Wikipedia is a work in progress. By definition, the user-written online encyclopaedia is in a state of flux, since every entry in its database is open to change, writes Jim Colgan.

In the four years since its creation, the site has amassed over 450,000 articles in its English-language version alone, and an increased profile in the past year is sending it millions of visitors a day.

With the added attention comes added scrutiny, and recent criticism of the site sparked a debate about how much it should rely on experts.

Last month a co-founder, no longer connected to the site, complained about its inherent "uneven reliability". Without a formal academic review process, he said, Wikipedia's future is uncertain. A month earlier, a former editor of Encyclopaedia Britannica wrote off the site as no more trustworthy than "a public restroom".

But as the debate about its quality persists, the encyclopaedia's expansion shows no sign of abating. It adds about 3,000 entries per day in 187 languages.

Wikipedia enthusiasts (known to each other as "Wikipedians") hail the project as a model of democratised information. The name comes from a tool called a 'wiki', which allows easy editing of content based on a model akin to open source software (the original Hawaiian word simply means "fast").

Many newcomers to the site are at first impressed by the wealth of information, but soon become confounded when they learn of its open nature, says Wikipedia's co-founder and current president, Mr Jimmy Wales.

"Five years ago no one had anticipated that you could have a website that anyone could edit like this and it not turn out to be junk," Mr Wales says. "When you think about the model that we use, it's insane."

To its millions of readers and the press outlets that cite it as a source (24 this month by the site's own count), Wikipedia is anything but junk.

The absence of barriers inherent in conventionally-edited encyclopaedias means that articles incorporate the latest available facts. The entry for tsunami, for instance, details the casualties of the Indian Ocean earthquake and links to video footage as well as describing the science behind the wave.

The collaborative system means the potential authorship encompasses the entire internet population.

By relying on this vast pool of knowledge, the theory goes, there will be an expert or enthusiast on any subject. If you don't agree with what's written, just click the edit button and amend (but prepare to be challenged if it's not deemed to have a "neutral point of view"). But when anyone can tinker with the content, a question of reliability remains.

With time, its proponents say, each entry will evolve into a comprehensive article.

But this is where some of the recent criticism arises. An established 'expert' may expand or correct an article, but any reader without credentials can dispute or alter that information. Mr Larry Sanger, who set up Wikipedia with Mr Wales in 2000 before they parted ways three years ago, says the site must embrace scholarly input. Dr Sanger now teaches philosophy at Ohio State University, specialising in epistemology, the theory of knowledge.

Though he stops short of claiming that the site is unreliable, he believes librarians, teachers and academics still view it that way - and this perception is just as damning. Unless the site can implement some kind of "deference to experts", he says, the project will be taken over (since the content is free to use).

"I'm quite sure that most of my colleagues, professors and graduate students wouldn't want to spend much time working on Wikipedia, and probably haven't," Dr Sanger says.

He says articles continue to be plagued by 'edit wars', where users change each other's work so the most recent version contains their words.

Mr Wales, who also runs the site's parent foundation, Wikimedia, disagrees. "We're strongly supportive of quality controls," he says. "But if someone can wave their PhD and then everyone has to stop editing articles, then that's not a respect for expertise, that's a respect for credentials."

He points to recently-adopted rules, such as an arbitration committee to resolve editing disputes.

Charges of unreliability are nothing new to those involved in the project and Wikipedians are accustomed to defending its principles.

Dr Sanger's remarks surprised many on account of his role in the site's inception. But his critique echoed similar comments in the offline encyclopaedic world. Late last year in an online article, Mr Robert McHenry, former editor-in-chief of Encyclopaedia Britannica, lambasted the site as unreliable. He claims that Wikipedia's open source model has no place as an information resource.

"Abandoning that model is the only way in which Wikipedia can possibly begin to build a reputation for true usefulness," Mr McHenry claims.

Though users intent on sabotaging an entry are free to do so, the site's administrators are used to dealing with such "vandalism", and frequently ban problem users. But critics say it's the minor details that escape oversight.

In his criticism, Mr McHenry pointed out a problem with the birthdate of a historical figure. But the entry's history page shows it was corrected within a day of his essay's publication.

Dr Sanger says he does not share these views, but he does agree on one point. Both men say the repeated changes don't always improve the quality of articles, and sometimes they are "edited into mediocrity".

Mr Wales points out an aspect of the site that promotes quality content. A group of editors with an established reputation in the community can bestow "featured article" status on high-quality articles.

But once an entry reaches this level it is still open to degradation.

"We are monitoring the best articles to see if they do in fact decline, but as an empirical matter, it's just not true so far," Mr Wales says.

The site's managers say they are working out better ways to test the content, and once they have a "stable", vetted version, they plan to lock it in place.

This standalone resource will be called Wikipedia 1.0, and will form a separate site. Mr Wales says that if this works, he plans to distribute a print version to poorer countries.

As well as the encyclopaedia, the Wikimedia foundation maintains less popular sites using the same model. Most recently, they transferred the principles to journalism, launching an experiment called Wikinews last year.

As Mr Wales observes, the sites will remain as good as the sum of its writers and editors.