From what I have seen the FNV hash has been around for some time now and seems to be quite well respected. HumphreyW (talk) 01:02, 29 September 2009 (UTC)[reply]
  • The criteria for an article are WP:RS and WP:N, not whether you think it's a good hash. Notability is not earned by association: there is a precedent that being used by a notable project doesn't make the hash notable. And I looked at fasm but it doesn't even mention FNV. As for reliability, all of the sources are just Noll. Phil Spectre (talk) 01:12, 29 September 2009 (UTC)[reply]
    • Sorry, I didn't realise the fasm article does not mention it. But the authors site has it mentioned here. HumphreyW (talk) 01:57, 29 September 2009 (UTC)[reply]
      • The way I see it, the fact that fasm didn't, and still doesn't, mention FNV means that this hash isn't all that important a part. Either way, there are still no reliable sources and it's not notable. Phil Spectre (talk) 02:21, 29 September 2009 (UTC)[reply]
        • Well of course an assembler is not going to make big news about the internal hash functions used, that wouldn't make sense. I am not sure how one would really objectively judge notability, but a quick Google search returns 154000 results. If even 1/10th of those results are unique then that seems significant to me. There are a large number of pages linking to the Wikipedia FNV page, deleting the FNV page would break a significant number of websites. In the Google results we can find mentions at NIST and MSDN. HumphreyW (talk) 02:44, 29 September 2009 (UTC)[reply]
          • MurmurHash was recently deleted, and it has about as many mentions on Google, so that can't be enough. The precedent is that you need reliable sources to confirm notability, not just tens of thousands of hits on Google. Phil Spectre (talk) 11:17, 29 September 2009 (UTC)[reply]
          • As Phil Spectre says, counting Google hits is not research. Google hit counts are estimates, not actual counts. They have no value. So stop counting hits and start citing sources. That's the way to make an argument for keeping that holds water. Sources! Sources! Sources! Uncle G (talk) 15:12, 1 October 2009 (UTC)[reply]
• Keep as per discussion with original proposer above. Both the Wikipedia page and the original source pages are mentioned and linked in too large a number of significant websites for me to consider the source as not reliable or not notable. HumphreyW (talk) 13:29, 30 September 2009 (UTC)[reply]
  • With all due respect, you're certainly entitled to your own opinion, but not your own facts. The NIST link is not a WP:RS, since all it does it point to the same old Noll site, along with the one for MurmurHash, which was deleted due to a lack of reliable sources. If it doesn't count for that hash, why should it count for this one? The MSDN link is not only a blog, but it contradicts WP:N in that the MS employee states that MS has no intention of ever using FNV, and the comments from the peanut gallery point out FNV's terrible avalanche characteristics. If you want to keep it, you absolutely must prove its notability with reliable sources. Fortunately, this is not a vote. Phil Spectre (talk) 22:59, 30 September 2009 (UTC)[reply]

• This AfD nomination was incomplete (missing step 3). It is listed now. DumbBOT (talk) 13:39, 1 October 2009 (UTC)[reply]

Sorry about that. It's my first AfD and I forgot. Thanks for fixing it. Phil Spectre (talk) 22:58, 1 October 2009 (UTC)[reply]

• Keep There is certainly plenty of discussion of it online with lots of people referring to it as widely used. The only questionable part is the "reliable" requirement of WP:N, but that is largely for verifiability and the large number of medium-quality sources (university lectures, peer-reviewed academic publications using FNV, tech sites, etc.) make this quite verified in my mind. I think it's unreasonable to expect newspaper articles and things about even notable algorithms--I google the extremely notable heapsort and the source quality is not obviously better than for FNV. Gruntler (talk) 07:15, 3 October 2009 (UTC)[reply]

Help me out here. How is FNV any more reliably sourced and notable than MurmurHash? Phil Spectre (talk) 03:29, 4 October 2009 (UTC)[reply]

FNV hash in google scholar [1] gives 105 articles. They are relevant. I see peer-reviewed articles, conference proceedings, lecture notes, etc indicating this algorithm's usage. Murmurhash in google scholar [2] gets three hits. Or compare searching Murmurhash on the .edu domain [3] versus FNV hash on the .edu domain [4], the sources are quite obviously better for FNV especially once you click past the first page. Gruntler (talk) 04:27, 4 October 2009 (UTC)[reply]

As far as I can tell, both hashes have some mentions on Google, but none of these sources have been included in their respective Wikipedia articles. The numbers are ambiguous, with MurmurHash having twice as many .edu references as FNV despite being more recent. Ultimately, if these are reliable sources, they should be added, not talked about here. This way, specific references could be researched and checked for reliability and relevance. Until then, we're only comparing one type of hand-waving against another. Phil Spectre (talk) 19:38, 4 October 2009 (UTC)[reply]

Not on my internet, the linked searches on site:.edu above returned 1580 for FNV and 584 for Murmurhash. (You're searching "FNV" and not "Fowler-Noll-Vo," right?) Beyond numbers, the quality is just obviously better for FNV. The .edu search for murmurhash quickly tails off into file listings and directories and the first several pages (at least) of the search for FNV consists of presentations, papers, conference proceedings, and such. Gruntler (talk) 19:02, 5 October 2009 (UTC)[reply]

And here I thought we were using the same Internet! For what it's worth, I get 160 for MurmurHash, and 225 for FNV. As for quality, neither set is much good. For example, even the first page for FNV contains irrelevant links, such as a page about primes that mentions FNV only incidentally.

Noll's claim to fame is his discovery of large primes, not his use of small ones in poorly-distributed hashes. I would not object to FNV being mentioned on his biography page, but it's just not important enough, or reliably sourced, to have such a large, detailed article. For all the links you mention, none of them are used in the article. Since it's really just a restatement of Noll's page, why not just link directly there and cut us out as middlemen? Phil Spectre (talk) 23:23, 5 October 2009 (UTC)[reply]

Gruntler, I cut the article down to a size that I would be happy with. Do you really think it ought to be bigger? Phil Spectre (talk) 23:28, 5 October 2009 (UTC)[reply]

The issue with the other sources is that they largely aren't *about* FNV hash, per se. They're of the form "we used the FNV hash to do X in our work on generally unrelated topic Y." To me seeing a bunch of these *does* establish notability but it's not something that can easily be incorporated into the article.

Anyway, I'm ok with something resembling the version you put up. Gruntler (talk) 17:13, 6 October 2009 (UTC)[reply]

Although this is my first AfD, I'm fine with it "failing". Arriving at a version of the article that we can all live with is a better result than either deleting it or keeping it in its previous form. I'm wondering if this means we should revive MurmurHash in a shortened form. I'm also wondering how we can end this AfD. Phil Spectre (talk) 01:23, 7 October 2009 (UTC)[reply]

Maybe this can show how to end the AfD: Wikipedia:Non-admin closure HumphreyW (talk) 01:34, 7 October 2009 (UTC)[reply]

Thanks. I think this qualifies, based on how long it's been listed, the fact that I'm withdrawing it, and the lack of any visible support for deletion. I'm comfortable with performing a non-admin closure, by way of cleaning up my own mess. Phil Spectre (talk) 01:45, 7 October 2009 (UTC)[reply]

The above discussion is preserved as an archive of the debate. Please do not modify it. Subsequent comments should be made on the appropriate discussion page (such as the article's talk page or in a deletion review). No further edits should be made to this page.