Wikis are hit only accidently by spambots (which target blog and comment forms mainly). It is not necessary to contribute to the locked web with forced user account registrations, and especially CAPTCHAs should be seen as the last resort only.
Like the author mentioned, it is pretty simple to stop the currently rather stupid bots. It often suffices to introduce a forced save/submit delay or add a checkbox ala "[x] I'm no spambot". A very effective mechanism is also counting the added external links and rejecting the whole submission if too many links were put in (often seen in chinese spam).
Bayesian filters may become necessary, but aren't currently. And if a general distributed blocking system emerges, it will be a bit more effective by not blocking IP addresses only, but ISPs and web space providers or via reverse WHOIS lookups (block the spammers, not their pages).
Mass edits can also easily be reverted, so cleaning up a spammed Wiki is only difficult if you use the wrong software or database scheme.
The rel=nofollow is mostly non-effective, because spammers don't check for it. And then it is merely a protective invention by Google, who created the current mess in the first place by throwing their PageRank at the Web (and now tries to persuade other people to fix it). I don't even clean my SandBox anymore - bogus and misplaced links are entirely Googles problem.
The only real problem with WikiSpam are injections and manipulations of existing hyperlinks. But as this often happens manually, it can be easily detected and cleaned up by typical contributors and visitors.
Learn about Linux
Download Linux
Get Linux help
good enough solutions in place
Posted by: vtre17 on June 29, 2005 12:07 AMLike the author mentioned, it is pretty simple to stop the currently rather stupid bots. It often suffices to introduce a forced save/submit delay or add a checkbox ala "[x] I'm no spambot". A very effective mechanism is also counting the added external links and rejecting the whole submission if too many links were put in (often seen in chinese spam).
Bayesian filters may become necessary, but aren't currently. And if a general distributed blocking system emerges, it will be a bit more effective by not blocking IP addresses only, but ISPs and web space providers or via reverse WHOIS lookups (block the spammers, not their pages).
Mass edits can also easily be reverted, so cleaning up a spammed Wiki is only difficult if you use the wrong software or database scheme.
The rel=nofollow is mostly non-effective, because spammers don't check for it. And then it is merely a protective invention by Google, who created the current mess in the first place by throwing their PageRank at the Web (and now tries to persuade other people to fix it). I don't even clean my SandBox anymore - bogus and misplaced links are entirely Googles problem.
The only real problem with WikiSpam are injections and manipulations of existing hyperlinks. But as this often happens manually, it can be easily detected and cleaned up by typical contributors and visitors.
#