Topic detection and tracking? bayesian filtering? are there any successful strategies to fight wikispam anywhere?

Google think they have a solution, which is for blogs and wikis to put rel="nofollow" inside links. The Google web robot will refuse to follow these links, so that they will not boost the linked site's pagerank.

Ironically I can't post a link to the Google page about this here. I get "Comment Submission Error - Your comment submission failed for the following reasons: Your comment could not be submitted due to questionable content: g**gle.com. Please correct the error in the form below, then press Post to post your comment."

(Thanks for letting me know that google.com was being blacklisted; That should be fixed now.)

1. I think rel="nofollow" is hackish, and will not be effective--ever, but certainly not for some time. I expect better from Google.

2. I'm not sure how the nofollow hack is relevant to a wiki, where the spammer is the one who gets to insert the tag.

2. The wiki software needs to process those links, to add the rel attribute (same as in the blog comments, the context of the original google suggestion).

One of the reasons that Google /likes/ wikis is that they're strongly linked to other sites; rel="nofollow" doesn't sound like a particularly appealing when you actually want to let (most of) your users add external links. You could, I suppose, introduce a quarantine period so that links are only marked nofollow for a limited period until someone has checked whether they're spam, then remove them. Which sounds like a good idea, but bear in mind that cliki has been doing it (on a slightly coarser granularity, using META ROBOTS tags) for four months already, with no obvious effect. Chinese spammers can't read the text saying "your effort here is wasted"

My proposal to Google: http://ww.telent.net/diary/2004/10/#26.86194

Their 'piss off and stop bothering us, little boy' response: http://ww.telent.net/diary/2004/10/#26.896