More Anti-Spam Stuff
Lots of people have sent their condolences about the wiki spam, thanks.
I've been doing some research and fortunately there are a few options available:
Make the wiki only writable by people who have logged in. I did this initially as I figured out what my options were, but I had no intention of leaving it that way. If that really becomes my only option to keep the spam under control, then I might just ditch the wiki and move to a blog and see how that goes (bring on comment spam ... <sigh>).
DarrinEden told me about a program written by Richard Parker called WikiMinion. WikiMinion is a script which monitors the changes which occur on your wiki and if it thinks they are spam it automatically reverts it. Cool idea and I may end up resorting to it, but I'll use it as a last resort.
Moin has a black list for IP address and host names which are repeat offenders. I'm not a big fan of black lists (see SpamBlacklistsConsideredHarmful) but at least it's a method of control.
The primary method I've chosen uses one of the new features in MoinMoin 1.3, what they call their AntiSpamGlobalSolution. It's a grandiose claim but seems to be the best, and most wiki'ish, option available. It works by making the wiki periodically mirror the remote BadContent page into a local BadContent page. Then anytime someone tries to save a page which contains one of the patterns listed it gives them an error saying that it won't save the page because it contains such and such a pattern.
After all of this hopefully we are back with true wiki style functionality (three cheers for SoftSecurity) while still being protected from the majority of the spammers. Big thanks to the MoinMoin crew for all the work which made, and keeps, this wiki possible.
Well see how well it actually works.
Comments
I wonder if Vipul's Razor could be extended to include wiki edits that are considered spam. --DonPark
- Hrm, that's a pretty cool idea. I didn't check closely but I do think that most of the pages that got overwritten were overwritten with the same spam. Hrm, maybe an even better idea would be to do bayesian style spam detection based on what's been added to the page (eg. the most recent diff). -- Adam