Algorithm "Outs" Facebook Users
- Thread starter Lauren Admire
- Start date
Yeah sorry you seemed to have missed my point. I dont mean every intelligent person should work on medical equipment cause that would be silly. I mean people should contribute something that leads to the gain of humanity if they are as bright as these two seem. Instead they compose an algorithm to find gays on facebook. See what im geting at?chronobreak said:
Sorry, I've just had a few encounters during my life because I was so unsure about what I was. It kinda gets hard to tell the jokers from the serious when some people say stuff like that utterly ironically.danpascooch said:
What is wrong with the pursuit of knowledge for it's own sake? Not every thing needs to, or even does benefit mankind. Even so, I would argue that the pursuit is beneficial in it's own right. Take mathematics for example. Non-euclidean geometry was just a curiosity for mathematicians with no real world applications, that is until a friend of Einstein's recognized that the formulas Einstein was working with would fit neatly into non-euclidean geometry.DividedUnity said:
About the article in general, most people seem to be missing the point. The MIT students have not done anything new here. They simpled applied an algorithm to a data set to produce usable information. The point of the article is to illustrate how easy it is to do. Let's say someone created a profile, but did not list their sexual orientation, with this algorithm one could (with a certain margin for error) determine the persons sexual orientation. The same approach can be applied to anything. How do people think social networking sites make money? They do exactly that. It even has a name, "Data Mining".
Facebook data mines it's servers looking at the connections between people. It combines those connections with what is posted on a person's page and his friend's pages to build a profile of his likes and interests, then Facebook uses that profile to target ads. Facebook has even gone beyond that, and sells these profiles to third parties. This is all possible using algorithms just like the one the students used.
While interesting, I can find three things wrong with this idea:
1)Definition of "Friend". I know for a fact that there are people that accept every friend invitation they get, even if they've never met the sender. To these people, the size of their friend list is a thing to be proud of. Therefore, it is possible to have to have no idea about the orientation of most of your "Facebook friends".
2)Communities. Let's be frank, if most of someone's friends are black, there's a good chance that they're black themselves. As the article says "like attracts like". However, this leads to generalizations. You don't have to be black to have black friend, and you don't have to be gay to have gay friends. Which leads to my next point.
3)Allies. Individuals that are a member of an LGBT collective but themselves are heterosexual are deemed "allies". It would then stand to reason that said individuals would have a significant number of homosexual friends, as well as friends who fall into the other categories represented in the LGBT. The algorithm fails to take this into consideration.
Conclusion: the MIT guys still have a long way to go.
1)Definition of "Friend". I know for a fact that there are people that accept every friend invitation they get, even if they've never met the sender. To these people, the size of their friend list is a thing to be proud of. Therefore, it is possible to have to have no idea about the orientation of most of your "Facebook friends".
2)Communities. Let's be frank, if most of someone's friends are black, there's a good chance that they're black themselves. As the article says "like attracts like". However, this leads to generalizations. You don't have to be black to have black friend, and you don't have to be gay to have gay friends. Which leads to my next point.
3)Allies. Individuals that are a member of an LGBT collective but themselves are heterosexual are deemed "allies". It would then stand to reason that said individuals would have a significant number of homosexual friends, as well as friends who fall into the other categories represented in the LGBT. The algorithm fails to take this into consideration.
Conclusion: the MIT guys still have a long way to go.
Can anyone say "how bored to tears were you?". I mean, it might be cool that you can figure out what kind of person is by looking at their friends list, but it is creepy at the same time because, if it's all true, it can tell things about someone they don't want to know, but the hints are spread out enough that no one could figure it out.
1.) People you interact with on a regular bases are people you'd probably consider as friends. So filter out "Friends" with no pm's or im's or that you block updates from. It can be fine tuned from there.C_Topher said:
2.) You are correct when you say "...you don't have to be gay to have gay friends". The students are using a known fact about human psychology and the information that is available about a person on Facebook to make an educated guess about said person's sexual orientation. They can be wrong, it doesn't diminish the point the article is trying to make.
3.) As addressed in point 2, the algorithm can make mistakes (called false positives). Even so, with enough data and a complex enough algorithm, the number of false positives can be made rare enough that they aren't worth considering. If students , with just the information available to them on Facebook, are able to write a program that produces useful ( though not necessarily 100% accurate) results, imagine what Facebook can do.
I think it's old fashioned to think that anyone with gay friends is gay or anything along those lines. Sure when homophobia was more prominent it might have been heavily divided, but these days it's just impossible to tell orientation just by looking at someones friends.Ironlenny said:
At most they'll probably snag a few people that 'might' be gay or bi but it's likely they'll get more wrong than right.
The algorithm in the article is about statistics and human behavior, not whether or not it's outmoded to think having gay friends makes you gay.
If the algorithm says it is likely (with the given data) that an individual is gay, that doesn't make the person gay. Besides, that is not the point of the article. That is just an example. The same approach can be applied to anything. Need I remind you that this was a project for a class about ethics and law? The students aren't trying to out people, therefore they don't care if it is 100% accurate. It is simply a demonstration of what can be done with an algorithm and some data. It's about privacy, people make information available with out them even knowing it.
If the algorithm says it is likely (with the given data) that an individual is gay, that doesn't make the person gay. Besides, that is not the point of the article. That is just an example. The same approach can be applied to anything. Need I remind you that this was a project for a class about ethics and law? The students aren't trying to out people, therefore they don't care if it is 100% accurate. It is simply a demonstration of what can be done with an algorithm and some data. It's about privacy, people make information available with out them even knowing it.
Why is being gay still such a big deal? Why do people have to be 'outed'? The human race seriously needs to grow the hell up and start being more accepting of divergent sexuality, to make homosexuals feel welcome instead of insecure.