Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!


Forgot your password?

Slashdot videos: Now with more Slashdot!

  • View

  • Discuss

  • Share

We've improved Slashdot's video section; now you can view our video interviews, product close-ups and site visits with all the usual Slashdot options to comment, share, etc. No more walled garden! It's a work in progress -- we hope you'll check it out (Learn more about the recent updates).


Researcher Predicts Your Next Facebook Friend 66

Posted by samzenpus
from the 29-points-of-compatibility dept.
itwbennett writes "Stanford professor Jure Leskovec knows who your friends will be before you've even met them and has won a Microsoft fellowship for his analysis. 'Data shows that who will be our next friend on Facebook is not so random as we think,' he said. Based on information about the personal networks of users and their communication he was able to tell in advance half of the new contacts they would add shortly after. In the future the rate of correctly predicted new friends could be even higher, he said. 'We are able to train the analyzing methods,' Leskovec said."
This discussion has been archived. No new comments can be posted.

Researcher Predicts Your Next Facebook Friend

Comments Filter:
  • by steevven1 (1045978) on Wednesday August 10, 2011 @11:54PM (#37052308) Homepage
    "Hey baby, software predicted that we'd become friends..."
  • by Anonymous Coward

    Stanford professor Jure Leskovec knows who your friends will be before you've even met them

    There's a saying that goes something like this: an expert is someone who knows a great deal about very little. Perhaps the professor knows who my friends are before I meet them, however he completely fails to understand the meaning of the word "friend".

  • If he is talking about Facebook then has forgotten two highly crucial variables in his complex data analysis and methods:

    1) Farmville
    2) Mafia Wars

    After he factors that in, I would *love* to see him predict my next friend in the 7000's range. Shiiitttt.. I'll bet $20 and give him 10:1 odds.

    P.S - I hate Facebook, but have loved Mafia Wars. Way I saw it, I was messing with their ability to predict precisely that. Fight the power.

    • So basically you are saying that Facebook stands or falls with the quality of the games they offer?
      • No.

        He's commenting on the fact that many users of Facebook who ALSO play Farmville/Mafiawars/something else will find themselves inviting strangers to be their friends to increase their whatever. Given the fact that there are a myriad of users utilizing these applications the odds of accurately predicting whom one might invite or accept an invite from next are much greater than were they to stick the numbers racket for clearly defined boundaries of acquaintances.
  • Advertising (Score:4, Insightful)

    by slashqwerty (1099091) on Thursday August 11, 2011 @12:05AM (#37052356)
    The obvious application of this is advertising. Every business wants to sell something. If this research tells them how to convince you to friend them they will be all over it. Political organizations will do the same. I have to wonder if the basic concepts have broader applications outside social media.
    • by Q-Hack! (37846) *

      When I first read the headline, I had the same thought process. This whole concept is just... scary.

    • The obvious application of this is advertising. Every business wants to sell something. If this research tells them how to convince you to friend them they will be all over it. Political organizations will do the same. I have to wonder if the basic concepts have broader applications outside social media.

      While you probably could use it to say "do this to friend X" and there are probably some broad things that reach a large enough population it may not be all that useful for advertisers unless they want to target vary narrow groups. Now if they could put together enough variables so that only a small percentage of each is necessary to reach an audience and the entire set reaches multiple audiences then it may be useful; but we already call that marketing and advertising. They'd probably be interested in th

  • I quit Facebook all three times. Can this guy predict my first friend on Google+? It has been a lonely, lonely 6 weeks.

    I don't want to pay the research cash for a Stanford guy to make this prediction for me. So, maybe I can get a discount and have some MIT geeks figure it out for free?

    And yeah, you don't have to mention the MIT grads I know who won't accept my G+ invites. We don't need to discuss that. I just want to know who and when!!!!!!

    So, so lonely here online. So lonely....


  • FRINKIAC 7 [youtube.com]

  • "In the future the rate of correctly predicted new friends could be even higher." ...or it could be lower.

  • Who's my next friend?
  • by Centurix (249778) <centurix AT gmail DOT com> on Thursday August 11, 2011 @12:47AM (#37052492) Homepage

    We've added these friends to your profile automatically. You were going to add them anyway. You're welcome, because I knew you were going to thank me.

    Love Mark.

    • by Inda (580031)
      Message me when Facebook can predict a 'friend' is not going to pay back that 20 quid.
      • by Rich0 (548339)

        Oh, that's been figured out. What you do is get 40 random strangers to front you 50 cents each, with an offer to pay them back 1/40th of any payments that come in at 4%. Then you only loan your friend $18 and keep $2 for yourself. While you're at it you give $18 more to every homeless person you can find, and there isn't any way that anybody could lose money on such a deal. But, if they're worried they can all promise to pay each other $500 for each homeless guy who defaults.

    • We've added these friends to your profile automatically.

      If you do not want that we add these frinds automatically in the future you can opt out. This easy 30-steps-opt-out process is described somewhere in our FAQ.

  • "he was able to tell in advance half of the new contacts they would add shortly after" So the other half of the time he was wrong. Maybe he was trying to predict who you wouldn't befriend...
    • You got that wrong. Half of the time he was wrong. The other half he predicted that the person who just got friended (and who he predicted wrongly) will friend the first person back.

      So. Where's my grant?

    • TFA doesnt actually specify which of the two types of wrong he was, ie false positive or false negative. False positive means that he predicts you will befriend bob, but you dont. False negative means that he didnt predict you would befriend bob, but you did. If his "wrongs" were only false negatives and he didnt care about false positives, then its pretty easy to get a 0% false negative rate, predict you will befriend everyone!
  • If analysts can now predict relationships without FB, then they can make money off those relationships without FB. So turn off all the servers and give me back my free time and shut down FB please - it's no longer needed as a profit source.

    • The application I see for this is not predicting who you will add as a friend. Rather, it would be in predicting who will accept your friend requests. I suspect those, who are all about making their friend list as big as possible, would like a tool that automatically selects the people who will accept it. I suspect Slashdotters would like the tool even more if it filters the list of people who accept friend requests until it only includes attractive females.
  • That Facebook is rapidly becoming unrelevant? Also fuck Mark Zuckerberg.
    • by Sir_Sri (199544)

      True. Although obviously Facebook has an algorithm to try and do this already (their "Find Friends" result, which they changed recently). It's a pretty weak algorithm though.

  • I think the research is bullshit and the professor is an idiot. What the hell "shortly after means"? Some time in future? Yeah there is a certain probability that I'll add some of my friend's friends in future. Pick a random one. It's not hard to be correct half the times.

  • How do we know the researcher couldn't get the same results picking people at random? The article doesn't describe the method or contain a statement of the problem. What's the null hypothesis, sample space, control group?
    • To me, it seems like a simple Machine Learning problem. The number of common friends will obviously be the most influencing input parameter followed by school, work place, etc.

      Unless someone makes me understand the complexities underneath, I am going to assume that anyone can solve this problem by taking one course in Machine Learning.
      • by RussR42 (779993)
        Facebook seems to have solved it already. As much as I hate it, it's still good for finding lost people and many of them were suggested by facebook.
        • Yes and no. Where it fails is when it suggests friends of friends on the periphery. For instance, I have friends and acquaintances from school who were friends with people I couldn't stand. Facebook seems to suggest these people constantly. They would probably rank highly based on the fact that I "went to school" with them and have lots of friends in common but I have no interest in being their friends after the fact. Similarly, I'm friends with my wife's cousin on Facebook. Facebook suggests her friends to
  • what surprises me the most is that there is a study about this.
  • So Facebook has become self-referential now?

    Sorry, 90% of the people I add as friends I do so because I've met them IRL. How are you going to predict that?

    But, apparently, I'm in the minority and too many people have begun to consider the people they add on Facebook to be their friends instead of the other way around. Am I getting old, or is the world getting stranger?

    • Most of the people you meet in real-life, someone you know has met in real life before, and has added to facebook. It's not that hard to see that if someone suddenly becomes friend with people in your social circle, they are likely to become friends with you too.
  • And so do researchers. It shows how much gray matter is being devoted to social networking that could be spent on better things. That's because the social networking problems are easy. Curing cancer is, you know, hard work.
  • Can someone actually find anywhere in the article where it states exactly what it means by "tell in advance half of the new contacts"? Does this mean they literally pin-pointed the exact person 50% of the time ahead of time? Because that sounds like a heap of bullshit. If it just means it predicts some aspects of the next person you will friend/be friended by, then I would find it much easier to believe yet much less impressive.

  • The most recently added individual on my FB friends list is an illegitimate third cousin that no one in the family knew about. He took his step-father's surname. While he does, in one sense, fall within their "six degrees" model, I doubt that any algorithm could have discovered him. There were certainly no clues on Facebook.

  • Would have to be real friends for this to reliably work, wouldn't it?
  • Why wait until someone pisses you off...?

If you steal from one author it's plagiarism; if you steal from many it's research. -- Wilson Mizner