Slashdot Mirror
Researchers Developing An Algorithm That Can Detect Internet Trolls
An anonymous reader writes Researchers at Cornell University claim to be able to identify a forum or comment-thread troll within the first ten posts after the user joins with more than 80% accuracy, leading the way to the possibility of methods to automatically ban persistently anti-social posters. The study observed 10,000 new users at cnn.com, breitbart.com and ign.com, and characterizes an FBU (Future Banned User) as entering a new community with below-average literacy or communications skill, and that the low standard is likely to drop shortly before a permanent ban. It also observes that higher rates of community intolerance are likely to foster the anti-social behavior and speed the ban.
Disqus, and the comment section at The Atlantic.
Best Slashdot Co
Automated censorship. Eh, saves us the trouble, I guess
“He’s not deformed, he’s just drunk!”
Anybody who tells the truth that the scum in power don't want you to hear, apparently...
Trolls are usually above average literacy and trying to skilfully cause a fight. It's easy to identify "illiterate" people and humans are way too quick to judge someone who cannot spell as having nothing to contribute or (worse) malicious, but these are not trolls. This is just another classist meme where the person is judged positively by the overcomplexity of their language and convolution of their sentences, as this must mean they have been educamated right.
BTW I went to a £30k/year British boarding school, so I have no axe to grind, nor insecurity about describing things as they are.
I don't want to talk to you no more, you empty-headed animal food trough wiper! I fart in your general direction! Your mother was a hamster and your father smelt of elderberries!
Get free satoshi (Bitcoin) and Dogecoins
within the first ten posts after the user joins
So, this algorithm only needs nine more posts than a troll will actually make per throwaway account, then?
That's some mighty fine police work there, Lou!
What Cornell's researchers found is way to id posters who are likely to get banned.
... at the term "Future Banned User" (FBU).
Isn't that just a nice way of saying "somebody our algorithm flags as a complete asshat"
Your mistake is in using the "classic" definition of "troll" - somebody who sets out to deliberately cause fights on a forum. Trawl through the archives of Slashdot and you will find many instances of this kind of trolling - and yes, the people doing it are often highly literate (and, when they do it right, sometimes very funny with hindsight).
But the term "trolling" has gone political these days and is routinely used to describe any form of online behaviour that the speaker doesn't approve of. So everything from outright criminal behaviour (eg. threats of immediate violence) at one end of the scale through to disagreeing with a forum's established groupthink (however respectfully) at the other.
And yes, it has become a favourite term of the intellectually insecure, whenever they want to shout down an opposing point of view without engaging with it. In fact, conflating those two extremes I mention above under the same term is outright beneficial for the easily offended, as it allows them to group polite dissenters together with the mouth-foaming loons.
Internet pre-crime.
There goes Gawker.
It also observes that higher rates of community intolerance are likely to foster the anti-social behavior and speed the ban.
If automated an intolerant core could try to get users expressing opinions that they don't like banned. The fact that they are subjected to intolerance would make the algorithm more likely to ban them.
It might be useful to inform an admin to look at suspicious postings, especially if they can get the accuracy higher. BUT I hope no one uses such algorithms to automatically stop suspected trolls. This can only lead to unforeseen consequences and stifling of free speech (unless of course stifling is not an unforeseen consequence, but an intended one).
Many Slashdotters already complain about the Lameness-Filter, this has the potential to be a hundred times worse.
The technology will of course be developed, let us hope its impact isn’t too negative.
In a somewhat related note, have you noticed how the automated answering at phone centers is getting more aggressive keeping you from a real representative and wasting huge amounts of your time when it doesn’t know how to process your query? Even hanging up on you when your issue is not resolved. My last experience with Verizon was a nightmare in this respect last time dealing with a technical problem with our phone. The more these things can be automated, the more they will – customer friendly or not </vent spleen>
Letter To Iran
int troll;
troll = found("goatse.cx");
OK. There's your algorithm.
I can't wait til the anti-bullying crowd lobby for something like this wherever their children might be on the internet. Sounds like it is early days though, 20% false positives is pretty darn high.
The article defines a troll as someone who has been banned from an online group.
You can be banned from a website such as redstate for being an Obama supporter. People are often banned from websites solely for having minority viewpoints.
If I were to rank the publicly-accessible online forums I participate in these days, from most civil to least civil, Slashdot would be top of the pack by a long, long way. Seriously, that's how bad it is now.
The unholy trinity of culture wars, console wars and overbearing admins have ruined many other discussion sites that were perfectly good 3 years or so ago.
Researchers have detected researchers detecting an algorithm detecting researchers researching.
We play the game with the bravery of being out of range
What they identify isn't people who "troll", it's people who get mobbed and ostracized by a community. There's a big difference between the two. That's not a question of "false positives", it's a question of whether people lose themselves completely in group think.
Of course, in practice, there is little chance this will actually go anywhere. Although content creators and ideologically biased readers frequently denounce as "trolss" anybody who disagrees with them, sites actually like controversy because it increases ad impressions. That's why sites continue to use systems like Disqus, which have rather ineffective moderation.
Does the algorithm account for the fact that the Troll designation is applied by some specific person who (a) has mod points, (b) strongly disagrees with a given post and (c) is in many cases part of a group who is looking for antagonists to some cause that group really believes in.
If (internet) then troll_present = true;
Done, just that easy.
Lost at C:>. Found at C.
https://xkcd.com/810/
Seems relevant.
Looks like my days on the internet are numbered.
I have a personal rule not to read comment sections. There are a small number of exceptions, but, in general, whenever I ignore the rule and browse the comments, I invariably encounter some insanely stupid comments that make me want to bang my head into my desk repeatedly. Too many people seem to be able to operate their brain or their fingers/mouth, but not both at once.
My sci-fi novel, Ghost Thief, is now available from Amazon.com.
The word troll is a pointless word which is misused by people who mainly want to villify those who disagree with them, and excuse for people who do not want anyone else to be able to express opinions except for the ones they approve, to censor anyone elses opinions they do not like. Thus, the marking in almost all cases is abused and has no real purpose except for censorship. Obviously, since a message board should be a place for discussion and expressing of differing views and opinions, such is contrary to the purpose of message boards to begin with, to express ones views and to debate subjects.
The fact is, expressing a view someone else disagree with is not something we should censor, and the tr*** accusation is just an excuse for censorship. As long as the poster honestly believes in what they are posting, its not a tro**, their are posting their view to express their position for the sake of the issue itself, rather than to annoy anyone. Maybe, a tr*** might be someone who posts things they do not agree with for the sole purpose to annoy. However, since it is impossible for anyone to know whether or not someone posting a message honestly believes in what they say, it is impossible to determine if a message is a tr***, or not. it is also impossible to know if someone is posting a view just because they are interested in a subject and have a view on it, rather than trying to annoy anyone.
The fact is, if someone is annoyed by something, the person responsible for being annoyed is the person who is annoyed, its all in the eye of the beholder, some people will agree with something and others will disagree, you have to allow for a difference of opinions and views. It is always the case that someone will disagree with someone else says, it does not mean that the message was posted with the sole intent to annoy, but the reader of a message may still misconstrue or assume that even though it is impossible for them to truly know that. It is okay and important for people to be able to post messages they know will annoy others, because, anything can annoy anyone, its impossible to post a view or position on anything if one has to fear annoying someone.
The tr*** thing could only apply to messages written with the sole intent to annoy, But as I said, its impossible for anyone to know if that was the sole intent, to be the sole intent, the person would have to not honestly believe in what they say, otherwise they are posting because they believe in what they say and think that its important.
That is why the marking on a message cannot be used legitimately and fairly, there is impossible for anyone to know if a message is a tr8**. Thats why, we should remove the marking from messaging and bullitin board systems. As I said before, in 100% of cases the marking is abused, it cannot be used in any proper, fair way, because it is a fundamentally flawed feature.
It would be best policy on these matters is that bullitin boards should have a rule against computer generated and mass posted advertising, but thats about it.
A persistant "Whoosh"ing noise seems to be getting worse accompanying Slashdot comments. I wonder what it is?
Have gnu, will travel.
I second what you've said!
Linux is for people who don't mind RTFM.
The 80% figure, which is the AUC (area under the curve), refers to threshold tuning. In order to make that usable in the Real World, you'd have to crank it so that it has nearly zero false positives (and thus very few detected trolls) or else you'd have to make it flag posts non-fatally, perhaps with nearly impossible captchas, which immediately defeats its anti-troll utility (not to mention angering all of the falsely identified trolls!).
The article, like the paper itself, ends on this note:
Use my userscript to add story images to Slashdot. There's no going back.
and volunteer to help test. We have a steady stream of trolls available for review, a truly endless supply.
spacefem.com
So, posters on message boards deemed "anti-social" or that have views that are not tolerated by the community are now the definition of troll? Wow, that's a good way to make sure opposing viewpoints never get heard. The "algorithm" will just drop any message that goes against the "party line".
I'd imagine there are plenty of places where if you stand up for your individual rights and privacy you'd be marked a subversive and the community wouldn't tolerate your presence. How about speaking of the value of equal civil rights for all humans on the kkk.org forums? I'm pretty sure the community would be pretty intolerant to your viewpoint there!
When did the definition of "troll" become someone who has a viewpoint that you don't necessarily want to hear? A troll, for those who don't know, is someone who legitimately goes out of his or her way to get a rise out of someone just for laughs. Someone who is passionate about an issue that may go against popular opinion or a political narrative doesn't make them a troll. Not having the best grammar and writing skills doesn't make you a troll.
This is purely a push by establishment websites to automate the moderators who go through and do the wholesale deletion of any and all viewpoints that go against the "party line" with robots so in the case of any kind of populist backlash against the predetermined narrative and status quo, they won't have to rely on humans that (possibly) have critical thinking skills going outside the box.
Me too!
I can see, how this may defeat (ab)users trolling for fun and not suspecting automated detection before it hits them (though, with only 80% accuracy, I dread the thought of the methods expanding out of the virtual realm).
But what about people "trolling" professionally — paid and/or otherwise compelled into it by a state or corporate actor pretending there to exist some kind of "grass-roots" movement? How would it deal with thousands of fake accounts mounting a coordinated assault, posting (while "liking" and "following" each other)?
Some times you may be able to catch accounts posting identical things at the same exact time (and ban them all in bulk), but Russians seem to have fixed that bug in their bots now...
This is turning into another battle like that, in which spammers have fought the best Information Technology minds into a standstill. I doubt, progress against forum-spammers will be much better than that — not when mere technology, however clever, is up against interests of a reasonably powerful state.
In Soviet Washington the swamp drains you.
... something like this:
int is_troll( const char* username ) {
if( !whitelisted(username) ) {
return 1;
}
}
Log in or piss off.
This won't work nearly as well as the authors expect. The moment such system gains adoption, the rules will change and anti-detection and algorithm poisoning techniques will be adopted. For example, proposed approach would likely be completely defeated by first making 10 "constructive" FAQ copy-paste postings. Also, spam is much easier to detect than trolling, since spam is not unique. Still it took years and complicated spam-detecting analytical algorithms to reduce it to manageable levels.
I'm secretly hoping the 80% success rate indicates that everyone else has a 1 in 5 chance of getting randomly banned, troll or not.
I feel this would noticeably improve many online communities.
But I was really just trying to disagree with someone's point of view.
Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
article = new nonsensefilledstory();
article.addStrife();
article.addContraversy();
article.stoketribalisim();
article.allowAnonymousComments(true);
stack_of_trolls *users = article.create();
forall users as user (
if (user.isTroll() == false && user.respondsToTrolls() == true)
(globalBanList.addUser(user));
)
Unfortunately, many people think that if you express a different viewpoint or opinion than the masses that you're trying to start an argument or a fight. Why is society so hell-bent on crushing dissenting opinions? And not merely silencing them, but villifying them?
I've often been tagged as "trolling" because I don't agree with the crowd. If you knew me personally, you'd know very well that I'm not trying to start a fight, just expressing my opinion. Just because it is not the popular viewpoint doesn't mean my views aren't valid.
Here on Slashdot, I often see people flagged as being trolls just because they don't follow the masses. You'd think a site full of outcasts and oddballs like programmers and technologists would be more accepting of alternative views, but the exact opposite seems to be the case.
I do not fail; I succeed at finding out what does not work.
And yet, quite routinely, I have seen people getting flagged as trolls right here on slashdot, myself included, for doing no less and no more than precisely what you are describing here.
Clearly what matters is not so much what actually makes one a troll as much as whether other people, particularly people with power, inflluence, or control actually *believe* a person is a troll... Any non-trollish intent of the poster is entirely irrelevant.
File under 'M' for 'Manic ranting'
Police officers arrested Mr Weston, mid-speech, for failing to comply with their request to move on under the powers of a dispersal order made against him.
He was further arrested on suspicion of religious or racial harrassment.
So yes, he was arrested for failing to comply with a police order and yes, he was arrested for his speech as well.
the rest of the white race
Oops, what a giveaway, you racist twat.
To have a right to do a thing is not at all the same as to be right in doing it
You call it Troll; They call it Advertisement; Somebody else will call it Spam;
Casteism
aka Flamer? http://www.newser.com/story/19... http://www.nytimes.com/2014/08...