[Bug-wget] The Wget Wgiki, and spam issues

[Micah Cowan]

Hey folks,

So the Wget Wgiki is still alive and kicking, but dealing with spam is
getting out of hand.

Moin anti-spam works with a combination of global blocklisting, and
"textchas" (text questions designed to be easy for humans, hard for
robots).

The problem is that eventually a human gets involved, checks out the
questions, and then feeds the correct answer to his bot farm. Then all
of a sudden spam is being added to the Wiki at a heavy rate. I've got
to go find what textcha they've been getting through with, and remove
it (and of course the spam, and the users that were created to make
the spam).

Well, at this point I've gone through quite a few textchas, and I'm at
the point where I have trouble coming up with textchas that will keep
out the bots, but not real wget users.

Does anyone know of customizations to Moin Moin (the wiki engine the
Wgiki uses) that greatly improve spam-blocking? I'd certainly be
interested in effective capchas - textchas are used by Moin to avoid
discrimination against the vision/audio impaired, but I'd be fine with
having people who can't use capchas contact me or this list directly
and ask for an account if it made it easier to keep out the spammers.

Alternatively, it may be time to move to a different wiki engine that
already has more effective anti-spam measures in place. MediaWiki
wouldn't be my first choice, but it might be an option (don't know
anything about its anti-spam measures, though).

Another possibility would be to disable new user additions completely,
and just have myself (and probably another admin or two - Giuseppe?)
directly approve new users to the current wiki. That's probably by far
the easiest for me to implement, but wanted to get a feel for what
others thought about making the wiki no longer publicly editable.

The wiki gets low edits (particularly, non-spam ones), but it's
definitely still helpful for new contributors to Wget to be able to
get on-board and update the Wgiki when appropriate. And it'd really
suck if someone contacted me to get a user account, and I missed it
because it fell into my spam box or something. :-/

For reference, here's all the textchas that have fallen to the
spammers. There is currently _one_ textcha still in operation. If that
falls, I'll switch to an unanswerable textcha (no one can edit) until
I can figure out a better long-term solution.

# u'What program is this wiki about?': ur'wget',
#u'Spell Wget backwards:': ur'tegw',
# u'WMD stands for Weapon of Mass ______:': ur'destruction', cashloans11 got
#u'What is the default TCP port for the HTTP protocol (spell it out,
DO NOT use digits):': ur'eighty',
#u'What is the last name of the main characters from The Hobbit and
Lord of the Rings?': ur'baggins?',
# u'Name an internet protocol supported by wget:': ur'https?|ftp',
#u'What\'s the acronym of the software license used by wget?': ur'gpl',
#u'The wget command is used on the c_____nd l_ne:': ur'command line',
#u'Wget is useful for d_______g web pages:': ur'downloading',
#u'What project, founded by the Free Software Foundation, is Wget a
part of?': ur'(the )?gnu( project)?',
#u'What is the long form of wget\'s -A option?': ur'(--)?accept',

(Note, PLEASE don't respond to the list with more ideas for textchas.
I'll entertain suggestions by private messages, but keep in mind that
this still allows the possibility for spammers to submit ideas to me
;) ... it's really not a long-term solution.)

-mjc