Feed the trolls

August 4, 2016 1:53 PM

Troll hunters: the Twitterbots that fight against online abuse (New Scientist, h/t TrollBusters)

TrollBusters previously

posted by Johnny Wallflower (6 comments total)

It's interesting to contrast this with the "Chatbots of the future will automatically harass you for dollars a day" articles. Those assume that near-Turing test passing chatbots (whose design will likely require resource intensive neural networks) would be good at researching and bothering a specific person for small amounts of computer time from Amazon or anyone with a bunch of idle graphics cards sitting around.

This, however, is getting human trolls to split their attention between a bunch of targets. And building simple taunts using random adjectives is easier than trying to create a strong series of personal attacks on somebody. It's sort of a Sybil attack.

However, there's two big problems I see. Unless the owners of these bots let them age and have them post realistic and banal content before they're found out, trolls will learn to recognize them. And while the reduction in slurs datapoint is promising, I think some trolls get a narcissistic charge out of trolling, and more targets wouldn't necessarily discourage the troll so much as waste a bit of their time. Maybe spending a bunch of energy trying to fluster a rando and realizing you were essentially yelling at an ATM menu is demoralizing, though?
posted by mccarty.tim at 3:58 PM on August 4, 2016

The future is going to be a planet devoid of humans but an internet still active with armies of angry bots doing endless battle, isn't it? Finally, the noble space messiahs will come, shake their heads over our sorry end, then discover to their horror, that their AI web has been corrupted by billions of autotrolls, and the last hope of the galaxy will die out. That is our legacy, mark my words.
posted by GenjiandProust at 4:37 PM on August 4, 2016 [7 favorites]

Munger wanted to target misogyny as well as racism but gave up when he found words like “bitch” and “whore” were so widespread that it was impossible to distinguish genuine abuse from casual chat.

Goddammit, internet.
posted by a car full of lions at 5:49 PM on August 4, 2016 [5 favorites]

The future is going to be a planet devoid of humans but an internet still active with armies of angry bots doing endless battle, isn't it?

Sorry, a bit of bad news: This is that future. We are those bots. Polite interactions are just noise in the algorithms.
posted by ArmandoAkimbo at 7:01 PM on August 4, 2016 [2 favorites]

"create bots that would admonish people who tweeted racist comments"

Is admonishing the only or even the most effective approach for all the abusers? I was intrigued when someone set a psychology bot onto some gamergate sites a while ago, and it had the intriguing effect of people getting into introspective conversations, often saying things like "this is the first time someone has actually listened to what I had to say".

Maybe he should experiment with some more subtle interactions and see if he gets better results.
posted by eye of newt at 7:30 PM on August 4, 2016 [4 favorites]

The Communications of the ACM had a front page article on social bots in July, which examines positive and negative aspects of bots in social media from an academic perspective.
posted by katta at 11:21 PM on August 4, 2016 [1 favorite]