[a_small_island]

>"2. Being able to identify troublemakers in a community (such as a forum or a social networking site). I'm sure a lot of administrators would love to know if that suspicious looking new guy is the alias of a banned troll from a few weeks back (posting through a proxy server)."

Orwellian.

[noonespecial]

That's especially troubling because the easiest and cheapest thing for any commercial service to do is to ban forever and without appeal any user that presents a hint of trouble to the algorithm. If these services have near monopolies, or share data with all of the other services like CLUE does for insurance... Its like a digital death sentence.

Ask anyone who's ever lost an ebay or a google account to an algorithmic burp and was essentially banned for life without appeal or even human oversight.

[zghst]

What about Russian trolls that have infected the internet, trying to manipulate public opinion or just collect data on regular citizens? Or Correct The Record? Or just about any other mass propaganda campaign? This would be very helpful

[CM30]

So, what do you when a troll persistently and constantly attacks your community website?

As in, flames everyone to a crisp, posts as much porn as possible, tries to incite a civil war between a few members that might not be on good terms with each other or the staff and registers hundreds of accounts, some of which stay semi dormant until they strike?

Because that can happen very easily online, especially if you get the ire of someone with a lot of free time and very few morals. Or if your site ends up at war with a troll site/gets raided by 4chan.

Do you avoid the hassle now, or wait until the situation blows up and half the site is now in the middle of it?

[abstractbeliefs]

Do we really need machine learning deanonymisation tools to identify trolls so unsubtle and obvious?

I voted to release the tool, but you've set up this false scenario with the intention of knocking it down easily and discrediting the opposing argument.

[DeadHitchhiker]

Ban behavior, not people. You can convert non-productive people to productive people, most of them only want to be noticed or accepted and past a certain point there is only so much you can do to block anyone anyway. Stylometric analysis would just be another simple hurdle to cross for a persistent person.

The idea that this tool would be useful for community management is terrible.

[ubernostrum]

You can convert non-productive people to productive people, most of them only want to be noticed or accepted

I'm a moderator of multiple online spaces.

A few months ago, in one of them, a user got too heated and started flinging insults at someone else. As was standard policy for the place where it was occurring, I issued the user a ban of a few days (enforced cooling-off) and pointed to our guidelines on how to behave.

This user then proceeded, over a period of months, to continually harass me, send me increasingly graphic threats, and try to track me down in real life.

Pray tell, how exactly would you go about "converting" such a person to be productive? I come to you since you are apparently quite the expert on it, or else you wouldn't be giving out advice to just "convert" people.

[dekimir]

In my day, we'd use moderators: put the forum/list in a mode where every new post goes through a human review first. After a while, things calm down, and you revert to the anything-goes default.

I don't know how well that would work for high-traffic forums of today, but it can scale pretty easily by employing many moderators.

http://www.netlingo.com/word/moderated-mailing-list.php

[jessaustin]

ISTM that any site that is capable of controlling spam, is also capable of dealing with trolls. Sure spam is more repetitive than trolls, so it's easier to notice automatically, but there is also much more spam, so chances are that at least some will get through. If you have moderators authorized to zap spam, let them zap trollage as well.

[Zuider]

It would certainly run the risk of false positives.

[CM30]

This is a good point though. Even a 5% chance of it misidentifying someone as a troublemaker would be a problem for an online community site, and it's likely the actual chance is a fair bit higher than that (because it's likely not been tested on a particularly large scale in this context).