Context : SpammingThoughtStorms
Response to Graham. (mainly historical interest)
yep, what a FUCKING PAIN. Literally, as I've had to give up rolling back the spam as I was getting RSI twinges. Will finish it off soon. The big question, should I leave the site locked while roll back is in progress?
This attack was nasty. I wonder what the idea of hitting so many pages is? To basically do so much damage that you hurt the morale of the wiki-owner or community so they don't try to fix the BrokenWindows? Are the spammers trying to knock out the community altogether so a wiki just becomes an uninhabited ghost-town, graffitied with GoogleJuice?
Clearly there's a danger for spammers in such a strategy. Too much virulence will simply kill the host. If my wiki ends up like this, why don't I just take it down altogether? So maybe the most virulent spam-bots are counting on some kind of inertia that means hosts will leave desolate vandalized wikis up.
But I'm not going to give up without a fight. Neither on ThoughtStorms nor wiki as a freely accessable writing space. But maybe I have to take note of ClayShirky's injunction in CreatingCommunities : that the special inner group need private channels of communication and tools to defend themselves.
My first, obvious thought, is that this attack would be less serious if I had a script that could automatically roll back all entries from a particular IP address on a particular day. Once identified such a spammer could immedietely be airbrushed from the wiki. So, I'm looking for or looking to write such a tool. (Update : RichardP wrote it)
: Might be easier if you used something like PHPWiki which stores stuff in a database.
: Could then do a periodic DELETE WHERE some condition, in this case the presence of certain characters. (SJ)
That immedietely opens up a larger can of worms. Is this the beginning of the end of the relationship with UseMod? Ideally, I'd write an external tool to handle this, rather than hack the UseMod code itself. But even that would require ScreenScraping. It would be be even better if the wiki engine supported other ways of accessing the data than HTML pages. OddMuse is tidier, and thus easier to modify. but it's basically an obscure fork of UseMod that's been refactored. And in order to port from UseMod to OddMuse I still need to change the database. In which case, why not go somewhere else? Maybe time to take another look at a PythonLanguage wiki code-base.
The second, I still haven't given up on my ideas about "variable geometry HTML" and "immune systems". Thinking how something this could actually work.