Though I would second the "fix it" philosophy, it can be frustrating to fix an article only to later come back and find that somebody who thought he knew better than you, but actually didn't, has "corrected" it back.
I will add, though, that the mathematical articles I've looked at did not suffer from this problem, in general. There are occasionl exceptions with "hot button" articles, like the one on the equality
1=0.999..., which every now and then falls prey to:
- idiots who think they know better than standard mathematics;
- mathematical purists who confuse the issue by getting all stringent and pædantic with the oh-so-important issue of "what really constitutes a true proof".
That said, this particular article is fairly good and accurate, on a good day. Even on a bad day, it's only a small fraction of the article that usually turns into gibberish or agenda-pushing, and most of the time that's easy to spot (for the informed eye).
Recently, Wikipedia has been tightening its standards of verifiability, which has been a mixed blessing. On one hand, it encourages more throrough referencing. On the other hand, sometimes whole paragraphs that were correct but happened to not be referenced are deleted and replaced with less accurate material, just because some editor found one inferior source that supports the latter version. (This refers to articles of non-mathematical content, though).
I do think, though, that for some, really, really controversial and complicated topics, Wikipedia is only good for the links to other sites. I'm thinking of stuff like the article on human races, which is currently an unintelligible and often misleading mess, because in the effort to be "balanced" the editors end up allowing the article to become a patchwork of opposed claims, with no effort to weight them against each other.