Wikipedia talk:Essay on the notability of software/Archive 1

Very Rough Draft

 * Inspired by this AfD, which points out that we don't have any sort of metrics to use for software projects. As mentioned by me in that AfD:
 * There has to be a line somewhere between Apache and Joe's File Renaming Script such that projects on one side are included, and projects on the other side are not.
 * This proposal is a first stab at generating some guidelines for the inclusion of software projects. --Stephen Deken 16:15, 29 December 2005 (UTC)

Reasonable
This sounds quite reasonable for a start. I think the main goal should be to filter out (most) self-published shareware and freeware, as well as indie projects that haven't actually released anything yet. There's far too many of both on the web, and while there are some quality (and notable) indie games and tools out there, about 95% of them are utter crap. That includes both "cut-n-paste job" games such as most created in Klik-n-Play or Games Factory, as well as the zillion of near-identical utilities to perform a trivial task such as icon editing. Radiant_ &gt;|&lt; 21:37, 17 January 2006 (UTC)


 * I added a suggestion that programs that have some user base but not are notable enough may be merged with others of the same kind. Of course, I excluded that is used only by its authors and their friends. - Liberatore(T) 17:34, 26 January 2006 (UTC)

Heavily Downloaded - From Where?
The current version of the page (as I write this) lists a standard of 100,000 downloads from, specifically, Download.com or Sourceforge. I don't think that these should be the only "blessed" download sites; in particular I think Freshmeat.net is a better standard than Sourceforge (which is technically a software development site, not a download site). I'm adding freshmeat, but I'd prefer to see discussion of which sites should qualify. 168.12.253.82 15:34, 18 January 2006 (UTC) I have modified the bullet by saying that the number of download can be used to assess the number of users, as the number of downloads do not apply for commercial software and software that is not distributed from freshmeat, etc. I have also removed the Google test, as other guidelines appear not to rely much on it any longer. - Liberatore(T) 10:21, 29 January 2006 (UTC)
 * I'd say that the download sites shouldn't be explicitly stated, but rather described -- something like "100,000 downloads from an independent software download site not run by the creator of the software." Of course, how much do you trust the number of downloads on any given site? What about ballot box stuffing -- authors re-requesting the download to boost the number?  --Stephen Deken 15:47, 18 January 2006 (UTC)
 * I'd take out the download/user bullet completely. 1) It's too easy to spoof download counters (for example, by running a bot against a favorite download site just to drive up the counter).  2) It ignores churn (having 100k users is irrelevant if people try it and dump it, leaving the site/tool/software with only a few users at any one time).  And, 3), it's extremely problematic to verify.  We can not, for example, rely solely on the company's own website or reported usage.  I think the other bullets on the page are probably sufficient for the vast majority of cases.  The exceptions can be handled as, well, exceptions.  Rossami (talk) 14:37, 22 January 2006 (UTC)

Genericization
I think that the "genericization" item is redundant: if the name is so popular as to became generic, surely there will be a couple of publications in major media. Mukadderat 19:07, 18 January 2006 (UTC)
 * Well, it's certainly not harmful, but I can't think of a single piece of software to which this criterion would apply. - ikkyu2  ( talk ) 20:47, 3 March 2006 (UTC)
 * It's completely redundant. I removed it. --Sharcho 17:06, 19 November 2006 (UTC)

Linux distributions
A software package should get bonus points for being in an OS distribution (Linux, FreeBSD, etc.). For the more selective distributions such as Fedora Core or Ubuntu main, just being in the distro should make it notable. Debian is too big for a software package to be included in Wikipedia just because it has a package in Debian; the Debian Popularity Contest is a great resource; we could choose a percentile above which packages are considered notable. &mdash;Quarl (talk) 2006-01-22 19:37Z 


 * Agree. Inclusion in a major mainstream distro is notable. - Just zis Guy, you know? [T]/[C] [[Image:Flag of the United Kingdom.svg|25px|  ]] AfD? 22:36, 25 January 2006 (UTC)

How is the list of distributions determined? The use of the word major appears to make that criteria subjective. lists the top 10 as: Also, the criteria is qualified by "The more packages a distribution includes, the less notability is implied by inclusion in that distribution." It would be useful to provide package count stats or links to stats to help wikipedians see which distro's hold more weight. John Vandenberg 20:19, 15 January 2007 (UTC)
 * 1) Ubuntu
 * 2) Mandriva Linux
 * 3) SUSE LINUX
 * 4) Fedora Core
 * 5) Debian GNU/Linux
 * 6) Knoppix
 * 7) MEPIS Linux
 * 8) Gentoo Linux
 * 9) Slackware Linux
 * 10) Xandros Desktop

User Expectation
I would like to throw out the idea of user expectation as at least something to consider. I don't think anyone expects to find new software on wikipedia, but for something that is very important, like Apache they may expect to find information about the history of the project, for example. If there is a software project that is so small that you can't imagine anyone trying to find information about it on wikipedia it might not belong, it will certainly be difficult to maintain if that is the case. - cohesion&#9733; talk 02:05, 23 January 2006 (UTC)


 * We have extensive Category:Lists of software that, arguably, visitors use to browse software. -- Perfecto 02:19, 23 January 2006 (UTC)


 * Well, if enough people use that in searching for software to install, maybe. I would argue that isn't the purpose of an encyclopedia though, "Wikipedia is not an indiscriminate collection of information" (WP:NOT). But others might feel differently. Personally I would say we shouldn't be a software directory, but if we are going to be one we should include everything. - cohesion&#9733; talk 02:34, 23 January 2006 (UTC)


 * Last I saw Apache was the dominant player in the web server market, so would be included under the "top ten" criteria currently listed. - Just zis Guy, you know? [T]/[C] [[Image:Flag of the United Kingdom.svg|25px| ]] AfD? 22:37, 25 January 2006 (UTC)

IMO bad idea
IMHO this is a bad idea. Articles on your own software comes under a vanity article. Under those critieria, GeneRally (With it's wealth of community contend and community competitions as well as community members) would not get an article. It should be on community size, amount of user content (where applicable) and so-on. Duke toaster 19:38, 25 January 2006 (UTC)
 * Which is a bad idea? GeneRally arguably belongs to WikiProject Computer and video games, not this one.  Software usually has no "user content" so I understand not what you're saying. --Perfecto  00:04, 26 January 2006 (UTC)
 * Computer games are software; these rules are as applicable to game software as they are to any other software. The GeneRally article appears to have a large number of contributors, none of whom appear to be the author of the software.  If you think that people who own a copy of the software or who use the software aren't allowed to contribute to the article, you've misinterpreted the guidelines.  -Ikkyu2 02:12, 1 February 2006 (UTC)
 * Nevertheless, I think GeneRally is under WikiProject Computer and video games. If video game articles requires inclusion discussion, I most certainly will look for it there, not here.  I still do not see the connection (as what you're saying) between the documented count of the software's users and the count of the contributors of its article here in Wikipedia.  AFAIK, having a dozen contributors does not guarantee that a topic stays here.  And, again, which is a bad idea?? --Perfecto  04:49, 1 February 2006 (UTC)

Media coverage
If there is a Que book or a "Dummies" book that's a fair indication of a substantial market, as written that's not taken into account. - Just zis Guy, you know? [T]/[C] AfD? 22:38, 25 January 2006 (UTC)

Historical interest
Bear in mind that software is beginning to have a history. An article like Sabotage (computer game) may well be appropriate for historical interest or archaic-game aficionados; it will never meet a download criterion, because it was distributed on floppy disc.

I think notability is probably less important for software for obsolete platforms, for the same reasons it's less of an issue for long-dead people, but it's well to consider all the angles when formulating a guideline. -Ikkyu2 01:45, 26 January 2006 (UTC)


 * I disagree that obsoleteness itself makes software notable. I have changed your text:
 * The software is obsolete (or written for an obsolete platform), no longer for sale, and of general historical interest; for example, old games still playable under emulation such as Sabotage (computer game).
 * to:
 * The software is of general historical interest -- being obsolete does not in itself make the software non-notable, as it may have been notable in the past.
 * What do you think of that version? &mdash;Quarl (talk) 2006-01-28 06:56Z 


 * I prefer the new version over the old one, but I still object on the last clause of point 6 ("as it may have been notable in the past"): as it is stated, it gives the impression that software is notable only if it has been notable at the time it was used. This is obviously not the case: the software may have been non-notable at the time but is notable now because it affected the following development. For example, we have an article on the B programming language, which is notable as the predecessor of C.
 * Actually, I think that this point should not be part of the critera itself, but rather a note, because it of its "negative" form (it says that obsolescence is not relevant to inclusion). - Liberatore(T) 13:08, 28 January 2006 (UTC)
 * Thanks for your consideration of my points, folks; I support the new revision entirely. -Ikkyu2 00:05, 29 January 2006 (UTC)

Comparison with WP:BIO
WP:BIO seems to imply a notability threshold at 5000 copies for a book. As such, the 100,000 users proposed in this guideline seems too high a number. I would propose to lower that at 5000 as well. As for the number of downloads, that's another story. - Liberatore(T) 17:21, 26 January 2006 (UTC)
 * Change done. The amount of 5000 was also the threshold for the size of forums in WP:WEB before the latest change . - Liberatore(T) 17:36, 26 January 2006 (UTC)


 * I consider the change to be a mistake. First, the 5000 threshold on WP:BIO is controversial and now considered by many to be too low even in that setting.  (A long-simmering discussion is trying to decide if that number should be raised (and if so, to what) or simply deprecated.)  Furthermore, production, distribution, sale and readership of a book are all more complex and expensive than the distribution of software.  5000 people buying a physical book represents a significant total investment of time and effort.  I can get 5000 downloads of software simply by loading it myself a bunch of times.  (Yes, some of the download sites do try to compensate for this abuse but it's still fairly easy to game their systems.)  It also ignores all the users who download, try for 5 min and then delete.  There is no measure I know of to separate out those people from the real users.  I think that this measure will tend to default to download count (and in fact, the notes further down explicitly say that we should assume that download counts can be a proxy for usership).  I consider download counts to be a very poor measure of actual usership.  I would prefer to remove the criteria altogether because I don't think it is a particularly good metric for the question we're trying to answer.  But if I can't convince you to remove the criteria, the thresholds should certainly be set much, much higher.  Rossami (talk) 23:19, 10 March 2006 (UTC)

A significant amount of software is moving towards web delivery - eg Web 2.0, Web applications and Java applets. I think this whole 5,000 downloads idea is a bit narrow. Here are four illustrated examples from software served by my company:
 * FORscene is a software service for video post production. Companies have accounts - here is a guest account. The software is embedded in a web page, and loads (usually from the web cache) each time the customer starts using it. Each TV programme made typically has hundreds of hours of source material (usually less than 1,000 hours) and can have twenty or so simultaneous users of their account. So each account can end up with thousands of hours of use - but no conventional downloads or installation is required.
 * At the other end of the spectrum is this FORscene customer. They have published a video using FORscene on their web site. We don't usually reveal usage figures, but you can see from their page count that 39,000 people have been there. Almost all (I have the figures from the web logs) of these people have downloaded (and run) the FORscene software which automatically plays back the video when they open the web page. So hey - one small customer publishes one video on FORscene, and we have easily passed the 5,000 unique users/downloads mark.
 * And at another end of the spectrum is the forbidden website itself www.forbidden.co.uk, which had around 1,000,000 hits per month for a couple of years now. A lot of these run the video on our home page (so we're already notable software just from our own website!), but most of the hits are from this demo, which gives one hit every 10 seconds of watching. This video is very popular. Some people watch it for a few seconds, others for hours on end trying to determine for themselves whether it is really live video!
 * FORscene also has a mobile component. This can be downloaded from the website, but most (I'm guessing here) users get it via Bluetooth from another mobile handset. There is no record of the number of people who send the application to each other in this viral way.

Some users use it a lot, lots of people use it once (and probably don't know they are using it, though this shouldn't matter for notability), lots of people use it deliberately but casually, and an unknown number of users run it on their mobile phones.

Of course, I think the software is notable, but the consensus in some areas is that it can't be notable because it doesn't have an article in Wikipedia! But counting downloads will miss all four cases, and not allow for web delivery or wireless delivery. The volumes illustrated here shows how significant web delivery is for software use, and not just from people like Google, which also has no downloads.

Downloads are significant BECAUSE so many people can do it. Web delivered software is significant BECAUSE so many people can do it, often without even realising it. You can't then adjust for this by saying (I paraphrase/exaggerate to make the point clearer) Google is easy to use, so we won't count their huge number of hits. Forbidden software is notable because it has been used millions of times, and it has been used milions of times because it is easy to use. In other words, it is notable because it is easy to use - approximately the opposite conclusion to Rossami. Stephen B Streater 11:44, 11 March 2006 (UTC)


 * As I said above, I would support a removal of the number of users from the guideline. First, it is difficult to evaluate, and Stephen's post shows also that in some cases it may not make sense at all. Second, it is arbitrary (why 5,000 and not 4,500?). However, in this case we should also clarify (as a note) that a similar threshold is intentionally missing, and that a program may very well be notable with 100 users while another is non-notable with 100,000 users (or downloads). - Liberatore(T) 14:48, 11 March 2006 (UTC)

Programming-Languages should be more NOTABLE
I propose we change Notability (software) to say that programming languages should be considered more NOTABLE (than the average software) simply because they're a programming-language. Those rules should not be enforced as strictly with programming-languages. Without programming-languages, no software would exist. BenRayfield 07:49, 30 January 2006 (UTC)
 * Nonsense, especially with recently-created languages that haven't become notable in any way. It's just software (albeit at an interpretive level between user and machine) like anything else.  RasputinAXP talk contribs 12:47, 30 January 2006 (UTC)


 * Agreed. There are plenty of programming languages out there that aren't notable, except perhaps as an illustration of some general concept. Basically all of the notable ones have (or once had) a large number of users, and would qualify on that point alone. Languages without a large number of users aren't notable just for being programming languages - these days anyone can write a language parser if they want to. 168.12.253.82 14:37, 30 January 2006 (UTC)


 * I agree that programming languages aren't inherantly notable. (BenRayfield, I suspect you've never written one.  You should try it, it's fun!  But almost as easy as writing software in general, so we have tons of them.)  &mdash;Quarl (talk) 2006-02-01 11:43Z 


 * I agree with BenRayfield that notability with software languages does differ from typical software and that programming languages should have other criteria to determine notability than just adoptance. There are a number of languages that may not have achieved widespread use, but are notable for the languages they inspired or the language paradigms or concepts they introduced.  I frankly have learned some rather interesting things from reading about programming languages listed here that probably would not cross above the threshold for notability based on some standard of adoptance. – Doug Bell talk&bull;contrib 00:34, 3 February 2006 (UTC)
 * Actually, I believe BenRayfield had an [|ulterior motive] for saying they should be more notable. If a lesser known language leads to a gerate known one, then by virtue of being the predecessor to a widely used language there'll be an article...I mean, who uses B anymore? But go to C++ and C and there it is.  RasputinAXP talk contribs 01:17, 3 February 2006 (UTC)

I think the question is the same as the section below: giving a fixed number of users only makes sense for software of general use (e.g., a text editor). For all others, we have the 8 other bullets of the guideline. I think they reasonably cover all other possibilities for a software to be notable. I have added a note about this. Another possibility is, of course, to delete the criteria about the number of users, which I also would support. - Liberatore(T) 12:18, 3 February 2006 (UTC)

Notable and popular
I have made a small change in the last bullet to reflect the fact that "notable" and "popular" are not necessarily the same. My point is that this guideline covers all software, not just that used on the average computer by the average user. A very specialized software may be notable even it is used by few people. Should this be explicitely remarked? - Liberatore(T) 10:55, 1 February 2006 (UTC)
 * I suspect the situation you're alluding to is something like astrophysicists all use software X, so it's notable, even though the total number of astrophysicists is small. I am not sure how one would phrase this point unambiguously such that someone can't say "I and my 12 friends all use the instant messaging software I wrote, so it's notable".  &mdash;Quarl (talk) 2006-02-01 11:48Z 
 * Yes, the example of astrophysicists is a good one. More generally, notability of programs used by professionals (for example, SPICE) should not be evaluated using the same threshold on the number of users of programs of general use (for example, an icon editor). I am not sure how to word this either.
 * Incidentally, is creating an article about software you have written yourself discouraged? (in the same line of WP:AUTO) - Liberatore(T) 12:11, 1 February 2006 (UTC)
 * Yes, I think the standard vanity/autobiography guidelines apply. It could be mentioned as a redundant parenthetical.  &mdash;Quarl (talk) 2006-02-01 12:30Z 
 * OK. I wrote a note saying that 5000 refers to software of general use. I think we have no way of making a bound for specialized software, so I just said that the other bullets have to be used instead. - Liberatore(T) 12:21, 3 February 2006 (UTC)
 * We could include market share as a concept here. So, if loads of astronomers use a piece of software, it's probably notable. Stephen B Streater 12:17, 10 March 2006 (UTC)
 * I'm also concerned with software which is notable because it transforms an industry. My personal interest is with FORscene, which has had loads of press articles in the industry press (and national press for that matter), saying (essentially) that this will transform video editing. However, as a relatively new and specialised product, only released in Europe and Japan (ie not the US), you've probably never heard of it. People who find out about (including journalists and big customers) think it's notable though. How would services like this be captured? Stephen B Streater 12:17, 10 March 2006 (UTC)
 * While I'm here, it's also worth mentioning that FORscene has significant technology, inlcuding specialised codecs for video editing and a Java/web interface. How are software products which are notable for their technology content to be included? Stephen B Streater 12:20, 10 March 2006 (UTC)

Latest addition Bullet #9
"The software has some important feature not present in other, more notable, software"

I don't care for this. It seems like a loophole to allow promotion of non-notable ware. "FooEdit allows you to edit the spellcheck dictionary BY HAND -- an important feature not found in Microsoft Word!! Only $19.95!

Comments? -Ikkyu2 16:12, 1 February 2006 (UTC)
 * Hurrrrk. No way. That's a loophole and a half. "My software doesn't have bugs, it develops random features."  RasputinAXP talk contribs 18:44, 1 February 2006 (UTC)
 * I've removed that item. &mdash;Quarl (talk) 2006-02-03 13:36Z 

Bullet 4
Bullet 4 reads: Has a forum or mailing list with a consistent number of members. I'm not exactly sure what is meant by "consistent number of members". 'Always 5 members' is consistent, while '300 members and growing 200% per year' is inconsistent. What is the real metric here? -- ShinmaWa(talk) 02:23, 5 February 2006 (UTC)


 * I think the word "consistent" was intended to refer to bullet 3. I've changed it to Has a forum or mailing list with a significant (5000) number of members.  &mdash;Quarl (talk) 2006-02-05 10:16Z 


 * I also intended it that way, and agree with changing the word to "significant". As for the number of members, it is perhaps a little strange to require 5000 members while 5000 users are also required. Considering that I use lot of software without being a member to any forum, I would propose to lower this number. Leaving it out would also be good, but then we have the obvious problem of someone saying that 10 members are a significant number.  - Liberatore(T) 11:16, 5 February 2006 (UTC)


 * I agree. The Apache HTTP server (arguably the most popular web server in the world) has only a few hundred people on their mailing list.  I think it would be VERY hard to find any software product mailing list with 5000 members. -- ShinmaWa(talk) 20:41, 5 February 2006 (UTC)
 * What about announcement lists? &mdash;Quarl (talk) 2006-02-05 22:27Z 


 * I believe some software is notable, but if any of them have forums I'm not a member of any of those. (Except Firefox, I admit I joined that.)  Perhaps I don't like giving my email address away as these forums ask me to register. --Perfecto  23:11, 5 February 2006 (UTC)


 * Not sure what your point is. I'm on a large number of software mailing lists.  &mdash;Quarl (talk) 2006-02-06 05:09Z 


 * As far as I known, there is no way, in general, for evaluating the number of people subscribed to a mailing list, other than asking the maintaier. I also have no idea of how much the number of subscriber compares with the number of users, but I would guess that the subscriber/user ratio depends on things that are not related to notability, such as whether the software is free or stable (I think free software and software that is not stable has larger mailing lists). - Liberatore(T) 09:20, 6 February 2006 (UTC)
 * By that I do not mean that we cannot introduce the size of a mailing list as a possible criteria; it's just that I think we cannot assume subscribers/users=1, and that we can accept the number of subscribers only if there is some way to reliably evalute it. - Liberatore(T) 09:22, 6 February 2006 (UTC)


 * I agree; the whole time I was assuming that this bullet could only be used when the number of subscribers was easily verifiable, as the subscriber list is public on some mailing lists. &mdash;Quarl (talk) 2006-02-06 11:26Z 

From this discussion I think the bullet, as written, is not very useful. How about we remove this bullet and change bullets 3,4 to something like:
 * Has more than 5,000 users
 * If the software is distributed via sourceforge, download.com, or freshmeat, the number of downloads might be used to estimate the number of users
 * If the software has a forum or mailing list, the number of subscribers might be used to estimate the number of users
 * &mdash;Quarl (talk) 2006-02-06 11:29Z 


 * The two sub-bullets seems ok to me. Maybe they are better suited for the notes than for the criteria, since they actually are not criteria but ways for assessing whether a criteria is satisfied? - Liberatore(T) 12:48, 6 February 2006 (UTC)
 * Okay. &mdash;Quarl (talk) 2006-02-06 13:09Z 

The problem with market share and download statistics
Much of the debate above involves the metric called market share.
 * For software with major media (Oracle RDBMS, Microsoft SQL Server), "market share" is easy to verify(May 2005).
 * For freely-downloadable software, we can choose the largest free-download-sites (call them "Amazons"), and use the download stats (weekly, monthly or ever since) there to rank the software titles amongst similar ones (aka, Sort by bestselling) or amongst all software there (aka, Amazon's sales rank).

If you read Market share closely, or even that May-2005 article, the metric refers not to "units sold (or downloaded) per period" or "units sold ever since". Oracle's 33.7% market share means that for every 1000 legal entities (individuals or corporations) that are Oracle's potential customers (qualified prospects), Oracle serves 337. This has no bearing whether Oracle had 10 or 10 million units sold, or 10 or 10 million downloads in the month of May 2005.


 * Note that download stats in free-download-sites are affected by a snowball effect.
 * Note that, unless you have one download source, like XP Service Packs, stats are hard to compile.

So let's not. I propose we stop interpreting download statistics as a metric of market share or of notability. Let reliable sources do it for us. If a PC Magazine article features ToDoList 5.0, saying it has 5000 downloads, 5000 users, or 5000 forum members then we call ToDoList notable by a nontrivial mention in a reliable source. --Perfecto 00:08, 6 February 2006 (UTC)


 * Thanks for the market share links. I'm not sure yet about your PC Magazine idea because there may be lots of software I would consider notable, but hasn't been featured in PC Magazine - I would think that being featured in a popular computer enthusiast magazine would automatically make it notable without having to look at the download count in the magazine article (and do they actually report download counts?).   Would it be possible to require 1. a certain minimum market share, and 2. a minimum market segment size?  After all, if someone defined the market segment for their instant-messaging software product as "people using web forum XYZ", they could easily claim 100% market share.  &mdash;Quarl (talk) 2006-02-06 05:17Z 
 * Market share? As I propose, market share and market segment size figures require reliable sources and cannot come from our interpretations (or the submitter's interpretations) of download or usage stats somewhere.  Now if you want the market share figure both verifiable AND above a specified minimum on a nontrivial market segment size..., well that's even better.  For me, if a 1994 Chicago Tribune article says that DiskCopyFast has 3% market share, it's enough for me. --Perfecto  06:22, 6 February 2006 (UTC)
 * Clesh has been included on .net magazine this month, and with FORscene (the professional version of Clesh) has dozens of independent press articles in multiple countries and is widely used. However, the AfD people still think it is non-notable. It is notable under both the old and proposed rules, but still the AfD consensus was that it was non-notable software. So I think these guidelines are either not understood by the users at large, or are not capturing a true consensus amongst Wikipedians. Stephen B Streater 18:55, 10 March 2006 (UTC)

taking anti-vanity a step further
''Creating an article about software you have personally developed is discouraged but not forbidden. It is indeed easy for an author to overestimate the notability of their work. If such work is notable, someone else will eventually start an article about it.''


 * Let's take it a step further. Articles whose only contributions had been from the software developers themselves should always be marked ASAP as stubs, and possibly a template indicating the contribution has been solely self-contributions.  Or something like that.


 * I would be happy for such articles to be allowed, but marked as such. Articles could say how many people have contributed to them as well, for example. My article about FORscene was removed primarily because I'm involved in FORscene, even though it meets both current and proposed notability criteria, depriving the Wikipedia community of any article on the subject. An appropriately flagged article would be better than no article. Stephen B Streater 18:47, 10 March 2006 (UTC)

What about software pages ?
It will work like advertisement for commercial products - against the wikipedia policy.

My software have much more than 5000 users and count of downloads at download.com more than 100000, marked as popular etc. (software is really specialized ). But using such values is bad idea. Maybe community voting will be better - like deletion log now. Is it good software or not. Discussion page is good place for this voting and if software receives a high count of negative values - it will be removed. But what about program info page at wikipedia? Article without external links will be good reason to not create just spam articles, but the what about the importance of this article? who will need it?

Wikipedia visitors does not generate any sales, software authors use wikipedia pages as keyword catchers for google etc. like Winpopup_LAN_Messenger, Softros_LAN_Messenger , ShixxNOTE , this is the reason for spaming wikipedia. The other reason - there is no rel="nofollow" tag in the external links. --Crea7or 18:42, 20 February 2006 (UTC)


 * Expereince shows that objective measures of notability are much easier to work with than any criteria requiring extensive debate and voting on each page. Also the criteria should be based on notability, not quality.  You're right about the reason for spamming and I agree with you that external links should have nofollow &mdash; however, Wikipedians in the past have decided against it, see Nofollow. —Quarl (talk) 2006-02-20 18:57Z 


 * Yes, It's better than nothing. But how to moderate pure advertising? Nofollow problem is really big, I think it will appears more and more, because google start to place the wikipedia results at the top of search.

--Crea7or 20:51, 20 February 2006 (UTC)


 * Feel free to bring it up for another big vote at WP:VP; it's been a year since the nofollow vote, and the vote was like 87-55 in favor of not utilizing nofollow - hardly a consensus. I think another Wikipedia-wide vote on this is warranted.  But the idea of using 3rd parties to establish notability is pretty established on Wikipedia, and it has worked pretty well in brushing aside debate - see WP:MUSIC for a nice way to routinely delete the dozens of un-notable musicians and bands that get articles posted every month.  Oh, and bad software certainly deserves its own article if it's notable.  (insert your favorite joke here about Microsoft, Apple, IBM, Lotus, the United States Government, etc.)  208.57.241.20 23:46, 2 March 2006 (UTC)


 * I'm assuming talk pages aren't followed, as people don't delete illustrative links. I find some links which have been removed as spam are useful, and others which remain are not. A solution for links in articles is to get the people who try them out to judge whether they were worthwhile with a little box (8/10 click throughs thought this was worthwhile). Automatically delete from article when it drops below some threshold. Similarly, we could have a "considered for inclusion" sub-page for each article (with no follow) which get promoted if people like them. Stephen B Streater 09:33, 12 March 2006 (UTC)

Long version
I propose a modified benchmark for notability. A piece of software is notable if it meets any of the following criteria: I would expect any software truly the "best" to have measurable spin-offs. In this case, the international fan mail helped me win a High Court case to receive unpaid royalities from a summer job. I also got to know Sophie Wilson, who joined the Eidos board when I floated Eidos in 1990. Eidos got access to all the prototype ARM chips we could use during the 1990s and were the one of the first companies to get the StrongARM prototypes. And I got to demonstrate and sell Eidos technology to Larry Ellison, (which we demonstrated at the launch of Oracle's Network Computer) and our deal with Oracle led directly to the acquisition of Tomb Raider. The rest is history, as they say.
 * It is the first (but not only) one in its class. More than one group must have an implementation. This is a slightly lagging indicator (as befits an encyclopaedia), which would pick up the first spreadsheet, but exclude the only wibblegrommet, which never caught on. So Forbidden's video streaming to mobile phones in spring 2002 was not notable at the time, because it was the only streaming to mobile phones available. But now other people can do it too, it is notable retrospectively.
 * It is measurably the best in its class (usually for technical reasons). In the late 1980s, when the Mandelbrot set was all the rage, generating speed was the big issue. I wrote a Mandelbrot set program which was 1,000 times faster than naive IBM PC implementations being published at the time. It was faster for three technical reasons:
 * 1) it ran on a computer using the new RISC CPU, the ARM, which was very fast for its time;
 * 2) it was implemented in machine code using highly advanced (for the time) and ARM specific optimizations;
 * 3) it used mathematical information about the fractal not generally exploited at the time (I was a maths student).
 * It has a high volume of users (direct and indirect). By indirect, I mean to allow, for example, a compiler used by a few people, but whose code is used by millions, or a piece of FEM software used to design 20m cars. Often the direct use is not for the most innovative pieces of software, but subsequent generations made by copycats may have better marketing and have learned from experience - Microsoft Windows is the cannonical example.
 * It has significant independent press coverage - this will pick up any reasons not in the more specific criteria above.

This is simpler than the current list. It excludes non-notable software from a big company (which may have dozens of such products), and removes the need for arbitrary numbers of users which strongly depend on the market. How many people buy nuclear attack detection software? But it better work because one false alarm and we're toast! A lot of the current list will lead to press coverage, so needn't be written separately.

Summary
In summary (though these could all be relaxed to count the top few):
 * First in class
 * Best in class
 * Most popular in class
 * Multiple independent press articles

What do people think? Stephen B Streater 09:27, 12 March 2006 (UTC)

As we are working on guidelines and not prescriptive rules, a long and complex set of rules will be less effective than a handful of simple ideas to guide the debate on any individual piece of software. Stephen B Streater 09:35, 13 March 2006 (UTC)


 * I think that proposing modifications to guidelines which directly affect disputes in which you are actively engaged without declaring your interest is something you should avoid. Not that I am necessarily opposed to an influence guideline. Just zis Guy you know? 09:40, 13 March 2006 (UTC)
 * I don't think there is a dispute ongoing. I've declared my interest all over the place. If you think otherwise then please be more specific (perhaps on my talk page if it is not relevant here). Stephen B Streater 12:17, 13 March 2006 (UTC)
 * No, I had missed the comment a couple of discussions up, and I apologise, but I am still uneasy about you, a publisher of software the article for which was deleted recently for failing these guidelines, arguing for a change to these guidelines which would allow your product to have an article again. I guess you can see why. Just zis Guy you know? 18:31, 13 March 2006 (UTC)
 * Yes - I am not opposed to Wikipedia ethos either. But it is worth pointing out again that under existing guidelines, FORscene is notable. See here under "Criteria for products and services": "The product or service has been the subject of multiple non-trivial published works whose source is independent of the company itself... This criterion includes published works in all forms, such as newspaper articles, books, television documentaries, and published reports by consumer watchdog organizations." This is so true of FORscene its hard to know where to start, including multiple reviews in industry press and websites, long reviews in multiple languages I don't even understand, and multiple articles in the national press. Even since the AfD, FORscene has has another independent website review (unsolicited I might add), and been included on the magazine disc of .net magazine, and had a whole sector created for it this web applications website. TV Documentaries have been MADE on FORscene! FORscene is notable under ALL the proposals given on this page. This is why I thing the article failed because of VSCA. You've seen my style ;-) And why I think it's OK to be involved in this discussion. Stephen B Streater 18:55, 13 March 2006 (UTC)
 * Woo-hoo! Obsessive capitalsiation - yup, WP:VSCA alright :-P Just zis Guy you know? 00:17, 14 March 2006 (UTC)
 * One of these days I'll include a video comment :-) (Available for editing in true Wikipedia style, of course, here). Stephen B Streater 08:40, 14 March 2006 (UTC)
 * Obviously this doesn't really work without uploads enabled and accessible to everyone. But that's only a matter of a few years away. Stephen B Streater 09:13, 14 March 2006 (UTC)

Back to the topic at hand: most people who vote in notability discussions don't read the guidelines. And if they do, don't follow them. Rather, they use their own metrics and hunches. Why is this? Perhaps it is that the guidelines are too long and complex, or too specific and prescriptive to be deemed relevant. So This is why I propose short set of guidelines to inform the debate rather than dictate it:


 * I disagree with the first three bullets. They completely depend on how a "class" is defined. As a trivial example: a non-notable text editor may become notable if someone decides that that "text editors with automatic lorem ipsum generation" is a class on its own. A more realistic example: "this is the first compiler for a language I have invented"; this is not a proof of notability: the language may not be notable. We had a similar problem with a bullet that said that software is notable if "it has a feature not present in other software". This bullet was removed because it depends on what is a "feature". - Liberatore(T) 12:21, 14 March 2006 (UTC)


 * The updated version (below) allows for the exclusion of obscure things eg "The only text editor with feature X" by insisting that it is the first of more than one. This doesn't answer your main point as to where the classes would begin and end though - this would have to be up to the discussion. In fact, all the first three points would be subjective - but that's because I can't see how a few guidelines written now can guarantee to give an objective answer which will work for all possible types of software. And if discretion is to be used, why not admit this up front. It is notable if people think it's notable [amend] - but they need some guidelines as to what to think about. Stephen B Streater 13:57, 14 March 2006 (UTC)


 * Indeed, these are guidelines and not policy. More specifically, these are proposed guideline, not even official ones. However, once an agreement over them is formed, they will likely be followed in most cases, especially because doing that makes articles to be treated equally. My objection is that defining notability in terms of classes is just too vague and open to gaming (you've been fair when your article was nominated for deletion, but some other people aren't) for being part of an official giudeline. Regarding the updated version of the proposed bullets, I'd remark that software being alone in a class may be a sign of notability of the software and may be a sign that the class is chosen impropertly. In my opinion, this shows that these bullets are just too imprecise, even for giving people a rough idea of what is notable. - Liberatore(T) 15:32, 14 March 2006 (UTC)

Updated Summary
Software should be deemed notable if it:
 * is the first in its class (but not only in class)
 * is the best in class (in some measurable and important way)
 * is one of the most popular in class
 * has multiple independent press articles

Stephen B Streater 09:26, 14 March 2006 (UTC)
 * I think the first reason everyone wants to delete the FORscene article is because it's adcopy and written by someone out to promote the product rather than written by a neutral, experienced, wikipedian. You may think you had written the article neutrally but there were  many telltale signs.  If you stick around for a while and write lots of articles on unrelated subjects then you'll know how to write a good non-adcopy article and people will be more likely to trust you.  However, it is almonst always a bad idea to write about yourself or your own work.  If it is notable then someone else will eventually write about it.  —Quarl (talk) 2006-03-14 10:45Z 
 * Well, it was my first try at Wikipedia. See WP:BITE. And the tag line says "Anyone can edit", "Remember, our motto — and our invitation to the newcomer — is be bold". So what did people expect from an enthusiastic newcomer? Stephen B Streater 10:58, 14 March 2006 (UTC)
 * That is a good point. The most common reason pages are deleted is in fact that a biased person wrote about a non-notable subject (in Wikipedia jargon, vanity).  I commend your civil and rational reaction -- some people lose their cool rather than work within the system.  I suggest getting more general Wikipedia experience before coming back to articles about yourself or your products or notability policies.  Bringing up FARscene again too soon would also likely evict an "it was just deleted" knee-jerk reaction. —Quarl (talk) 2006-03-14 11:22Z 
 * Absolutely. There is no deadline to meet on Wikipedia, let the dust settle, contribute to other articles on your area of expertise, continue as a calm and well-informed editor and before you know where you are you will be contributing to an article on FORscene which somebody else started - and then you will know you have arrived :-) For what it's worth I am pretty much persuaded of the notability of the product, but I'm not going to cause trouble by re-creating it right now.  A few more months should see some further coverage in the press and so on.  Time can only help your case here.  Just zis Guy you know? 11:44, 14 March 2006 (UTC)
 * Reading the FORscene article and its discussion on AfD, it's clear that its deletion has little to do with this guideline (only the note about "software you have developed" matters, but this note also says that doing that is not forbidden). I agree with Guy that this software indeed is notable, and that someone else should recreate it sooner or later. - Liberatore(T) 12:42, 14 March 2006 (UTC)
 * This advice is pretty consistent, which is good. Life will run its course and what will happen will happen. Stephen B Streater 19:09, 14 March 2006 (UTC)
 * Back to the specific proposal. Stephen's broad guidelines seem reasonable to me with one change.  The first three bullets should be "or".  Something could be notable for being first or best or most popular.  The fourth bullet must be an "and".  It is not sufficient for the author to claim that something is first.  We must always have verifiable and reliable sources to build the article on.  Those sources must also be independent and non-trivial.  That means excluding press releases, sales literature and any article or website that merely parrots the self-published copy.  (I suppose it's theoretically possible that something could be notable solely on the basis of bullet 4 but no examples jump to mind.)  Rossami (talk) 15:04, 14 March 2006 (UTC)
 * The bullets in the guideline have always been in "or", actually, so I took it for granted for the ones proposed here. Anyway, I disagree. Apart from the fact that a class can be defined in various ways, there is also a point regarding being the "best" or "most popular" bullets. If there is a reliable source considering a software when looking for the "best/most popular software of a class", that can be considered as an independent review of the software, which makes it notable. - Liberatore(T) 15:43, 14 March 2006 (UTC)
 * Incidentally, is the proposal to dump the whole of the current proposal and to replace it with the new one or to add the new bullets in or with the current ones? In the first case, we should also discuss the removal of the current bullets. - Liberatore(T) 15:50, 14 March 2006 (UTC)
 * I was planning "ors" between the points, with the press as a catch-all in case software is notable for a reason not in the list.Stephen B Streater 19:06, 14 March 2006 (UTC)
 * The (slightly bold) proposal is to acknowledge the essence and ideas learned from the current bullets, but rather than just make everything longer and longer, replace them with something shorter and less prescriptive. It seems a bit one section per person, but no simple overarching concept. So the discussion about how precise we want to be is highly relevant. The independent press coverage seems to be the most verifiable information we can go on, but although it almost certainly covers "Most popular", it doesn't necessarily cover the more interesting "First" and "Best". The ARM chip was notable soon after a few lucky techies had access to it in 1986. Now it is in every mobile phone and has outsold Intel cores. Perhaps this would have come under "Best". OTOH, The Psion organiser series 1 was the first PDA. Not very good, but it's a big industry now. So this would come under "First". Stephen B Streater 19:06, 14 March 2006 (UTC)
 * I agree with Paolo that we have traditionally viewed all these bullets as "or" bullets. And I think that in some cases, it's worked against us.  This has been especially true in claims made about WP:BIO and WP:CORP.  New users claimed that their pet article met one of the inclusion criteria but they selectively ignored policies like WP:V.  When we explicitly included the discussion about multiple independent sources in WP:CORP and excluded all press-releases and self-generated copy, many of the problems seemed to go away.  That would seem to be appropriate here as well.  The ARM chip may theoretically have become notable in 1986 but it would have been inappropriate to attempt to write an encyclopedia article about it until there was solid information saying so from independent, verifiable sources.  An encyclopedia is, by definition, a tertiary source. I share Paolo's concerns about the definition of a "class" but hope that we can wordsmith that to an acceptable rule.  But frankly, if we explicitly require independent coverage of the claim, I'm less worried about it.  It will no longer be possible for the author to self-define their market down until they are the "first in class".  The definition of the class will have to have been made by the multiple independent sources.  Rossami (talk) 22:21, 14 March 2006 (UTC)
 * Yes. I was going to date the ARM from 1986, when a group of us outsiders first saw it in action, but I added "shortly after" because by then the computing magazines had picked up on it and tested it against pre-existing solutions - independent and also reputable sources again. I think that it is necessary that any classes should be defined by people not involved with the software for all the reasons stated, but that if independent and reputable sources have defined these classes that may be sufficient. Another reason why first in class cannot be only in class - though I suppose if it is extraordinary enough, people could make a class with only one entry ;-) Stephen B Streater 22:39, 14 March 2006 (UTC)

Quality of guidelines
I wanted to mention something here that I mentioned a while ago on Wikipedia talk:Notability (academics). A notability guideline is only as good as its effectiveness in AfD debates. The guideline needs to (1) have strong consensus behind it, (2) be clear and informative, and (3) be unquestionably reasonable.

I see problems with the current guidelines with respect to my points 2 and 3, particularly when it comes to the 5000-user criteria or 5000-person mailing list or forum. First, it's not clear why having 5000 users makes a piece of software notable. Wouldn't a 5000-user hardware driver be totally non-notable? Whereas, a 5000-user game is borderline notable, and a 5000-user business firewall is pretty damn notable (because users probably equals businesses, and that's a lot of market share). And a 5000-member mailing list or forum? It's a slippery slope, and the way that this implies notability is not obvious. The guideline should attempt to address different types of software differently, if only to be informative.

Side points: I don't think a 5000-person forum is notable in itself, and even mention of one on wikipedia is sort of dubious... but this criteria suggests that such a forum should be mentioned in software articles. Furthermore, I think the main point about the importance of WP:V here is being lost: if there's something unique to say about a piece of software and it can be backed up with reliable sources, this makes for notability. Also, the guideline should probably carry a caveat about reviews: even totally minor piece of software can be reviewed by external sources, so a review doesn't imply notability. However, multiple different independent reviews or reviews in publications that review relatively few products are stronger evidence. Mangojuice 22:04, 29 March 2006 (UTC)


 * I agree the current draft guidelines are a bit lacking. In the section above (Influence) I was trying to capture in a few simple ideas what makes software notable. This may have nothing to do with number of users, and when it does, threshold of number of users depends on the type of software. Do you have any opinion on the four points under Summary above? Stephen B Streater 06:58, 30 March 2006 (UTC)


 * The point about the 5000 users was already raised (see a section above). I proposed to leave it out altogether, which you (MJ) seem to agree. What changes do you suggest regarding the reviews? - Liberatore(T) 07:21, 30 March 2006 (UTC)


 * (To SS): I think those points are good, except for the caveat about independent coverage I mentioned. However, the info must be verifiable and independent: IE, an external reliable source must say that this piece of software is the first of its type, or whatever.  The article creator can't be the one to say that, or it's original research.


 * (To Lib): Yes, we should leave the 5000 number out altogether: it's impossible to verify, anyway.  Frankly, I think we might want to scrap the idea of going with notability here, and just embrace verifiability.  If someone can write an article on a software product that goes beyond its basic function, version history, and who produces it (and/or distributes it), and the information is verifiable from reliable (external and reputable) sources, it's good enough for me.  Mangojuice 13:53, 30 March 2006 (UTC)


 * Interesting points. I was using notability as a partial proxy for verifiability. Unlike an article on, say, Beethoven, software is written by people who may edit Wikipedia. This means that the software must be notable enough that other editors also know about it and can correct any bias which creeps in. Having a large user base means that the articles are more likely to be accurate. But being widely written about (by, for example, being the first) also means that there will be more independent coverage to correct inaccuracies. So verifiability is the key, not notability. Claims about insignificant software will automatically not be verifiable. Stephen B Streater 16:56, 30 March 2006 (UTC)


 * Is this a motion to drop this guideline altogether? When this guideline was initially created, I also didn't believe so much that it was really needed; however:
 * this page is currently linked from 95 AfD debates (mostly closed, of course), which seems to indicate that some people are actually using this as a reference;
 * deleting it would mean that people may decide to go for notability anyway in AfD, using their own standards; as a result, some articles may be deleted even if they do meet these criteria; this can happen anyway, unfortunately, but having a guideline reduces the probablity of this happening;
 * the verifiability and reliable sources have (often) been intended on AfD as necessary but not sufficient conditions for articles to be kept
 * Apart from the number of users (which we apparently all agree to remove), I'd say that this guideline requires little more than WP:V and WP:RS while giving details about them in the specific case of software.
 * I have removed the mention to the number of users, but also added a not about why the number of users is not there. Comments? - Liberatore(T) 18:27, 30 March 2006 (UTC)
 * I like the changes you have made so far. Notability is important in some form, otherwise every jotting of code will be in. My feeling is that notable software will be written about by independent people if the software publishers want this: journalists always like a story. The latest software I have been involved with, FORscene, has dozens of articles as it is notable in many ways, even though it doesn't (yet!) have the user base of the big applications. Not being written about yet doesn't guarantee non-notability, but the feeling I concur with is that Wikipedia can wait for things to become clear. Stephen B Streater 19:29, 30 March 2006 (UTC)

I further simplified the guidelines just now. Basically, I got rid of all the sub-points to bullet 1, and I made an "awards and honors" section; most other notability guidelines include this kind of thing. Bullet point 2 is the important difference: it basically says, "if this software has any interesting properties, and you can back that up with a reliable source, it's notable." So this would cover the first software to do X, the best at X, software that influenced the development of X, et cetera, which lets us drop the "historical significance" thing; it's included in #2 now, and with the added bonus that we're reminding people it has to have sources. Mangojuice 21:59, 30 March 2006 (UTC)


 * I like the idea of simplifing the guidelines as much as possible, and most of the last change seems ok to me. One thing that seems useful is to extend the list of examples of "non-trivial" publications. The ones that are currently in are sufficient for software of general use (e.g., Microsoft Word), but may not cover "specialized" software. How about: "...is independent of the software developer, such as a major media news media, a book, a peer-reviewed scientific publication, an article in a reputed technical magazine. A single such publication that is specifically about the software is sufficient, multiple works are necessary otherwise." This is mostly the same as the last version, but changing the list of example "non-trivial publications" is easier.
 * Reading again, I realized you have also made another change, which is the mention of "lower profile sources". What do you mean exactly by that? In my proposed change above I took that to mean "a work that is not specifically about the software". Is this as intended? - Liberatore(T) 23:02, 30 March 2006 (UTC)


 * What I was thinking was, minor coverage such as in a local paper or an e-zine would be "low profile," while coverage in a major source like a national magazine or paper would not be. I wasn't thinking specifically about the degree to which an article is about a piece of software; I figured that whole business was covered under the word "subject". Mangojuice 23:26, 30 March 2006 (UTC)


 * These latest changes look like a big improvement to me: more concise; includes software interesting enough to get independent articles; and verifiable. Stephen B Streater 05:38, 31 March 2006 (UTC)


 * To MS: this looks reasonable, but the guideline should be more explicit about what is a "low profile" publication. I also believe that multiple mentions in "high profile" publications that are not specifically about the software prove notability. - Liberatore(T) 07:45, 31 March 2006 (UTC)


 * I agree with you. I'm not sure how to write that compactly though.  Want to take a stab at it?  Mangojuice 12:44, 31 March 2006 (UTC)


 * I gave it a try, and the result looks reasonable. Whether a publication is "high profile" or "low profile" is still somehow a matter of opinion, but I don't think this will be a major problem. - Liberatore(T) 14:11, 31 March 2006 (UTC)

Please! No More RuleCruft!
Enough already. There are plenty of guidelines, about guidelines about guidelines. There is virtually nothing that is verifiable that needs rules like this. Please stop. For great justice. 00:40, 20 April 2006 (UTC)


 * Hear hear. This is just plain stupid. I notice that articles are being removed on rules. And I don't feel like wasting my time reading all this "cruft", let alone writing them... Guaka 19:31, 22 April 2006 (UTC)


 * So how do you propose to ensure some consistency in Wikipedia? Stephen B Streater 19:47, 22 April 2006 (UTC)
 * What are you worried about? If you find a field that doesn't have enough covereage, write about it! There are tons of policies that rule out nonsense. Don't waste everyone's time on rulecrufty instruction creep. This basically repeats verifiability, only more confusingly, and with more POV. It's crap. 67.109.101.226 22:48, 26 April 2006 (UTC)
 * Many verifiable subjects are currently deleted partly because they are not deemed notable enough - including an article I wrote about a piece of software which I think needs coverage. WP is not just a collection of information. Guidelines on what notability is help ensure consistent standards. Stephen B Streater 07:25, 27 April 2006 (UTC)
 * The problem is that notabilty is an inherently POV idea. What is notable to you may not be to me. What is notable to a software engineer is not notable to a clasicist, and so on. If we apply verifiability we will only end up with things that are well referenced in credible places, and that will remove nearly all of what the moral panic people are worried about. For great justice. 16:01, 27 April 2006 (UTC)
 * Yes. That's true.  That's exactly why we NEED guidelines, so that we have some idea what the middle ground of consensus is.  If you and I disagree about notability of something and there's no guideline, no one knows how to choose between our arguments, apart from simply voting.  But if there's a guideline that reflects community consensus to back up one of us, it makes things much clearer.  Mango juice talk 16:33, 27 April 2006 (UTC)
 * We HAVE GUIDELINES! Verifiability, Sources, WP:NOT, etc, etc. Everyone agrees on the. I don't simply diasgree with you about 'what is notable', I disagree that the concept of notability is valid for an encyclopedia. The existing guidelines are enough - give me a real-world example of what gets through them that you are worried about. For great justice. 16:57, 27 April 2006 (UTC)


 * Correct. It's an expression of how the community in general interprets WP:NPOV, WP:V and WP:NOT in the context of individual subject areas.  Software which is adjudged not notable by these criteria has insufficient verifiable mentions in reliable external sources to ensure NPOV, or is unencyclopaedically indiscriminate, or is possibly being added as a way of selling the product (WP:NOT).  Some people think that "nothing verifiable" should be removed, but actually I don't see anybody wanting to include every verifiable fact or item, however trivial, in the end we all just have different personal thresholds for what constitutes content which merits inclusion.  These guidelines allow us to achieve some kind of consensus view on where the community as a whole wishes to draw the line.  In general it's pretty permissive - things will meet these criteria long before they get close to any other encyclopaedia. Just zis Guy you know? 16:50, 27 April 2006 (UTC)
 * So why not just use the criteria of which software has sufficient verifiable (in the sense of the WP policy) material to write an article about? Why bring in your POV? Re your comment on 'everything verifiable' - please read the policy. It does not cover 'everything verifiable'. For great justice. 16:57, 27 April 2006 (UTC)
 * Yes. And why not giving some hints as to whether software is sufficiently verifiable? Or whether an article about it qualifies as indiscriminate collection of information? I'd also like to see examples of articles deleted under this guideline that should have not been. - Liberatore(T) 17:35, 27 April 2006 (UTC)
 * WP:V should be enough. If it's not, we can take it to the talk page there and refine it. For examples, see 70% of AFD. For great justice. 17:59, 27 April 2006 (UTC)
 * A lot of software is rejected for non-notability even though it is verifiable. As an example, look at Diva. Stephen B Streater 18:13, 27 April 2006 (UTC)
 * Yes, a lot of bad decisions get made in the face of policy. That doesn't mean it has to continue. We don't have to enshrine bad habbits as policy... For great justice. 18:21, 27 April 2006 (UTC)
 * WP:SOFTWARE is not policy. It's a guideline (actually, a proposed one). It helps having consistency among AfD discussions by reaching consensus in a centralized place instead of discussing general principles every time. What do you think is bad habit? Consistency, consensus, or centralized discussions? - Liberatore(T) 18:37, 27 April 2006 (UTC)
 * Exactly. WP:SOFTWARE is not policy, and will never be, because it reflects the POV of a few people. It is 'what we like'. Using that as deletion criteria, rather than NPOV policy, is a bad habbit. Making rulecruft (masses of pseudo policies that a minority of people actually support) is confusing, unhelpful, and a bad habbit. For great justice. 18:43, 27 April 2006 (UTC)
 * Policy does not tell which articles to include, and specifically say that not everything that is NPOV and verifiable is to be included. Without rulecruft like this guidline, people would have to take decisions at every AfD, which means that they can very well base decision on 'what they like'. Guidelines do not only tell that some articles have to be deleted, they also say that other articles are supposed to be kept. - Liberatore(T) 06:37, 28 April 2006 (UTC)
 * Policy does spell out which articles should be deleted there is no need for more rulecruft, since the existing policy allows for deletion of anything inapropriate. I've asked elsewhere, and got no response - but perhaps you can give me some real-world examples of the kinds of things you want to delete but existing policy doesn't allow?
 * How many articles which have been deleted should have been preserved though? If each AfD makes up its own interpretation of WP:NOT: "Wikipedia is not an indiscriminate collection of information", this is also a concern. Stephen B Streater 18:06, 28 April 2006 (UTC)
 * Of course it's not an indiscriminate collection of information, but what does that mean? It gets trotted out whenever anyone wants to justify deleting anything. A structured, detailed article about something that you are not interested in is not an indiscriminate collection of information. For great justice. 18:32, 28 April 2006 (UTC)
 * Re afd, many people on afd ignore policy and guidelines anyway, the point is to try to keep things off afd, which is also a destructive waste of time, and make people spend more time improving and editing articles. For great justice. 17:20, 28 April 2006 (UTC)
 * AfD is a learning and teaching experience too, so it's not all bad. Stephen B Streater 18:07, 28 April 2006 (UTC)
 * It's a pretty bad way to learn or teach. It wastes a huge amount of time that could be spent writing or improving articles that should never end up there. It not only wastes the time of people who list things, but also of those who have to watch for stupid listings, and spend time explaining why they should not be. It's not simply things which are deleted that should not have been, but the huge number of listings that don't get deleted. For great justice. 18:32, 28 April 2006 (UTC)
 * I also think too many things are deleted, but most of the things I see on AfD are flawed. I would do things differently, of course - include more articles, but also have a quality rating for each article which would reflect the number of independent people who had edited it and seen it. Articles with only one contributor eg Nokia N93 will not have the depth or sophistication of articles with wider contribution eg Go. Stephen B Streater 19:20, 28 April 2006 (UTC)
 * It's better to have a guideline that expresses a rough consensus of what most people think constitutes an appropriately encyclopaedic subject, than to have endless debates about what precisely constitutes indiscrimiinate, verifiable, etc. And although AfD does make wrong decisions (much more often than most of us would like) it also acts as a rapid-improvement forum in a way which, sadly, the article improvement drive is not.  I have seen many articles rescued from why should I care? to acceptable or even excellent articles during AfD.  And I have also seen rafts of vanispamcruftisement rightly deleted.  WP:SOFTWARE exists to reduce acrimony.  And yes there are those at the ends of the spectrum of deletionism and inclusionism who disagree with all the various guidelines, but in the end that's just an artifact of the fact that Wikipedia is a very disparate community which works on respecting the views of those you disagree with.  Much better to have an agreement which most people support than to have endless wars and POV-pushing. In an ideal world policy would be adequate, but interpretation of policy can be subjective, and where that happens disagreement causes friction. Just zis Guy you know? 22:36, 28 April 2006 (UTC)
 * No, I disagree. The fact that a plurality of people in the room share a POV does not make enforcing that POV on the articlespace appropriate. There is not anything like the scope for disgreement on what verifiability or adequate sources are than the mess that happens when you introduce a concept like 'notable'. Using AFD as an article improvement drive is just wrong. That's not what it's for.
 * Verifiability as an inclusion criteria respects the fact that any three people will have three different views on what is notable, and refers to the bottom line of can the facts be verified. Anything else is simply pouring fuel on the POV fire.
 * vanispamcruftisement is exactly the kind of rulecruft that we don't need more of. The article it was created to delete is not verifiable by the WP policy definition. Case closed, no rulecruft required. For great justice. 22:50, 28 April 2006 (UTC)
 * The idea of rating an article based on activity on it is interesting - being able to see how many people had an article on their watchlist would also be useful. For great justice. 20:37, 28 April 2006 (UTC)
 * I like that too. You can get some idea from "What links here" eg Links to N90, but I think your idea is better as if more people are keeping an eye on something it s likely to be better. A good measure would take into account how active users were - on average a vandal would be spotted in x seconds. It would be better for systems to work automatically than for lots of rules to try and force things to happen. Stephen B Streater 21:38, 28 April 2006 (UTC)


 * I support having a notability guideline for software. There is just too much software written in the world.  Having an article on every single piece of code in the world leads to editor fatique, reader fatique, etc. Johntex\talk 00:45, 10 May 2006 (UTC)
 * It might, if you chained them to a chair and made them edit/read them all! But the great thing is that you don't have to read what you're not interested in! And, if you later find that you are interested in something that you were not at first, it hasn't been deleted! For great justice. 16:35, 10 May 2006 (UTC)

Most software has plenty of documentation to make it verifiable. Verifiability alone is not enough of a criterion for what should be included in an encyclopedia. —Quarl (talk) 2006-05-10 19:37Z 
 * Please read WP:V and Reliable Sources. Simply having the manual lying around does not meet the existing standards. Likewise you dog, toenail etc and all the other spurious examples. For great justice. 20:08, 10 May 2006 (UTC)

Self-edits, a developer's perspective
Hi; I contribute to Wikipedia in areas unrelated to software, but two programs I've written, pydance and Quod Libet both have articles about them. My feeling is that both articles are unencyclopedic (as is almost every article on Wikipedia), but I'm not going to AfD them because obviously someone else disagrees. The problem I have is that the articles get out of date very quickly for active projects, and for active small projects that means no one upates them. Quod Libet is very unhelpful for example, having had two inaccurate comparisons to other media players, and having been at stub status for a very long time. The Russian article is more complete, but also 9 months out of date. I've updated pydance myself to improve the quality a little, but I felt uncomfortable doing it; enough so that I don't want to touch Quod Libet, since music player flamewars are common already. There's a huge music player feature comparison table somewhere, which is really out of date and probably was wrong even when it was written. Users are getting bad advice about what software serves their needs, and developers are basically having their software lied about.

I guess I don't really have a conclusion here, except: If you're going to put a thousand active free software projects into Wikipedia, you're going to need to keep a thousand active free software projects up to date, which is going to be hard if you don't let involved developers edit. piman 02:42, 6 June 2006 (UTC)


 * The point of notability is that enough people will keep them up to date. Several acceptable courses of action are available to you:
 * Mark the article as disputed accuarcy with
 * Write a full summary on the talk page, and wait for someone else to add it
 * Quote directly from WP:RS, with references, showing the latest upgrades you have made


 * The third is probably the best as no one can dispute what was actually printed about your software.
 * I'm working on a article here - it is much easier for me to find references than a typical editor. I am wondering how many more references it needs.Stephen B Streater 06:46, 6 June 2006 (UTC)
 * As I said, my software isn't notable. And it isn't just mine - there are some people that seem hell-bent on adding every piece of free software to Wikipedia. See Gaphor, File roller, Liferea, Gconf-editor, Gnomesword, Katapult, KFTPGrabber, Thunar, and undoubtedly dozens more. I use some of this software, and have heard of almost all of it outside of Wikipedia. But none of it is notable, even though it meets these guidelines, because they're in "mainstream" Linux distributions. Most distributions contain thousands of pieces of software. Free software fanboys add these and then never maintain the entries. And I have better things to do than add yet another place to visit every time I release a new version of my software. piman 19:12, 6 June 2006 (UTC)

The appropriate level of notability
The recent edit here has significantly tightened the notability guideline. Previously any criterion listed could be considered sufficient for notability. The new edit says this may or may not be sufficient. As this is a proposed guideline, I thought that some discussion was appropriate. I prefer the old version for two reasons: Does anyone else have an opinion? Stephen B Streater 19:11, 26 July 2006 (UTC)
 * This is a potential significant tightening of the proposed guideline which could exclude notable software
 * The new proposal is so vague as to render the guideline less useful


 * As the editor of the new version, I do have an opinion. I'm going to reproduce and sign your points below, and reply to them. Feel free to reply to my replies.--Chealer 23:48, 26 July 2006 (UTC)


 * This is a potential significant tightening of the proposed guideline which could exclude notable software Stephen B Streater 19:11, 26 July 2006 (UTC)
 * The proposal isn't meant to justify excluding non-notable software. It's meant to confirm that some software is notable by certain criteria. If the previous proposal confirmed that a certain software was notable and the new proposal doesn't do that anymore, it means that evaluating the notability of that software will be left to people's judgment, not that the software will be necessarily considered non-notable.


 * The new proposal is so vague as to render the guideline less useful Stephen B Streater 19:11, 26 July 2006 (UTC)
 * The old proposal is already vague (read: subjective). I'll comment the criteria to highlight that:
 * The software has been verifiably the subject of non-trivial published works whose source is independent of the software developer, such as a major media news piece, a book, a peer-reviewed scientific publication, or an article in a reputed technical magazine. A single such publication that is specifically about the software is sufficient; for publications that mention the software while not being specifically about them, and for publications of lower profile (such as a local newspaper or an e-zine), multiple such works are needed.
 * Saying that multiple such works are needed already makes it impossible to justify notability by "publications of lower profile" and "publications that mention the software while not being specifically about them".
 * The software is/was innovative, significant, or influential in some specific way, and this is verifiable from reliable sources independent of the software developer.
 * The software was clearly significant for its developer(s) and influenced the life of its developer(s). Saying that it wouldn't be significant or influential in general would be subjective.
 * The software is among the core products of a notable software developer or vendor.
 * The software is included in a major operating system distribution such as Debian or Fedora Core, and the maintainer of the distribution is independent from the software developer. [...]
 * The product is so well-known that its trademark has suffered from genericization. (ibid.)
 * This is one point I would say is not subjective, but at this point, nobody will challenge its notability, so this point is IMO superfluous.
 * The software has won an independent award or honor, including mention as one of the top ten applications in its class in an independent published list.
 * So rather than having debates about the subjective terms, let's be honest about what these criteria are: helpers to prove that a software is notable. A subjective criterion can't prove something.--Chealer 23:48, 26 July 2006 (UTC)

The above analysis is in part faulty, being partly based on a misguided conception of "more notable/less notable" and partly based upon the error of thinking that an accumulation of trivial works adds up to a non-trivial one. No matter how many magazines publish trivial works of the form "Version 6 of Acme Corporation's AcmeWord word processor is now retailing for $400", that doesn't add up to a non-trivial work about the word processor. A non-trivial work would be a full-length product review. However, I firmly agree that the "significant or influential" criterion should not be included. The primary notability criterion does a perfectly good job of reflecting significance or influence in the world at large without requiring Wikipedia editors to make their own judgements of how significant or influential something is. Uncle G 15:28, 11 August 2006 (UTC)

Awards
There is no real need for an explicit awards criterion. Awards are just a single special case of the primary criterion. When computer magazines publish their awards, they often have accompanying articles, and when they don't what they do publish is of no worth in creating an encyclopaedia article. The PC Magazine Editors' Choice Award given to WebTrends was accompanied by a full and detailed review, for example, thereby fulfiling the primary notability criterion by being a non-trivial published work created and published by someone other than the author of the software. In contrast, where awards are just bare lists of product names, such as is the case for the Codie awards, the award doesn't provide any sort of source material that can be used to create an article. There is no way that the 2005 Codie awards can be used to write an article on The Everdream Management Suite, for example. Thus the package fails the primary criterion because the published work is a trivial one that is no more than a directory listing. The point of including secondary notability criteria is to fill in the gaps where the primary notability criterion fails. (The secondary notability criteria in WP:CORP, for example, are to ensure that our coverage of the Fortune 500 et al. is comprehensive, in the exceedingly unlikely event that the primary notability criterion should fail.) In most cases, and for most alternative notability criteria that people can come up with, the primary notability criterion doesn't fail. Uncle G 15:28, 11 August 2006 (UTC)
 * Many software awards are not given out by computer magazines. For example, the FORscene product won a prestigious industry technology award from the Royal Television Society. These organisations don't write articles. So award criterion is not included in article criterion. Stephen B Streater 15:46, 11 August 2006 (UTC)
 * These organisations don't write articles. &mdash; Then the existence of the award has no bearing upon whether an article can be written, and a notability criterion that explicitly includes all packages that are the recipients of such awards is effectively giving us a whole raft of unexpandable "X is a software package that won award Y." permanent-stub articles, and is a bad criterion. Uncle G 15:55, 11 August 2006 (UTC)
 * Furthermore: FORscene was the subject of non-trivial published works from sources independent of its authors, e.g. the review by Gavin Aimes in Showreel magazine that describes the package in detail. Once again, the primary notability criterion is sufficient on its own, without need for any secondary "awards" criteria. Uncle G 16:07, 11 August 2006 (UTC)

Core products and operating system distributions
Wikipedia is not a directory. It is not a directory of all of the software packages that come bundled with Debian, for example. Neither of the "core products" and "included in a major operating system distribution" criteria are well conceived. (The latter is particularly egregious, as it effectively introduces a mathematical formula that involves dividing the number of distributions that contain the package by the total number of packages in all distributions. That's a wholly unworkable metric.)  As pointed out above, the point of including secondary notability criteria is to fill in the gaps where the primary notability criterion fails, in order to ensure that Wikipedia's coverage of certain areas is comprehensive. In other words: They add directory-like qualities to Wikipedia for certain limited areas. So we should be asking ourselves: What areas are we trying to ensure are comprehensive with these two criteria? WP:CORP's secondary criteria are aimed at ensuring that all of the links on List of Fortune 500 will be blue. Wikipedia has no "list of all software packages included in Debian". Nor has it "list of core products by notable software developer X" articles. So what gaps are the proponents of these criteria actually trying to fill? Uncle G 15:53, 11 August 2006 (UTC)


 * Presumably, Fortune 500 corporations are important because they affect many shareholders and customers even if they lack media attention. If a software package is in the core of a linux distribution then it's affecting millions  of developers and users.  If a magazine publishes a "list of 500 most important software packages" then would you be happy with inclusion in that list meriting a Wikipedia article? —Quarl (talk) 2006-08-11 21:54Z 

insanely biased towards free software.
getting into debian is not that hard, with a little determination pretty much anyone can become a debian dev and package thier own software to thier hearts content. I'd imagine fedora is similar though i don't have experiance with processes there.

the only other ways in essentially require you to be a big powerfull company or to have had the luck to get into print and have someone here recognise you (web review sites seem to get disregarded here unless they are the web arm of some print operation in almost all fields). Plugwash 21:47, 15 September 2006 (UTC)


 * I think the bias may be because of this. Free is being interpreted as free formats as well as free content. And preferably free software. Stephen B Streater 22:13, 16 September 2006 (UTC)


 * yes and i have no problem with wikipedia being written as free content, stored in free formats and run on free software. I also don't have a problem with help pages etc being pro free software (since the NPOV requirement afaict does not extend outside article space). However any policy that pushes this bias into the editorial side of wikipedia goes directly against wikipedias core principles. Plugwash 23:03, 16 September 2006 (UTC)


 * I agree. I've already negotiated with a free fan that "free" should not be in a "good" colour (green), with "patented" in a "bad" colour (red). Stephen B Streater 09:42, 17 September 2006 (UTC)
 * I'd say anything self-published as freeware or shareware is non-notable unless it picks up a substantial following somewhere.  &gt; R a d i a n t &lt;  21:54, 17 September 2006 (UTC)
 * What counts as a "substantial following" though? imo getting into a "package everything" linux distro doesn't make you any more important than getting good ratings on major freeware/shareware sites. Plugwash 19:39, 19 September 2006 (UTC)
 * In my experience the Google Test works surprisingly well for determining cult followings of recent software.  &gt; R a d i a n t &lt;  20:09, 19 September 2006 (UTC)
 * The flip side of this is that software included in a substantial number of distros is almost certainly notable, but that isn't discussed at all. And yes, releasing your software for free and having it incorporated into various larger free-software projects is a cheap and easy way to gain some notability.  But the point is that it does gain you some notability!  The only question is, how much?  (And not, for example, whether it's "fair" that small, obscure companies can't compete.)  Xtifr tälk 22:03, 26 October 2006 (UTC)
 * And the flip-flip side is that Linux distributions have hundreds of programming libraries and system administration tools that aren't notable at all outside of a technical manual.
 * Also, explicity listing "major" operating systems such as FreeBSD or Debian seems to be biased criteria, because their user-bases are comparatively tiny relative to (say) the Dell Computer freeware-bundle du jour. 64.171.162.77 04:33, 25 November 2006 (UTC)
 * The size of the user base has been rejected as a criteria for assessing notability (see Notability (software)). That some topic is "technical" has also never been considered, in general, as evidence of non-notability. Tizio 16:42, 25 November 2006 (UTC)

Call for Amendment
I, CherryT, motion to amend the Software Notability policy to include:
 * The Community
 * If the author/supporter of the article runs a poll/fourm post (preferably on Wikipedia in the Talk page for the article) and the community acknowledges the existence of the software and further goes to say that the software is notable or important AND the voting peoples are NOT a part of the company/producers, then the article is protected from deletion.
 * Terms
 * The voting body must be of 10 people or more and must NOT be a part of the company/producers of the software in question.

Please feel free to amend this amendment, or if you think it is appropriate, reply with Vote - Aye if you accept. If you reject this amendment and feel it is inappropriate, reply Vote - Neigh and please explain your reasoning.

--CherryT 04:21, 27 October 2006 (UTC)


 * Not gonna fly, CherryT. As a note, this is in reference to Articles for deletion/Within a Deep Forest. Andre (talk) 04:07, 31 October 2006 (UTC)


 * Wikipedia is not the house of parliament. We do not make "motions" and call for aye or nay votes. Nor do we write guidelines as a book of law. Rather, we discuss the pros and cons of what you're saying. ( Radiant ) 14:45, 20 November 2006 (UTC)

Long Tail Considerations
I think an important consideration for notability of software is that much of the value of WikiPedia is the ability to get great content on highly specialized subjects. Given this, readers of these specialized articles would likely be very interested in software (yes – even commercial software) that in a general context are not particularly notable.
 * Give us an example please? People who write their own small-time tools that a dozen people use worldwide, for instance, do not get an encyclopdeia article. ( Radiant ) 14:46, 20 November 2006 (UTC)

What does this policy proposal need?
I've been nominating a lot of things for deletion lately that fail WP:SOFTWARE. However, since it's not official policy, it leaves the decision whether to follow its notability limits up to the AFD contributor, who I suspect can often be a fan of the game in question and therefore have a bias in favor of keeping.

So, what does this proposal need added to it to bring it to a consensus level? What's next? How can we make this official policy? Andre (talk) 21:26, 8 December 2006 (UTC)
 * We don't; however, we could make it "guideline" similar to the other notability guidelines. I think you should drop a note on the village pump for final comments on the current version, and it may help if you can cite a few AFD discussions that show precedent for this guideline. ( Radiant ) 14:35, 11 December 2006 (UTC)
 * The above discussion is preserved as an archive of the debate. Please do not modify it. No further edits should be made to this page.