Wikipedia talk:Spellchecking project

A few remarks: FvdP 18:36 Oct 18, 2002 (UTC)
 * is Ed Poor's section reserved to Ed Poor ? Since it has a format I like, I would be glad to use it too
 * spelling errors detection should be done automatically if possible. That is, a list of common misspellings should be periodically searched for in the articles, by an automatic "daemon", and the results be published for humans to correct (I'm wary of automatic correction).
 * in the meanwhile, what about using Ed Poor's list with "extinct" errors classified by date of last verification ? Errors are never really extinct, they are expected to eventually reappear ! So they should be rechecked periodically. Perhaps we could separate them in "common" and "uncommon" errors (according to how many of them are spotted on the same period of time; this information should be stored on the page as well.)


 * I agree with all of the above, and I'm going to take off the "Ed Poor" heading. --Ed Poor

(Prefatory disclaimer: I'll attempt to word this as diplomatically as possible, but sometimes I'm not known for diplomacy, so please don't take it the wrong way.  I like the spellchecking project.  It's a great idea, and a very worthwhile application of system resources.  The project definitely needs to continue in some form.  I don't like spelling errors in the Wikipedia any more than anyone else.)

However, I think the current implementation of the project is seriously flawed.

Articles which contain spelling errors almost always contain other problems, related to grammar, content, etc. A good example of this is the article on Randomized controlled trial. This article, while containing sound factual material, nevertheless needed major grammatical and stylistic work which, despite its being visited three or more times for spelling correction, was never done.

I realize that we are all limited in the time we're able to spend on articles, but wish there was some mechanism by which we could mark an article that needs further work, or at least skim it to see if it does need further work.

I left the article on Janissaries alone for three reasons. One, it'll take some research to get straightened out, and again, we all have only a certain amount of time in a day to work on Wikipedia. Second, if a spell-check were run on the article, and all the spelling errors were fixed, it's possible that it would be quite some time before someone else stumbled onto it and discovered it has major problems.

Third, it stands as a perfect example of what I feel is the problem with the spellchecking project as it currently stands. Doesn't anyone even bother to look?

Sorry for ranting, as I said I don't mean it to be taken personally, or to disparage the spellchecking project on the whole. I'm done now. :)

Hephaestos 22:36 Jan 12, 2003 (UTC)


 * When I correct spelling in articles I do skim the article to make sure it's not horribly written or completely absent of links to other articles. -- goatasaur


 * As do I... it's rare that I make a spelling correction without hitting some formatting, linking, minor rephrasing or other wikification. Though, since the spellcheck feature has been unavailable, obviously I can only check errors as I spot them. You make a good point, though; an article filled with spelling errors is likely to have other problems too.


 * What we need is a List of commonly incorrect factual claims or List of common POV statements that could be matched by some kind of fuzzy-logic to articles which employ such techniques :) That may not happen until Wikipedia becomes sentient and takes over the world, though. There's only so much we can do... maybe there should be some kind of rating system, though. Whenever I edit an article that I am either not qualified or motivated to verify, correct, or otherwise fix, I make a comment in the edit summary that it needs more work (so at least future editors can check the history and see that comment). But it'd be great if there were a voting system whereby we could indicate how much work we think an article needs, regardless of easily-spotted errors like spelling. -- Wapcaplet 00:06 9 Jun 2003 (UTC)