4.74 Degrees of Separation on Facebook

First time accepted submitter perryizgr8 writes "Facebook Data Team has taken all the friends data of everyone on Facebook and analyzed it, finding out the shortest distance between every two persons. They can now confidently say that the average degree of separation between any two humans is 4.74, not six as previously claimed by various entities."

Comments

Disagree

[Kazymyr]

Mine is infinity since I don't have a facebook account.

Re:Disagree

[Githaron]

That is the first thing I thought. It is like saying of the set of people I personally know, there is at most one degree of separation between any two people.

Re:Disagree

[Githaron]

Wait, is the degree 0 or 1 between two people that know each other personally?

Re:Disagree

[Anonymous Coward]

You also don't know anyone with a Facebook account, and no one you know knows anyone with a Facebook account, and so on? I'm not sure you understand what they are talking about, you read "Facebook" and just wanted to tell people you don't use it.

Re:Disagree

[rapidreload]

Lemme guess, you don't have a TV either but want to tell the world regardless?

Re:Disagree

[fisted]

yeah, so antisocial not to hand over all data to facebook.

Re:Disagree

[similar_name]

I think it would be interesting to see the average degrees of separation for each individual. One person might have an average of 9 degrees separation to everyone else while another individual might average 3. Cross that number with standard demographics data and look at any correlations. x being people not on Facebook you could still compare people with 3 + x degrees vs 9 + x degrees.

Re:Disagree

[foobsr]

is the degree 0 or 1 between two people that know each other personally

The question might be whether the degree between you and yourself is infinity and only approaches 1 after an enormous amount of training. Just a thought.

The implication is that any small number whithin the context given is worth ... well.

CC.

Re:Disagree

[DavidTC]

0 is the degree of separation with yourself.

Kevin Bacon is the only actor with a Bacon number of 0.

Re:Disagree

[Kjella]

I think it would be interesting to see the average degrees of separation for each individual. One person might have an average of 9 degrees separation to everyone else while another individual might average 3.

Extremely unlikely. Remember that the number of people connected grow exponentially. Your friends are few. Friends of friends are many. Friends of friends of friends is insanely many. Even if you're a tightly knit bunch already after 1-2 steps you're bound to have many connected to the "main" network. Personally I know I have people in my friend list that have gone through every class list since primary school - that's how I'm their friend. If you have *one* of those people as friends, or even friend of friend you're extremely well "connected" even if you can count your Facebook friends on one hand.

Likewise it's not likely to go as low as 3 because if you say 100 friends average then the most people you can reach in 4 connections is 100^4 = 100 million. The only reason I think you'd go as high as 9 would be if you're an isolated tribe deep in the Amazons with 3 degrees of separation to the few researchers that are there, that are 6 degrees from the rest of the world.

You can do the math the other way around, to connect 7 billion people with six degrees of separation each degree of separation must expand the network 7 000 000 000^(1 / 6) = 44 times. Is that likely? Yes. That doesn't mean 44 friends though, it's more complicated than that. The first degree is my direct friends, that is simple. The second degree is friends of friends minus those I'm friends with directly but only counting each person once. So if five of my friends went to the same school and know the same person (that I don't), he's only counted once. So the formula is

Unique persons brought into the network * 1 +
Shared people brought into the netowork * 1/n where n are the people shared with +
People already known to the network * 0 = 44.

That doesn't seem that unreasonable, to my friends my work mates and family are new, to my work mates my friends and family are new and to my family my friends and work mates are new. Different school history, work history, different family, lived different places... each degree brings plenty new. Take for example my study mates, very many of them studied abroad. Each of them is like a new boom of contacts entirely new to the network.

Re:Disagree

[dotancohen]

Lemme guess, you don't have a TV either but want to tell the world regardless?

Actually, I'm the caveman without a TV. But I do have my neighbour's wifi!

Re:Disagree

[Sqr(twg)]

Everyone's "average degree of separation" is infinite, because there exist uncontacted indigenous tribes in Brazil.

Perhaps facebook can partner with the OLPC foundation can fix this?

Re:Disagree

[leuk_he]

More interesting, who is that 0.75 person. A dwarf, a invalid without legs? a hobbit?

Skewed Data?

[Sharkyfour]

Wouldn't this be skewed by all the people who befriend random strangers to increase the size of the Mafia's or farm friends?

Re:Skewed Data?

[aicrules]

Even better than that, it's skewed to those who are on facebook. So you add those two things and we're back to 6 degrees. Dumbass story

Re:Skewed Data?

[Anonymous Coward]

Oh knock it off.

Assuming Facebook relationships are used as a model for our casual relationships outside of Facebook... it probably stands up better than any previously attempted method. It's not like they're working with an overly selective sample group. And what's more, I'm not aware of any major and consistent social differences between people with and without facebook accounts. The tiny percentage of people that don't have facebook accounts specifically because they're socially phobic isn't likely to be statistically relevant.

The real fault in it, if there is one, is that it's skewed towards 1st world people. A kid in some remote part of Africa that's never been to a large school, or even seen a computer probably can't claim to have met 300 people, by phone, computer or other personal interaction. Casual or not, they're just remote and wouldn't have been figured in.

Douchebag neckbeards just aren't relevant.

Re:Skewed Data?

[Spad]

However, given the perfectly valid observation that people on Facebook are much, much less selective about who they "friend" than in real life, the results will inevitably be skewed as a result.

Additionally, Facebook has ~800 million accounts of which an unknown number are inactive, fake, duplicate or for some shitty new product, which is less than 12% of the global population.

i.e. Whilst interesting data, it would be stupid to try and claim that it can be used to infer anything about peoples' general relationships outside of Facebook.

Re:Skewed Data?

[maxwell demon]

Assuming Facebook relationships are used as a model for our casual relationships outside of Facebook...

That's probably as valid as taking the opinions expressed on Slashdot as model for the opinion of the general population. In other words, not very.

Re:Skewed Data?

[meza]

Maybe I'm different from other Facebook users then, sure I have some people on my friends list that I only met once at a party and now don't even remember who they are. But in real life there are so many more people that I know casually and would say I'm "connected to" that I am not friends with on Facebook, such as: my hair dresser, my dentist, my boss, other colleagues, all the people I ever went to school with (of whom I've probably befriended less than 25% on Facebook) all the teachers I ever had, my neighbours, distant relatives, my siblings friends etc etc.

So I think if we included everyone we know in real life the degree of separation would probably go down, not up.

Re:Skewed Data?

[DavidTC]

Except that the idea of degrees of separation isn't 'friends', it's 'People who know each other'. No one ever said those people had to be 'friends'. No one's ever bothered to try to define exactly what that means, although at minimum you probably have to have exchanged words with them at some point, and have a way of contacting them.

Granted, on Facebook, it's probably slightly too loose even with that requirement. Apparently, some people on Facebook go around friending anyone who shows up as a likely friend, regardless of whether or not they actually know the person. And sometimes the other person accepts that request. Clicking on someone's picture and sending a request is probably not actually 'knowing' someone.

So assuming, on average, one of those bogus 'knowing people' per chain of '4.74' people, which caused the calculation to skip a number that really should be there (They aren't X's friend but they are the friend of the friend of X) ...it comes out to essentially what people have been saying all along.

Which is weird, because as far as I know, 'six degrees' isn't based on any scientific information...it's from playing a game with Kevin Bacon. (Which is not about who 'knows' each other, it's about who's been in movies and TV shows with each other.)

Re:Skewed Data?

[Sique]

Moreso, the average chain length has nothing to do with the original 6-degrees-of-separation claim.

The original statement was that there aren't two people on Earth for which the shortest chain of people which connects both is longer than 5 other persons. You could construct a situation where the average degree of separation is slightly above 1, and still have at least two people with a longer chain than 10 other persons. Lets say you have 990 people who know each other, making the average degree of separation of this group exactly 1, and 10 people where one only knows a single person, this one knows the first one and then another one, and the tenth knows the ninth and one of the 990 others.

Then you would get an average degree of separation of about 1.035 and still the longest chain is 11.

Re:Skewed Data?

[bluegreen997]

My thing with the 'data' is when you consider those Facebook users who friend *everyone* because that is just how they use the service. Case in point:

I was listening to a local radio personality who was telling a story about how he un-friended someone and the huge fallout that happened because of it. At one point he lamented over the fact that since he had over 800 friends, and asked like who dosen't!, how was he supposed to know that person was someone that he actually worked with. (Not directly but they are a part of the same company.)

To me while the whole 6 degrees thing has always been more of a fun social thought game, trying to say that what Facebook has done here is anything more than that does not ring true.

"Research" indeed

[Kittenman]

A good promo for Facebook ... gets it in the news without mentioning 'security' Dammit, I just did.

Re:"Research" indeed

[hellkyng]

Right! All this wasted research when anyone on 4chan could easily have told you how many friend requests it would take to find the neighbors hot daughter who just went off to college and joined a sorority...

So overall, it is 6...

[Anonymous Coward]

6 degrees of separation, not limited to any single medium

Just under 5 for any two facebook'ees
but to get to anyone not on facebook, you'd have to go one extra hop

Re:So overall, it is 6...

[Baloroth]

This would assume that Facebook is the most efficient separation path. I.e. there are paths of friends who aren't facebook friends (I have quite a few of those actually), who don't use Facebook, etc. So you cannot use this statistic for any person or people who are not on Facebook, nor can you draw any accurate conclusions about their degrees of separation.

Re:So overall, it is 6...

[BradleyUffner]

Actually 7 if you are going from someone not on facebook, via the facebook network, to a second person also not on facebook.

So assuming the traditional value of 6 in meatspace is correct, then the facebook network is actually less connected than the real world.

If both people are in meatspace there is no reason to ever hop out of meatspace and in to facespace, so back to 6.

Re:So overall, it is 6...

[TexVex]

I understood that six degrees was supposed to be the maximum, not the average.

Kevin Bacon

[Aaron32]

I wanna know my link to Kevin Bacon. Do you think FB would tell all of us how we get back to him?

Re:Kevin Bacon

[sco08y]

I wanna know my link to Kevin Bacon. Do you think FB would tell all of us how we get back to him?

Facebook can't, because ever since the introduction of CCTV and police cruiser dashboard cams, they've had to rename it the Six Degrees of Lindsey Lohan.

Minimum

[Zaphod The 42nd]

I'd be curious to know what the minimum number of connections necessary to link any two persons is, over the average.

Re:Minimum

[pipedwho]

I'd be curious to know what the minimum number of connections necessary to link any two persons is, over the average.

Probably infinity as there are most likely some small subgroups that don't contain a friend with an account that links outside that subgroup. That means that Facebook can't reliably measure a maximum connectivity estimate without saying 'infinity'. And even if they remove the infinity outliers, there would still be some higher number linkages simply due to a lack of active friending between people that do know each other, but don't want them to be 'friends'.

In the real world, one of those people in the outlier groups probably 'knows' someone outside of the clique group. But, within Facebook, they may not have an account (or may be intentionally 'non friended') so are not a measurable link.

This brings us back to the old claim of a maximum 6 degrees of separation between any two people on the planet.

no it's not

[Anonymous Coward]

it is 4.74 + 1; think about it...

Re:no it's not

[Tomato42]

4.75 + 1 in my book is very close to the "old, frowned up, value" of six...

Re:no it's not

[Zamphatta]

Where are all the mods? Somebody needs to up this guy's score 'cause it's statistically insightful. After all, if it's 4.75 between Facebook friends, then to connect someone without a Facebook account, one would have to add another 1 to it. Which totally blows away the argument.

Re:no it's not

[jrumney]

There are whole villages across Asia, Africa and South America that do not have an internet connection. Are you sure everyone on Earth knows someone with a facebook account?

Re:no it's not

[MurukeshM]

If even among Facebook users, between any facebook user and other facebook user there is 4.75 person, it is 4.75 and not +1.

As now you are jumping on the gun thinking that person who does not have a facebook, has 5.75 person between him and anyone else who has or does not have facebook account.

a) Either you don't understand why he thinks he should add the +1, or
b) You're trolling, or
c) You just re-read what you wrote, remembered what averages are, and had a facepalm moment.
I'm not sure which it is, though.

He's not saying that there are always 5.74 people between a non-fb-user and anybody else. He's just saying that, given a non-fb-user, we can use one of his friends (which, I admit, he may not have any of)) with an fb account to be, on an average, within 4.74 of another fb user. It might be that for specific cases the actual number (being an integer) is ore than 5.74 or less than 5.74. Let us say there's an average of 1.26 people between a person without an fb account and one with. So we get 6 degrees of separation. :) :P (What was that law about making up stats on the spot?)

Also, you gave a specific example, which doesn't count when it comes to Statistics. For every dropout who became a millionaire, how many do you think are starving now? Would you use a specific example of a homeless dropout or a millionaire dropout to justify remaining in or dropping out of college?

Re:no it's not

[RivenAleem]

+5.75 Insightful

Interesting

[koan]

So if you're on Facebook you're only 4.74 degrees from some maniacal jihadist, right-wing Christian extremist or a pedobear...

Re:Interesting

[gandhi_2]

.... on average.

Bad Claim

[Ibiwan]

1) As others have pointed out, not all humans are on Facebook. I'm sure the FB researchers would be hard-pressed to believe that, though.

More importantly,
2) The "six degrees" is supposed to be the MAXIMUM linkage between any two people -- not the average. Good job disproving something nobody ever claimed, guys!

3.74 I think you will find

[Anonymous Coward]

http://www.bbc.co.uk/news/technology-15844230

Cool

[Anonymous Coward]

I stopped using facebook way before it was cool to stop using it.

Re:Cool

[Dyinobal]

ha I've never used it so I got you beat.

Re:Cool

[Ravon Rodriguez]

I use Facebook ironically.

Re:Cool

[Anonymous Coward]

I use Facebook ironically.

No, in Soviet Russia you use Facebook. Elsewhere ...

Misrepresentation of the original research

[smellsofbikes]

If anyone wants to read a good analysis of the *original* six-degrees-of-separation study, Malcolm Gladwell wrote about it in The New Yorker about ten years ago. (You may wish to skip ahead to part 3.) The researchers -- and this was Stanley Milgram, of the infamous Milgram Experiment involving people's willingness to torture other people -- gave people envelopes addressed to a specific person, and told them to write their names on the envelopes then give them to someone they thought might know the addressee. When all the envelopes came through, they analyzed both the number of hops and the route. (The average was somewhere between 5 and 6 hops, with some being higher. There is no assurance this is the shortest route, but their initial estimates were 100 hops, not five.) The most interesting part was that of the envelopes that reached their destination, more than half came through just three people. It's the discussion of those people, the ones who know people in various different close-knit communities, that matters: they're the connection points.

this just in

[geekoid]

average is less then the most.

islands

[stud9920]

What I'd like to know is: "are there islands", i.e. groups of people completely unrelated to other groups? And if there is no such island, what is the max distance between any two people?

Re:Friends really? most are just passing acquainta

[maxwell demon]

If you were having a party or getting married you would invite your friends, 99% of the "Friends" on facebook wouldn't get an invite.

Not intentionally, at least. :-)

Incorrect conclusions

[ThePeices]

"They can now confidently say that the average degree of separation between any two humans is 4.74, not six as previously claimed by various entities."

Wrong.

They can confidently say that the average degree of separation between any two humans on facebook is 4.74.
Not only that, but "various entities" never claimed that the value was six for facebook account holders, they claimed 6 degrees of separation between all people.
The authors incorrectly assumed that every human has a facebook account.

Not exactly.

[raehl]

The old value is that no person is more than 6 degrees of separation from ANY OTHER PERSON, period. So, randomly pick any person on the planet, and you should be able to get to that person with no more than 5 intervening people.

An *AVERAGE* of 4.74 doesn't say anything about a 6-person maximum.

Re:Not exactly.

[Anonymous Coward]

Of course, that incredibly vague metric doesn't really explain what a degree of separation is.

Is it just people you personally know? Or does it count:
- people you've only met, even if you've forgotten them?
- people you work with?
- people you work in the same building as (regardless of whether you work for the same company)?
- people you work in the same company as (regardless of whether you've ever met them)?
- people you've done business with at some point (including the check-out clerk at a shop or their manager, whom you've never met)?
- people you've passed on the street?
- people you've ridden a bus with?
- neighbors, whether you've ever talked to them or not? (And in how large of a radius?)

So 6 degrees of ... whatever. You could say everyone shares only 1 degree since they've breathed the same atmosphere or you could massively increase the steps by making a "degree of separation" limited to parent-child genealogical links. It's all just lies, damn lies, and statistics anyway.

Re:Not exactly.

[AK Marc]

The definition of the "Bacon score" requires someone have an official credit in a film. This made me think back. I have a playbill with my name next to Nick Stahl's (he'd remember me if I ever ran into him again, I bet, though I'd rather run into his sister), and Nick was in a movie with Kevin. So I either have a score of 2 (having acted with an actor who acted with Kevin) or a score of infinity, as I've never been credited in a movie. Most people I mention this to claim "that doesn't count." So, until you receive credit in a movie, you have no score.

Re:Not exactly.

[Lorens]

An average of 4.74 for people connected to FaceBook. So? There are lots of people NOT connected to FaceBook. If anything this finding actually gives weight to the value 6 for everyone on the planet!

Re:Doesn't Matter if you have a FB account

[ChatHuant]

Yeah, well, what if you don't have an email address?

Perhaps closer to reality, what if you don't have any friends?

Outside the metric's domain; I'd already graduated

[tepples]

You also don't know anyone with a Facebook account, and no one you know knows anyone with a Facebook account, and so on?

The metric covers only Facebook, just as the Erdo"s metric covers only coauthorship of articles in scholarly journals and the Bacon metric covers only publicly exhibited feature films. I am meatspace friends with a lot of people who have Facebook accounts, yet because I don't have one myself, I am outside the metric's domain.

you read "Facebook" and just wanted to tell people you don't use it.

Facebook launched after I had already got a degree and lost access to my @*.edu. That's the excuse I've tended to give.

Celebrities

[Per Wigren]

How much is the result skewed because of celebrities with tens or hundreds of thousands of "friends"?

Well, as one of the authors...

[vigna]

...I made three days ago a submission with the correct number (3.74). /. couldn't care less. On average the degrees are 3.74, and this is why our paper is called "Four degrees of separation". 4.74 is the distance, which is one more than the degrees of separation. Unfortunately sociologists decided to count the "intermediaries" (so if we are friends, our degree of separation is zero), whereas computer scientists count "hops" (so if we are friends, our distance is one). This generated a lot of confusion. But, just to be clear, no, we did not round 4.74 down to 4; we rounded 3.74 up to 4. :)