Note: this article was originally written in Spring 2004, in response to the belief of some editors that notability or importance should not be criteria for inclusion at all. Some of it sounds odd to me today, because even fairly strong inclusionists now seem to accept that a notability bar exists somewhere. They just have different ideas about where to put it.
I am a convinced deletionist.
That doesn't mean that I want to delete articles that aren't very good, or every article on something I haven't heard of, or every article on a non-academic subject, or anything like that. It does mean I'm convinced that while it may be hard to define what is "notable" or "important" or "encyclopedic", verifiability cannot be our only criterion for inclusion. In some cases, the best interests of our readership are served by NOT including an article, even if the information it contains is verifiable.
Some think that our purpose as an encyclopedia is to provide information. That's not really true. By Wikipedia policy, we don't add anything to the net sum of information that's available out there (although in some cases we make it a lot easier to find.) Every piece of information in Wikipedia could be obtained elsewhere. A large part of our purpose is to provide organization and filtering of information.
Just as an individual encyclopedia article needs to be shaped to provide the most useful, interesting material, so too the encyclopedia needs weeding. The principles are very similar. We could have 30-page articles on broad subjects like physics. Why don't we? It's not about bandwidth; we could split the article into pages if we wanted. It's because the purpose of an encyclopedia is to summarize. People go to an encyclopedia because they don't know about a subject, and want to get an overview of the most important information about it.
There's also an issue of implicit inaccuracy. People want relevant, useful information. If we have an encyclopedia entry on something, we imply to our readers that this is something worth knowing. If, for example, I read an article about a rock band from Detroit, I would assume just from the article's existence that this band is one that many people in Detroit have heard of. If, in fact, the band exists in somebody's garage, I as a reader will be misled.
The more obscure a subject, the more difficult it is to verify. That means that in practice, most such verification will never be done, and when it is, it will be time-consuming. As an example, I spent a good deal of time trying to verify information about the card game Ambition, whose article was written by the game's creator. In the end, everything remaining in the article is verified, but it took a great deal of effort to sort out exaggeration and self-importance from fact; finding information on such an obscure subject is difficult. The amount of time required was far out of proportion to the usefulness of the article.
The more obscure a subject, the more likely that an article will fall badly out of date. I'm certainly not going to spend any time telephoning primary schools to verify their existance and update information about them. Are you? Between the possibility of hoax or incorrect information, and the likelihood of out-dated information, such articles simply aren't useful.
Also, it is important to understand that standards of importance for inclusion don't just mean the deletion of a few fairly harmless articles. They also discourage a potential flood of publicity-seekers, advertisers, self-important nobodies, and well-meaning but irrelevant nonsense. If we were to adopt a policy of allowing in anything that was verifiable (especially if we allow verifiable to mean "could in principle be verified, although difficult enough that noone would ever bother,") others would realize this. Friends would tell friends "Hey, you should write an article on yourself. I've got one, you should too!" Professional colleagues would do that same. Worse, every crank would realize that as long as he can prove that a) he himself exists, and b) that he is making a bunch of crazy claims, that we will accept an article about him and what he says. Failing to delete the trickle could easilly result in a flood.
Lists and categories would become completely useless. After all, if we had article on something, and it fits in a category, then naturally it ought to get the category tag. But then, how will anyone find the articles in that subject that are actually worth reading? If I wanted a learn a bit about the development of electronic music (a subject I know virtually nothing about,) I would never be able to pick out the well-known, influential groups from a list of several hundred band names. The more names are on that list, the less useful the list is to me.
Similarly, we'd need far more disambiguation pages than we currently have, which would add a lot of maintainance work. And all this would be for articles about things almost no one will need or want to know.
And do I even need to say anything about the bandwidth and server resources that this would waste? Every estimate of Wikipedia's growth rate has suggested that technical limitations are the main damper on our growth - when the site becomes too slow, people stop editing. Isomorphic 03:22, 26 Apr 2005 (UTC)
I sometimes get the idea that people forget about the rest of the internet, and even more radically, that there exist resources other than the internet. Rarely do we choose between having useful information on Wikipedia and having it available nowhere on the web. A lot of stuff really belongs elsewhere on the web, for one reason or another, and there's nothing wrong with leaving it elsewhere. We are under no obligation to provide anything that anyone might ever want to know. The words "potentially useful" and "encyclopedic" are not synonyms. In my Wikipedian nightmares, I see people creating things like Bus schedules for Topeka, Kansas and London School of Economics Class of 1998.
If blanket statements about inclusiveness still sound like a good idea to you, well, I don't know what else to say.