Boost logo

Boost :

From: Jeff Garland (jeff_at_[hidden])
Date: 2006-06-18 22:25:38

David Abrahams wrote:
> Jeff Garland <jeff_at_[hidden]> writes:
>> David Abrahams wrote:
>>> Jeff Garland <jeff_at_[hidden]> writes:
>> Two reasons. That requires a software upgrade including data conversion
>> to new wiki software. Second, based on my discussion with other wiki
>> administrators it doesn't stop spammers -- they just register with a
>> free email address and off they go.
> Of course a few will do that. But isn't most spam done by automatic
> webcrawling software that seeks out open wikis and forums? As long as
> those abound, fewer spammers will bother with the protected one,
> right?

Nope, from what I understand the email registration test just doesn't
work at all unless you want to put a human in the loop on all approvals,
which prevents the casual update. Remember, we are dealing with folks
that have hacked hundreds of machines, so a bunch of throwaway email
addresses isn't a problem for them. A typical 200 page spam attack is
done from 15-20 different IP addresses with each one spamming at about 1
page every 5 minutes. The reason they do this slowly is that many
wiki's added a feature to prevent robo spamming by only allowing a slow
number of changes per ip per unit time. So the spammers adapted...

> That must also be why image verification is so widely used.
> If spammers were content to deface these sites by hand and use
> arbitrary amounts of subterfuge to do so, pretty much nothing could be
> effective against them.

Content banning is the most effective and that's what we currently do
(see other mail). One thing I could do better is keeping up to date
with some blacklists, but I just checked the main one I know of
( and having the latest version of their database
wouldn't have prevented the last 3 attacks.

>> Bottom line is that today's systeam takes me about 10 minutes per
>> day. The only bad part is that it sometimes takes a bit more time
>> to roll back if I don't happen to be online.
> As long as you don't mind doing the work, I appreciate it, and of
> course it's okay with me.

In my ideal world we would upgrade the software and it would enable a
group of moderators to trivially manage/revert/stop spam. We'll get
there at some point, but the solution we have now is working well enough
and I'm busy enough that I don't plan on pursuing this for awhile.


Boost list run by bdawes at, gregod at, cpdaniel at, john at