Analyzing Tweets To Identify Psychopaths

nonprofiteer writes "Researchers presenting at Defcon next week have developed a psychopathy prediction model for Twitter. It analyzes linguistic tells to rate users' levels of narcissism, machiavellianism and other similarities to Patrick Bateman. 'The FBI could use this to flag potential wrongdoers, but I think it's much more compelling for psychologists to use to understand large communities of people,' says Chris Sumner of the Online Privacy Foundation. Some of the Twitter clues: Curse words. Angry responses to other people, including swearing and use of the word "hate." Using the word "we." Using periods. Using filler words such as 'blah' and 'I mean' and 'um.' So, um, yeah."

Comments

There's a rumor going around

[Dunbal]

I hear that if you measure their skulls you can also spot murderers. I will call this science phrenology!

Re:There's a rumor going around

[Anonymous Coward]

This article is absolutely ridiculous and I pray it is shot down my mainstream psychologists because it's not fair at all. I am a law-abiding citizen who works a full-time job and spends much of my downtime educating myself and hanging out with friends. I have hobbies, I have a beautiful family, and I am a very good person. Many of the things in this article have applied to me throughout my life and I am disgusted that they would even begin their psychopath search in this manner.

Fucking incredible.

Re:There's a rumor going around

[Charliemopps]

No, this system tries to predict the future by reading posts on twitter. Could a certain percentage of people who commit violent crime share the common trait of psychopathy? Certainly. Do all? No. In fact, a rather large percentage of the population are psychopathic, yet lead normal, non-violent lives. Far more people that are not psychopathic commit crimes than those with psychopathy.

The end result of this research is rather clear: Watch what you say on the internet, the FBI might flag you. And that's a far more dangerous threat than a few psychopaths walking around. If you look at the biggest mass murders in human history, every single one of them was a government official. Think about that for a while.

Re:There's a rumor going around

[ivoras]

Of course, the real insight is Reverse Phrenology! Kicking people in the head to make them better people!

Re:There's a rumor going around

[Oligonicella]

"whilst this system classifies people based on their actions and ideas"

No it doesn't. There's no follow up going on to match the words to people's subsequent actions. And please, the use of periods? What does that indicate? Grammar? How about my immediate excessive use of the question mark?

"(content of their character)."
Even further horseshit. There's no correlation betwixt grammar and character.

Asimov had it wrong, there is no such math.

Re:There's a rumor going around

[fatphil]

> Do all? No.

That's actually not the failure that makes this useless. The failure is that the dangerous psychos are such a small minority of the population that even if you throw away 99% (and I'm being generous) of the relatively normals, you're still left with a huge pool of normals with just a few pschos in it. You've not made the problem of actually finding the psychos significantly easier. Now 99.99% of the people who seem suspicious are dangerous rather than 99.9999%.

This is exactly the same flaw as the face scanner tech they were proposing at airports which Bruce Schneier et al. ripped to pieces when it was first proposed.

Re:There's a rumor going around

[AngryDeuce]

But how many people have an online persona that is vastly different than they're real life one? I have friends that act like they're Che Guevara on their social media postings that, in real life, will bitch and cry and complain for an hour if the person at Starbucks takes too long to prepare their non-fat mochaccinolatte, and don't even get me started on all the people that bitch and complain about "freeloaders" and "people living on the government teat" that I know for a fact are collecting government services themselves or have benefited from them in the past....that's another particularly LOL-worthy demographic these days. And then, of course, there's the "I'm ME and I don't care if anyone likes my opinions or not!" posts I used to see all the time, written by people that are, without a hint of irony, obviously searching for positive reinforcement from their social group.

The number of people I know that are actually 'themselves' on the internet, and not some idealized version of themselves that they invented to be popular, is quite slim. I don't bother with social media anymore because I got tired of dealing with cartoon characters and hypocrites. If people actually had to be themselves on the internet, warts and all, I bet the number of social media junkies out there would plummet overnight...

For those reasons, I fail to see the value in this...

Re:There's a rumor going around

[hey!]

Well, the test of something's value shouldn't be whether it strikes you as silly.

The difference between this and phrenology is that phrenology is based on a fundamentally flawed assumption: that the external shape of the skull is correlated strongly to the size of medium scale anatomical structures of the brain. It's easy enough to debunk phrenology by looking at few skulls and comparing the interior to the exterior. In contrast it's quite plausible that a personality trait would be *correlated* in some way to answers on a questionnaire.

So what we need to know is whether the methods and assumptions of the researchers stand up to critical scrutiny, which is something more than making a few hand-waving analogies.

Presuming this idea stands up to scrutiny, the problem with it is that the people who might want to use it (e.g. law enforcement) don't have enough mathematical education to be trusted to use it correctly. Most people treat tests like magical oracles. Cops aren't the only ones who do this. My daughter's pediatrician wanted a certain uncomfortable test done on her. I asked him what the prevalence of the condition being tested for was in the general population. He didn't know. I asked him what the false positive and false negative rates on the test were. He didn't know. In other words, he had none of the information needed to interpret the results of the test, and even if he had that information he lacked the mathematical skills to do it.

Proof of anything requires developing independent lines of evidence; in the hands of the mathematically uneducated even a far more reliable test than this one is bound to be trouble. Drug screening tests are reliable enough to have considerable value, but I think that most drug screening programs are garbage because those programs are mandated, designed and administered by statistical ignoramuses. A test like this is bound to be a lot less credible in itself than testing somebody for opiate use, and if cops and corporate HR departments can't be trusted with *that*, then they certainly shouldn't be entrusted with *this*. Still, an idea like this could plausibly be useful to somebody who actually knew how to reason from data.

Re:There's a rumor going around

[Maxo-Texas]

I've been using a lot of voice recognition to IM lately.

SO I called someone and got voice mail.. and I said...

"Okay, I got your call and I'll see you later today PERIOD".

I was face slapping as soon as i hung up.

Oh and for the benefit of NSA and twitter type scanners...

I don't know what this shit is. Fuck that. I HATE it when they scan my posts. Period.
I mean blah blah blah, yada, yada. I mean, I don't give a flying fuck, um, what they think.

Thanks--- I was, ah, feeling psychopathic, and um, didn't want to put that in a separate post.

Re:There's a rumor going around

[__aaltlg1547]

If you look at the biggest mass murders in human history, every single one of them was a government official. Think about that for a while.

That doesn't mean they weren't psychopaths, does it?

Re:There's a rumor going around

[bill_mcgonigle]

Could a certain percentage of people who commit violent crime share the common trait of psychopathy? Certainly. Do all?

Right. There are psychopaths who are perfectly fine and functional members of society. People often regard them as 'a bit off', but there is more to the psychology of a dangerous psychopath than just being a psychopath.

Check out this BBC Documentary about the neuropsychology of it. The documentary turns out more interesting than would be initially expected (not going to give it away).

That said, when designer babies are feasible, I doubt anybody is going to consciously select for psychopathy. Well, outside of the dystopian scifi realms anyway.

Re:There's a rumor going around

[sco08y]

But how many people have an online persona that is vastly different than they're real life one? I have friends that act like they're Che Guevara on their social media postings that, in real life, will bitch and cry and complain for an hour if the person at Starbucks takes too long to prepare their non-fat mochaccinolatte

But that's pretty much how Che Guevara did act in real life. The guy was a pampered rich kid who, when he couldn't hack it as a doctor, decided to play soldier. After a string of pitiful failures, he tried to prove how tough he was to his men by personally executing bound prisoners, or shooting his men in the back when they tried to get away.

and don't even get me started on all the people that bitch and complain about "freeloaders" and "people living on the government teat" that I know for a fact are collecting government services themselves or have benefited from them in the past..

We know that the IRS tracks donations to the treasury and there are virtually none. Therefore, people who want higher taxes don't actually donate to the treasury, they just vote for higher taxes. So their actions demonstrate the acceptable standard: play by the rules as they are, and vote for how you want them to be.

Thus the pro-taxers have clearly established that as long as you vote to eliminate government handouts, it's perfectly fair to take the benefits while the law stands. And even that's ignoring that you've already paid taxes for those benefits anyway.

Re:There's a rumor going around

[LaminatorX]

Most psychopaths are simply called "Boss."

Re:There's a rumor going around

[garyebickford]

It also conflates correlation with causation. The supposed fact that they were government officials could be the result, not the cause - psychopaths are likely to be attracted to positions where they can wield power, whether in social, tribal, corporate or government environment is just a matter of convenience and availability. Also, it's hard to say whether Genghis Khan or Alexander the Great would rightly described as 'government officials'. Or a large number of ancient Roman Caesars - as often as not they came up through the military, and often achieved the emperorship by means of social manipulation as well as the use of force. I don't know of a single Caesar who came up through the bureaucracy, which would be the province of 'government officials'. So I'm inclined to disagree with the parent's proposition entirely. The only one I can think of offhand was Hitler, who did manage to get himself elected.

A recent NY times article discussed work on discovering psychopathic tendencies in young children. It's early days but there seems to be some pretty good evidence that this problem can be diagnosed and eventually possibly treated early. Psychopathy is apparently about 80% inherited. The childhood 'calculating unemotional' (CU) syndrome is strongly correlated with later psychopathic behaviors, but about 1/2 of those with the syndrome seem to grow out of it or learn how to behave in their teens, so it's unfair to stigmatize them early on and this leaves hope that the percentages can be improved.

IANA evolutionary biologist, but the fact that psychopathy and leadership are so strongly correlated, and that the syndrome appears to be solidly 1% of the population (up to about 4% in high leadership positions according to another study) implies to me that there is an evolutionary advantage both individually and for a given population as a whole. Perhaps this characteristic, or some components of it, make it more likely that a particular tribe will succeed against its neighbors (such as success in war); or maybe psychopaths' manipulative abilities improve their chances of having offspring.

Re:There's a rumor going around

[BeanThere]

This also represents basically a new class of monitoring, that of continual monitoring of ALL human action by AI's (since that's basically what this is, being watched by an AI). It takes society to a whole new "Welcome to the Panopticon" level.

Algorithmic monitoring will soon be behind all mass-surveillance, e.g. automatically scanning every CCTV image to see if what you're actually doing might be considered 'suspicious' (e.g. looked at a young girl for just a few seconds too long? look 'fidgety'? look 'angry'? etc.). Algorithmic real-time monitoring of all over-the-wire voice conversations (e.g. telephone, cellphone, VoIP) will all very very soon be a possibility. AI's behind automated flying drones will be watching you in the streets, reporting you to authorities if you seem 'suspicious'.

The future is arriving very quickly, and we need to deal with this head-on and not blithely dismiss it, before it becomes widely entrenched - we need to be asking serious questions NOW about what kind of world we really want our children and grandchildren to grow up in, and how to achieve that. We need to be asking, who are all these technologies REALLY trying to protect - you, from the common street criminal, or will the kleptocratic financial cartels still be stealing billions care-free?

Re:There's a rumor going around

[s.petry]

Most psychopaths are not dangerous, the desire for a so called pseudo-science of trying to cure something obviously not broken is odd. Just like most people have some type of phobia, have narcissistic tendencies, some type of rational paranoia, etc... We don't run around trying to make magic bullets trying to fix it all for several reasons.

First, most of those traits are healthy and responses to the environment. Second, most people do not experience those tendencies for extended durations of time. If the environment is messed up, so are the people.. it's called a coping mechanism. Trying to fix the "problem" without really fixing the problem is something us humans have an extremely poor habit of doing, and the results are often much worse than working on the root cause (environment, education, social influences, etc...)

Instead of trying to fix symptoms, we should be trying to fix the problems...

Re:There's a rumor going around

[Afecks]

people that bitch and complain about "freeloaders" and "people living on the government teat" that I know for a fact are collecting government services themselves or have benefited from them in the past

The more money you take from a thief, the better libertarian you are. The fact that libertarians even realize that taxation is theft puts them miles ahead of anyone on the left or the right. The only way they would actually be hypocrites is if they had an opportunity to abolish the state and refused to do so. I recommend that libertarians put as much burden on the state as legally possible in order to speed things up. Take every dime you can from a thief.

Re:There's a rumor going around

[ShanghaiBill]

Now 99.99% of the people who seem suspicious are dangerous rather than 99.9999%.

If the suspect pool is the general public, then cutting it down by 99% still leaves way too many people. But what if the suspect pool is much smaller? What if there is a serial rapist in a town of 5000? You can use this test to cut the suspect pool down to 50. Of those 50, 20 are female, another 15 have solid alibis, 10 don't fit the description. Of the remaining five, three agree to voluntary DNA testing to eliminate themselves from the suspect list. Does this mean that one of the remaining two did it? Of course not, but focusing some resources on investigating them might be a better option than checking out the other 4950.

I am not claiming that this test actually works. I am just pointing out that a test that is wrong 99% of the time can still be useful under some circumstances.

Re:There's a rumor going around

[Jiro]

There's a better answer to that: those things have good effects (people can use them) and bad effects (money is taken from people so other people can use them) and you oppose them because you think the bad outweighs the good. You can't really "choose not to take them" because you can only choose not to take the good effects, but you can't choose not to take the bad effects. Refusing to take a college grant isn't going to get you a life without taxes that go to college grants.

Like the example goes, if the government confiscated $100 from everyone and used it to give everyone a $95 check, you would not be barred from taking the $95 check.

Not to mention that there are more negative effects than just directly being taxed. Perhaps by setting up X the government distorted the market and crowded out a private version of X that would have been more efficient. You certainly can't choose to take the private version; taking the less efficient government version is not hypocrisy, it's just cutting your losses.

Re:There's a rumor going around

[garyebickford]

To my mind that's just putting pragmatism as a philosophy or value system ahead of libertarianism. :)

Of course, real life does intrude on every philosophy, making it impossible to behave purely with regard to the philosophy. For example, as I understand it, the Jains (used to?) believe that killing even the smallest bacterium is bad, and make great efforts to avoid killing anything. When bacteria were discovered the most extreme believers began wearing masks so as to avoid inhaling and inadvertently killing bacteria that floated in the air. One wonders what they think now, to discover that by count our bodies contain 10 times as many bacterial cells as human cells, and zillions of those bacteria die every day in this very accelerated ecosystem that we call the 'gut'.

This is just one example that every 'ism' is wrong - 'isms' are an attempt to impose a rational structure on reality, which is inherently arational - or if you prefer 'beyond rational'.

Re:There's a rumor going around

[Anonymous Coward]

Probably don't want to hire a psychopath or marry one.

My wife loves being married to me. She did learn to tame me though, once she realized that sometimes I'm more like a beast than a man. She also has used me for protection from her family, who have dangerous criminals among them.

Plenty of women have dangerous dogs for protection with no problem to themselves. Human regularly tame animals far larger, stronger and more dangerous physically by applying fairly simple, easy to understand training methods. There is no reason why a woman can't marry a psychopath and be very happy, so long as she is prepared to dominate skillfully, rather than with the nagging, whining weakness that many men put up with. I'm much more capable than a dog, even though I require similar training.

I have defined winning the battle of the sexes as keeping my stuff, getting regular sex and in home access to my children in a world where the woman can at her whim move the battle to court and automatically remove all that. With this perspective, I am able to satisfy my emotional need to WIN by making her happy, since when she is happy she won't move the battle into court where I lose. As a friend remarked when I described this to him: "She probably doesn't even know she's losing".

I'm actually really good at pushing happy buttons. Much better than you, since I do it deliberately, consistently and with ruthless efficiency. It takes me about 30 seconds to turn the situation around if she's in a bad mood, face to face, by which I mean her throwing her arms around me in excitement. If you have a psychopath convinced that their self interest lies mainly in your happiness, you are in for a very happy life.

Re:There's a rumor going around

[Jane Q. Public]

"Well, the test of something's value shouldn't be whether it strikes you as silly."

Yes, it should. Truly silly things are quite valuable.

Using Periods?

[Anonymous Coward]

Bullshit. Using periods is simply properly using the English language. (And expletives have their proper place as well--although this one is more ironic than angry :) )

Re:Using Periods?

[cheekyjohnson]

What are you, some sort of psychopath? Flag him!

Re:Using Periods?

[__aaltlg1547]

i think the lack Capitalization and lack of Proper punctuation may be more Useful in Finding people suffering from Various Mental Disorders on the other hand it may merely Indicate a shortcoming In the posters education isn,t it obvious that when some one fails to Use punctuation and Capitalization correctly that they are a Tragedy in Waiting clearly the fbi is Onto Something or at least they Think They Are it cant be simply a case of people Being too lazy to punctuate their sentences or find the fucking shift key many messages on twitter come from mobile devices where it can be pretty damned inconvenient to apply capitals and punctuation marks as compared to a proper keyboard

Re:Using Periods?

[CanHasDIY]

i think the lack Capitalization and lack of Proper punctuation may be more Useful in Finding people suffering from Various Mental Disorders on the other hand it may merely Indicate a shortcoming In the posters education isn,t it obvious that when some one fails to Use punctuation and Capitalization correctly that they are a Tragedy in Waiting clearly the fbi is Onto Something or at least they Think They Are it cant be simply a case of people Being too lazy to punctuate their sentences or find the fucking shift key many messages on twitter come from mobile devices where it can be pretty damned inconvenient to apply capitals and punctuation marks as compared to a proper keyboard

What you did: it's there, and I see it.

Really?

[WrongSizeGlass]

The ability to predict psychopaths in under 140 characters? What have all these mental health professionals been doing? All these face-to-face interviews, "sessions" and observations. Maybe if they submitted their reports in only 140 characters they would have figured all this out a long time ago. /sarcasm

Re:Really?

[reub2000]

Probably is bullshit. The twitter responses where compared to a survey. In other words they calibrated their instrument by simply asking people if they're a psychopath. Lets see, one of the most well known trait of psychopaths is their tendency to lie.

Re:Really?

[WrongSizeGlass]

Lets see, one of the most well known trait of psychopaths is their tendency to lie.

Is that true ... or are you just a psychopath?

Re:Really?

[Anonymous Coward]

Yes. If somebody uses Twitter, he can’t be right in his head in the first place. ^^

(There are only two types of "people" that use Twitter: 40+ midlife-crisis people who want to be "cool" again, and marketing companies posing as celebrities. No exceptions.)

Re:Really?

[Anonymous Coward]

"The ability to predict psychopaths in under 140 characters?"

***Psychopath test***
You think your crap is worth following? Check!
You follow the crap of strangers you have never seen? Check!

If one or more of the above are checked, congratulations, you're a Psychopath.

Re:Really?

[Anonymous Coward]

The ability to predict psychopaths in under 140 characters? What have all these mental health professionals been doing? All these face-to-face interviews, "sessions" and observations. Maybe if they submitted their reports in only 140 characters they would have figured all this out a long time ago. /sarcasm

I have a crazy idea... Maybe they use multiple tweets? Mind blowing, I know, but 2 tweets add up to 280 characters! Maybe you could even use 3! Or even more?!

A longitudinal tweetment?

Re:Really?

[Anonymous Coward]

Lets see, one of the most well known trait of psychopaths is their tendency to lie.

And that's why many of them become politicans.

Don't dare to mod this as funny.

loudbot

[HighBit]

oh dear, I wonder what would happen if they ran the model against loudbot - http://twitter.com/loudbot -- it'd probably score off the charts

Analyzing myself

[KiloByte]

> Curse words.
Hardly any, but mostly due to many years on a MUD with strict rules. Abusing "damn" and so on, though.

> Angry responses to other people
Hell yeah.

> including swearing and use of the word "hate."
Got a bigger vocabulary, but yeah.

> Using the word "we."
Check, to a big extent.

> Using periods.
You mean, so those with no punctuation are not morons but normal people? Blah. Check.

> Using filler words such as 'blah'
See above.

> and 'I mean'
Check.

>and 'um.'
"hrm", "hmm" and "ghrmblah" (see also two paragraphs above)

So, you mean, is there any hope for me?

Re:Analyzing myself

[WrongSizeGlass]

So, you mean, is there any hope for me?

Sure there is. First, you'll seek twitter therapy and twitter assimilation resistance resistance. If that succeeds your tweets will be indistinguishable from 99% of the population (and most of your followers will finally be able to understand them).

If your TT&TARR fails you'll have to go to twitter court (most trials are very short) or simply go to twitter confessions and throw yourself on the mercy of 12 randomly selected twitter addicts. Once you're sentenced to twitter prison (where all you say and read are tweets) you'll be able to start paying your debt to the twitterverse. When the time comes for your parole review they'll analyze your tweets during your sentence to determine if you're ready for early tweetlease or if you need to try harder.

So yes, there is hope for you ... though not much hope for society at large if we're making DSM-V quality diagnoses based on twitter (unless we're determining that those who use twitter really have too much time on their hands).

Narcissism?

[humanrev]

rate users' levels of narcissism

If Twitter users are anything like Facebook's, then if you're trying to use narcissism as a rating for psychopathic behavior you're gonna get a LOT of hits.

Re:Narcissism?

[jaymemaurice]

My awesome facebook profile is narcissism free. I only use facebook to show others me doing the awesome things I have done so they can comment in ways that make me feel good... but the band Carcass is good, just not interesting as me.

Re:Good-Bye Civil Rights

[craigminah]

Try doing this with skin color, religion, race, sexual preference, etc. and you'll quickly find an angry torch-wielding mod at your front door. Switching things up to look at parameters perceived to be under the control of the person and it's ok. Kind of ironic people don't see this as an even bigger potential threat to civil rights than anything else. Like anything, this will start of with good intentions and be limited in scope (e.g. Twitter) but soon, if funded, it will be used for evil and be linked to Google's database to refill prisons around the USA. Yay.

Re:Bullshit.

[Black Parrot]

By these qualifications, every twelve year old is a psychopath. Also, seriously, using periods? You mean, as in proper fucking writing? What else, proper capitalization of sentences and pronouns?

Maybe "anal retentive" is a slippery slope toward "psychopathic".

Re:Bullshit.

[hcs_$reboot]

Maybe the same algorithm should be applied to AC comments on /. ...

I Fucking Hate...

[Anonymous Coward]

I fucking hate stupidity like this being passed off as "sciency". I mean, next up they'll claim to use palm scanners to determine my future actions. Blah! We need to vote these types of articles down. It's worthless shit!

I guess the only thing between me and a psychopath is that I don't use The Twitter.

Idiots!

So, basically, speaking English.

[Arancaytar]

Some of the Twitter clues: Curse words. Angry responses to other people, including swearing and use of the word "hate." Using the word "we." Using periods. Using filler words such as 'blah' and 'I mean' and 'um.

Speaking English either formally or informally and either without emotion or with emotion. I have also heard that almost all of them breathed regularly and wore clothing. Those are relevant clues that might help recognize the next one early.

On the other hand, that they all owned assault weapons is purely a coincidence.

Implying those qualities are illegal

[loufoque]

So basically, they're implying that if you're narcissist or machiavellianist, which are both valid of traits of human nature, you should be arrested?
What a nice display of tolerance.

Phew!

[Psicopatico]

Good thing I don't use Twitter then.
I think.
Um, yeah.

1984

[wbr1]

Big brother is watching you....140 characters at a time.

They never heard of Jon Ronson

[cellocgw]

"The Psychopath Test" author Jon Ronson pretty much makes the case that there isn't any test.

So, um, yeah.

[darkain]

So, um, yeah. Apparently we, erm, I mean I am a psychopath because of a few fucking words I like to say? blah. I hate articles like this.

Bipolar people are not psychopaths

[eggstasy]

And they hate everything all the time, as a result of passive-aggressiveness during the depressive phases, or mere aggressiveness and grandiosity on the manic ones, or simply a tendency to hyperbolize their description of the world based on their own extreme feelings and disinhibition. They will also, of course, get into angry arguments.