Kill the Stub Pages

From Meta, a Wikimedia project coordination wiki

You know, I'm noticing that if I start from the homepage and just randomly follow links, I often end up at stub pages. I personally usually am paying attention to the RecentChanges page, where the bulk of the links are of course longer, more frequently edited pages; however, I can see how someone new to the project and not aware of RecentChanges might browse around, get frustrated at the stub pages, and justifiably conclude it is all just stub pages.

The "Featured article" section of the main page gives newcomers great articles.

I'm not at all certain that this is a characteristic of the project that will necessarily diminish with time - I think it may be more due to the speed and ease of creating pages that are lists of topics, compared with the time and difficulty required to make just a few content-ful pages. For example, in the time it takes one person to write a decent article about Nebraska, another person can create a listing of 20 towns in the state with matching "stub" pages. I am wondering if we may be "stuck" with having a large proportion of stub pages, just due to the way wikipedia works, no matter how far along wikipedia is?

But isn't that the nature of all human knowledge? The bigger Wikipedia grows, the bigger its borders with unknown become.

For example, browse w:Language. Nice list, but only a handful of the items are filled in, and not many of the filled in ones have much detail. Or w:Agriculture, w:History, w:Cooking, w:Politics (now much improved), w:History of Science and Technology, w:Hobby, w:Music, w:Sport... Yes, these are very nice, reasonably complete lists of topics, all of which would be nice to have articles on, and yes there are some adequate articles (and a few _great_ articles) in there, but by and large these pages scream "WIKIPEDIA IS MISSING LOTS OF INFO!"

Since this was written, the edit links in the Languages article have been filled in.

If this is true, then it could be a difficult criticism to shake. I think up until now, these pages have served a very useful purpose, because wikipedia's structure was geared more towards the aim of gathering _writers_, not gathering readers. Writers NEED incomplete lists; readers need complete ones. But I would argue that we have plenty of writers, and that to get more, we want to shift Wikipedia towards the users. And that means we need to look more complete.

Two ideas spring to mind for how we can address this.

The first idea is cultural. Simply encourage the creation of longer pages over shorter ones. Adopt the stance of "No page is better than a stub page" and discourage people from creating a new topic unless they're willing to write at least a full paragraph on it, and definitely start discouraging the creation of more empty lists. Also, adopt practices that give positive encouragement to people who create long articles or greatly fill in stubs - this was one of my objectives with Wikipedia_NEWS, before we got so overwhelmed with traffic that I gave up updating it and just started writing long articles myself. :-)

I know the idea of discouraging stubs runs counter to some of the essays and policies on the site, so I understand this would take significant mulling-over and might not be acceptable, but I think this approach needs to be considered.

The second idea is more direct and radical: Encourage deletion of empty lists. Lists are not *that* valuable; as mentioned above, we can count on people creating many more such lists, quickly and easily compared with creating the more desireable lengthy pages. Only leave the few links that already have significant material attached. The absence of empty lists will give added weight to the filled in articles, for the first-time visitor. Or if nothing else, weed out some of the unfilled links, to bring the proportion of "long pages" to stub or unfilled pages to at least 50%.

As an example, consider the pages w:Archaeology, w:Transport, and w:Computer Science, which even though as lists they are incomplete, the fact that the majority of the items in the lists actually have filled in pages (and many of them quite lengthy) makes these pages *look* a lot more complete. And of course it's *easy* for someone to add a new topic onto the Computer Science page when they have something to write on.

One argument in favor of keeping long lists of empty topics is that it "gives plenty of openings to encourage people to write about". Agreed, however we have a disproportionate amount of empty topics listed, and it's going to make the user very frustrated, and even if we got rid of 90% of them, we'd still have plenty of empty links for people to write on.

-- BryceHarrington

What do you think of my rendering of w:politics? In case it's not still there, I moved the non-completed links to w:Politics/Wanted, and put a plea on that page for authors to write those articles and move them to the main page when they're more than just stubs.

For authors, there's now a useful list of articles to be completed. For readers, they get links only to the completed articles.


Wow, that looks a lot cleaner, plus it provides the same "articles needed" info as before. I like it. -- BryceHarrington

Like we have w:Requested articles, we could list stub articles on a w:Fill these stubs page. I might put an automatic "stubs" list on my PHP wiki. --w:Magnus Manske

I think this is a fantastic idea. I don't know where you come up with all these ideas, Magnus. :-) I think it's a great idea for us, those of us who care, to constantly be adding stub articles to a central page-o-stubs, and make it our mission in life to destubify them. --LMS

I have a third suggestion (I've been thinking about this a bit lately): some of us should concentrate our efforts on our "portal" articles, the catagories that are listed on the front page. Ideally, they should all serve as an extensive overview to a topic (w:Philosophy is currently pretty good). Personally, I'll be working on a w:Communication Studies article over the next little while. If other people could pick one of the portal pages they know something about, we can thread the web a little better. -- w:STG

I totally agree here, too. It seems to me I once kvetched about that. :-) --LMS
I too agree. Since people just "clicking into" the site will hit the portal pages first, making sure each of those pages are good and long would help a lot (and thankfully there's only about 50-60 such pages, so is a doable endeavor even for 1-2 people). -- BryceHarrington
I've just finished working on w:Geography. It had originally been just a couple sentences and then a list of (mostly empty) geographical terms, and I expanded it with a more detailed description of geography as a science. Could it be true that the portal pages are less provocative and less interesting to work on because they're so generic, and people prefer working on specific topics? -- BryceHarrington

Just my own little two cents worth -- one reason that no stub page is better than a stub page is that no page generates a question mark after the term when mentioned elsewhere, which is an admission that we don't have the information, and a "teaser" to all of us who can't resist the urge to write. :-) Whereas a stub page satisfies our thirst *just a little bit*, which perhaps doesn't encourage authors nearly as much.

Whenever we are issuing forth propaganda about the site, we should never use a count of articles that includes stub pages. It's much stronger to say that we have 8,000 articles longer than 100 words (or whatever metric we like), than to say we have 15,000 articles, but who knows how long they are.

--w:Jimbo Wales

Or the proverbial "500 word essay"? ;-) Btw, very good point regarding having no stub page might be more of a teaser. For me, at least, there's a certain enjoyment in the initial creation of a topic, that I don't get if a stub's already been put there. I've no idea if others feel this way tho. -- BryceHarrington
I get enjoyment from creating stubs. :-) Honestly, I think that something is better than nothing. The question mark placeholders give me the message that no one cares about the article; at least when it's filled in with a hyperlink I can tell that at least one person cares about the topic. So what if they only create a stub? I see no significant difference between being the initial author of an article and being the first author to replace a stub with something substantial. <>< w:tbc

Generally, I'm not really sure how to think about this. There are a number of different issues here, some of which are unresolved in my head, so I don't know what to think.

I agree, stubs suck and they make Wikipedia look bad. On the other hand, stubs represent content that someone was willing to donate, and why should we discourage people from contributing what they can? That seems to work against the very source of our activity. But, on the w:gripping hand, maybe Jimbo is right--maybe having the question mark is more provocative. I just don't know. Who really knows how people are thinking about this, short of doing some sort of study? One might just as easily say that the question marks, if there are many of them, overwhelm people by showing them how much there is to do. At least a stub article is a start, even if it's a pathetic start.

If some of us make it our mission in life to build w:Fill these stubs and then to do as the title says, I think we won't need to set any specific rules which might discourage people from participating.

I think w:Wikipedia/Our Replies to Our Critics will help at least somewhat to dispel the notion that Wikipedia is lightweight because it contains a lot of stubs.

Jimbo's suggestion that we advertise an even more restricted set of articles, as the ones we're willing to take credit for, will help a lot too. On the "what number should we use to advertise our number of articles" question, I'll post something on w:Malcolm Farmer/How many Wikipedia pages are there. --w:LMS

Yes, stub articles is a start, and I agree this is hard to think about because it's been our policy for so long that even minor stub pages are okay. But listening to the outside commentary we've been getting lately makes me think the stubs are doing us a disservice. I'm taking an intentially provocative stance in the above essay to stimulate thinking and am betting that others can help find less radical solutions that are still effective. Working on the portal pages, for instance, I think would do a lot of good. Or as another example, maybe we don't need to *discourage* stubs (and certainly we shouldn't berate people for making them) but perhaps we could just make it abundantly clear in the newbie docs that one should put something *meaningful* on a new page, at least a full paragraph. On the other hand, we discourage mere dictionary definitions already, for example, so maybe we could make a clear stance...
Regarding making a mission to fill in stubs - well certainly this is going to necessary, but please note that the growth of stub pages can greatly exceed our ability to fill them. So we need to slow that growth so we can catch up.
w:Wikipedia/Our Replies to Our Critics might answer some of the concerns about the lightweightness, but note that anyone who browses around enough to find that *particular* article will probably have run across some other stub pages. What we're concerned with is the many people who are looking to make a quick judgment by just link clicking from the main page. We want them to make the quick judgment that Wikipedia Has Lots Of Good Stuff, and dive in to find more.
Btw, thanks for giving this some thought! I've got confidence that you'll help guide us to a good solution. :-)
-- BryceHarrington

I wonder if there's an automatic way to search for articles that are smaller than a certain size! Even if not, it might be easier to program such a search than to create and maintain a w:fill these stubs page by hand.

I'll try to put links to the reply-to-critics page prominently in several places (the welcome page and the FAQ, at the very least). --LMS

Hmm, good idea. Here's code to do it:

First thing, nuke all the empty pages that just say "Describe the new page..."

 mkdir /tmp/emptypages
 cd [....]/wiki/lib-http/db/wiki/html
 for file in `find . -type f -size -1400c \! -name '*["]*' -print | \
   xargs grep 'Describe the new page here.' | cut -d: -f1`
   mv $file /tmp/emptypages

Review and delete the stuff in /tmp/emptypages at your convenience.

Now for the fun part

 cd [....]/wiki/lib-http/db/wiki/html
 for file in `find . -type f -size -1800c -print`; do ls -ol $file; done

This one gets a listing of all the pages which have fewer than about 500 characters worth of content. Knock the number up or down depending on where you wish to draw the line.

-- BryceHarrington

  Won't this delete redirect pages as well?

Well, I'm an admitted newbie, but I really like stub pages. I have a wide range of knowledge, but shallow in most areas not related to science or computers. I also tend to have a short attention span (blame video games). I think it's kinda neat to just cruise around Wikipedia and add stuff where I can.

For example, I happened across the diesel engine page. At the time, it only had about 2 sentences of info. I'm not an expert on these engines, and I'm not going to write a long treatise on their history, design, and operation. But I knew slightly more about them than what was currently exisiting on the page, so I added it. Someone else came along later and added more. Someone else editied it to make more sense. Now it's a decent entry, though by no means completely comprehensive.

But the process is cool. The process is fun... at least for guys like me.

Some people don't want to take 2 hours (or more) to write a decent-length essay. I think Wikipedia needs every little bit it can get, as long as it is correct.

-- ansible

The exact minimum length for a starter page (if you can't write a substantial amount anyway...) would seem to be just enough content to start a fight. Nobody wrote anything in the w:Java programming language unti I started it, then people ripped that sucker apart with new content. It would seem that many people are much more comfortable adding a to (or disagreeing with) a page than creating one, so if we can pique people's interest with a single paragraph, thats all for the better.

Perhaps to facilitate something like this, if you have something to add to a nonexisting page, add it. but also add some skeleton structure to facilitate other people adding trivia easily, because they see a place for it.

Dictionary definition==bad, enough to build on==good.

--w:Alan D

If you must write stubs (and of course you must! Do it! You know you want to!) see w:the perfect stub article for a few suggested guidelines. --w:LMS

I do have to respectfully disagree with the main premise of this article. We all have random facts, figures and info in each one of our heads. Quite often the knowledge we have is not enough to start off an article with a proper definition (that is, a good stub). However, if someone went to the trouble to try to synthesize something into a one line definition, then anyone can then input their random info about the topic and therefore expand the stub into an article. This, along with an improved "find and fix a stub" utility, and following Larry's "perfect stub" criteria should work. maveric149

I have seen some stubs gradually grow into good articles, and I have had several very enjoyable sessions of Stub Tennis with other Wikipedians. This is a crucial part of Wikipedia's strength: it isn't just peer review, it's collaboratively written. -- Tarquin

"The second idea is more direct and radical: Encourage deletion of empty lists. Lists are not *that* valuable".

Speaking personally, I feel the author's views (legitimately) express a frustration that Wiki is not yet complete; but I feel that frustration at its incompleteness should not manifest itself in the form of 'drawing a line in the sand' and cleaning up the database with the intention of throwing it open to a wider viewing audience.

As long as Wiki exists, part of its raison d'etre is to encourage people to play a part in it. This also means stressing the areas which are not adequately covered, in the hope of encouraging somebody to play a part. If all Wiki's weak areas were reduced to articles with no links to even stubs, then there is no incentive for knowledgable people to contribute - strangers may not even realise they have the opportunity to contribute.

The presence of stubs encourages people to fill in the information they themselves know. Explicitly deleting those stubs in the vague hope that somebody will one day create them anyway is virtually an anathema to me. -- xneilj

Just for correctness sake, this is Wikipedia. "Wiki" can be found here. Wikipedia is not Wiki.
I think we need to differentiate from stub pages (which are very short articles) from junk pages (which have zero or very close to zero useful information in them). If a page is created that has a decent one-line definition then we should rejoice that somebody has started that subject and then add to it. But if some creates an article for Somewhere, Some State and the only content is "Somewhere is a city in Some State." then that should be deleted because it has zero information in it that cannot be obtained from the title. Of course random junk (such as aflkjal;kfjds;al) or vandal-created pages whose content is "This is so gay" should also be deleted on-sight. --Maveric149
There's also a distinction between temporary stubs and permanent stubs. Temporary stubs can grow reasonably quickly into proper articles (though they may not, by quirk of contributors). Permanent stubs either can not do that, or will not do that. Examples: obscure characters from the Simpsons or Atlas Shrugged. (more controversial example: obscure people who died in 9/11). I still think merging is the best solution to permanent stubs. MyRedDice

Duhhh, this is so cut and dried its not funny - MBTI 101 J's hate stubs P's love them

I dont know if this suggestion is practical, or useful or even new, but what comes to my mind is: why can't we use another color on the links which are directed to stub pages? Of course it would require that "everyone" uses the {{msg:stub}}-command when appropiate, but then the newbies would, I believe, rather quickly learn that following those links are not going to give 'much' useful information - and they won't be as "scared", which I understand is one of the main reasons to dislike the stubs. What do you think? Much work to implement this? Mikez 19:11, 2 Feb 2004 (UTC)

You can already change the colour of links to stub articles. Change the Threshold for stub display setting in your preferences. See MediaWiki User's Guide: Setting preferences. Angela 23:45, 2 Feb 2004 (UTC)
Thanks for the info. But when considering the discussion above, maybe such a threshold should as default be a non-zero number? At least for those "newbies" who are not logged in.

[bold emphasis added by Stephen Brooks, as he thinks this is a good idea. Have levels of 'missing', 'stub' and 'complete article', the first two of which come with a sort of warning.]

I'd just like to say that stubs are not bad. It is you people who sit complaining about them that is. the lack of sensitivaty is reasonably understandable, but if they seem to pick your bone so why dont you do something about them instead of constantly bickering about how the seemingly "stupid stubs" and go searching for one on a topic you are interested in exploring. This would give you something to do other than complain, and you could further expand your brains.

What's wrong with stubs? I hate pages where people just go on and on about the same topic. we, as a people, should not have to sit in front of a computer for 10 minutes, straining our eyes to read tiny print, only to get a little information about a topic. I know some topics have a lot of stuff about them, but people need to get to the point. I think there should be a limit as to how long a page is, not how short. Kill the fancruft pages!!

      -Camille Long 

if you kill the stubs, the wiki proyect is never going to grow.

I think one problem with stub-killing can be seen in Wik/de. Short articles are rapidly deleted, not because they are wrong but, in effect, because they are stubs. Consequently, there are all sorts of h les in Wik/de. If people are concerned about links to stubs frustrating people, then here is a suggestion to fix that up: make a different designation for such links, e.g., color them green or have the word stub or a symbol for it in place by the linked word. - kdammers of wikipedia.

See Also: Kill the fancruft pages