Clever girl...

[Geoff_B]

Why not use a capcha system in the "Summon Inglips" style - one word and one not-quite-word? Or was that idea discussed already?

[Keab42]

I've used recaptcha before. Was about as effective as a chocolate teapot.

[Yaxley]

But not nearly as delicious.

[Evil Jim]

How about adding an extra field in user profiles that is to be intentionally left blank during profile customization, to foil the bots that autofill those things? Then we pick 'em off like fish in a barrel.

[Mara Kalat]

I have a background that also covers some captcha and bot behaviour stuff (yay, a use for my AI bachelor), and it's very hard (if not impossible) to make something humans can solve easily and bots can't.

Captchas started out as simple text-as-picture segments without any fancy and have now gotten as silly as diagonally shifting animated gifs with distractor lines phasing through it.
Point is, it doesn't work. All it does is set the hurdle a notch higher for the bots to jump.
So unless your goal is to breed the smartest possible bot (a goal I'd personally not be adverse to, but that's besides the point) none of this would work.

It also won't stop the human registrations handed down to bots either.
Screening or other checking methods where a registration has to be validated are too time intensive for either the registrating users or the mod team.

Where am I going with all this boring rambling?
I'm glad you asked.

Let the userbase monitor for spambots. Right now the behaviour of each bot is roughly this:

• Make it through the registration process
• (Waiting period to lower suspicion)
• Make a large number of topics or replies in relatively quick succession
• Adjust signature to also hold the advertising links (sometimes also set earlier by the stupider ones)
• Get banhammered by our team of vigilant moderators

An elegant solution, that I think isn't that hard within phpBB, is to block posting for users who have a specific number (or ideally proportion) of posts reported.
So if the first 2 posts of a user are reported it is automatically locked out of posting until the reports are handled.
When working with proportions it could just be set to a threshold of 50% or something. No regular user would get half their posts reported (and the theoretical ones that do, well... do we really want those?).

This way each bot only gets as much posts in as the first report, which would cut down the total number of spam posted by, I'd say, around 80% without pressuring or compromising the current authority structure.

Of course bussing is way more important right now, but it might be something to consider once that's over...

Too long, did not read / Bottom line: Set up a system that blocks users from posting if they have too many reported posts, meaning the userbase can actively defang bots themselves.

Mara.

...that was the most boring thing I've written here to date.
It needs a cat picture to compensate.

[Valliac]

Found out what's powering the spambots and how they've evolved.



They can learn and pass our defenses.

EDIT: ONE OF THEM MADE A POLL. WAT.

[Metcarfre]

They're adding avatars and pics to their posts, too.

[JackSlack]

I know that in the end the decision is Paul & Co.'s, but I still think the best weapon we have at LRR is the community: We have a bunch of smart, mature, dependable posters who can be trusted to wield a reasonable amount of power fairly, and we should use that. Identify as many people who want to be allowed to purge spam, and empower them to do that and only that.

We may not be able to stop spammers hitting the site, but we can limit the amount of time the spam is up and make its impact minimal. This solution would not be exclusive, of course, to Mara Kalat's, either.

[Valliac]

Try and say half of these spam thread names. You'll find it's wholly impossible to do so.

[Timelady]

Heh. I'd almost say leave those posts, if they weren't so foul. The irony is palpable.

[thatlaurachick]

Thanks for my morning dose of irony spambot.

[My pseudonym is Ix]

They are out in force this evening...