Facebook Wants to Own All Your Social Graphs, Not Just One


Updated: Facebook has popularized the use of the term “social graph” as a way of describing all the various social connections you have to people in your life, both online and in the real world. But Chris Dixon, co-founder of Hunch.com and an angel investor in a number of web startups, says in a blog post published today that there is more than just one kind of social graph — in fact, he argues that there are actually about half a dozen different kinds, including graphs related to location and recommendations. Whether he is right or not, one thing seems pretty clear: Facebook not only wants to own them all, but is well on its way to doing so.

Dixon starts off with a little math, by noting that the term “graph” refers to a “set of nodes connected by edges.” The biggest online graph of all, he says, is the web itself, where individual webpages are nodes and the hyperlinks between them are the edges. In social graphs, “the nodes are people and the edges friendship,” or the relationships between those people. Dixon also describes how Facebook’s graph is symmetrical (meaning both sides of a connection must agree) and how Twitter’s is non-symmetrical (since you can follow anyone regardless of whether they follow you back).

Then comes what you might call a taxonomy of graphs, which Dixon says include:

  • Taste: This is the kind of graph that Hunch is trying to create, by looking at questions that users have answered about a variety of topics (the company also has a demo that reveals what it knows about you based on your tweets). GetGlue and other services are also explicitly going after this graph.
  • Financial Trust: Payment services such as Venmo and even Blippy (which lets you share your purchasing habits) are interested in this graph, which relates to financial connections between people and companies. PayPal and other payment companies are also obviously focused on this graph.
  • Endorsement: Dixon says that this graph involves people recommending things — or other people — and uses the example of LinkedIn, which is trying to create an endorsement graph for people who are looking for work. Facebook is also going after one aspect of this kind of graph with its “like” button plugins.
  • Local: Companies and services such as Foursquare, Gowalla and Loopt are obviously targeting this graph, which creates relationships between people and other people — as well as people and services — based on their physical location. As Dixon notes, this graph is highly appealing to advertisers.

Which brings me to the next thought that Dixon’s post triggered: namely, that Facebook has a massive head start on owning virtually every one of these sub-graphs, with the possible exception of the “financial trust” graph — and with Facebook Credits rolling out, it’s likely the giant social network will get its hooks into that one soon as well. Certainly recommendation-based graphs powered by the “likes” of 500 million users could be fairly powerful. And when it comes to local, Facebook appears to be working on features in that area as well, although it’s not clear what form they will take.

One other thing that struck me as I looked at the different categories of graph is that Google is virtually absent from this picture. It is a giant web entity with a multi-billion-dollar market cap, and hundreds of millions of people use and rely on it every day, but apart from flawed experiments such as Buzz and Wave, it has no place in those relationship graphs — which might explain reports of ongoing internal pressure at the web giant to come up with something like Google Me.

Update: As Mahendra Palsule (@ScepticGeek) has pointed out on Twitter in a response to this post, Google does have a number of social graph elements in its arsenal, including the Social Graph API, which is similar to Facebook’s open graph protocol. Google scans for relationships between users and pages that use standards such as XFN and FOAF, and connects those with Google Profiles, and then uses that information in its social search results. But I don’t think there’s any question that Google’s efforts have been substantially less successful than Facebook’s in terms of mass adoption.

Related content from GigaOM Pro (sub req’d): Why Google Should Fear the Social Web

Post and thumbnail photos courtesy of Flickr user Luc Legay



There is a new social networking experience coming this fall. It’s called somethingcoolhappened.com. You can go there now to view a preview video and to preregister. With this site you can interact with friends, create your own unique avatars, upload videos, pictures or stories of something cool that happened to you or someone else! You also get full anonymity. It is going to be awesome! Check it out.

Think of this site as more like a combination of YouTube, Flickr, Facebook and then something totally brand new. With this site you will get your anonymity back, no more personal information floating around being sold off and you will get to be creative, compete with other people if you wish and just have fun in a great positive atmosphere.

Zach beauvai

I’m not sure I completely see the splintering of the social graph as a useful concept. All the subgraphs have at least one node in common… You.

This means they’re linked together, part of your network of social data. Not only that, but they’re linked further to the information giving them context: where they take place, on which site ( if online), under which area of interest or influenc’ personal or professional, etc.

Instead of seeing these as splintered graphs, I see them as parts of the whole, with different contexts. In technical terms, they’re linked but with different metadata.

I also recorded a podcast with Facebook’s David Recordon about their social graph, and the Open Graph Protocol. I was quite cynical about bf’s approach until I spoke with David. Now I’d say I’m cautiously optimistic that they’re at least considering he wider world and working with openness in mind.

Dr. Phil Hendrix

Maybe we should stop using the term social graph and even revisit the use of “friends?”

We regularly interview consumers on a wide range of issues, including their use of social networks. I find it interesting that the term “social graph” is used widely as if it has some meaning and relevance to users – while it may have a precise meaning in scientific circles, my sense is that it’s become virtually meaningless in many discussions. I have yet to encounter a single consumer who uses the term “social graph” – perhaps if we were to begin to describe social networks in ways that are more descriptive, meaningful and relevant to users, we might make more progress in helping them manage their relationships. By the way, the same goes for the term “friends” – while it once had a specific, widely shared meaning among individuals, the term has since been co-opted to refer to virtually any relationship. There is clearly need and opportunity for someone to innovate and disentangle the mess that our online social networks have become. Chris Dixon’s suggestions are a step in the right direction – I’d welcome thoughts on other ideas and solutions that have merit.

Dr. Phil Hendrix, immr and GigaOm Pro analyst

Dr. Phil hendrix

Rethinking how we define and manage social networks…
In a recent presentation (“The Real Life Social Network;” http://bit.ly/b6A03x) Paul Adams (http://twitter.com/padday) summarizes Google’s research on challenges and opportunities in managing social networks. Adams distinguishes between strong, weak and temporary (transient) ties and shows convincingly how difficult it is at present to “manage” and keep these networks separate (as we do in real life) online – e.g., on Facebook, Twitter and other social networking platforms. Excellent discussion, well worth studying (all 216pp) and considering the implications for Facebook, Google and others who aspire to help us “manage” our social networks. None of these companies have yet developed even adequate solutions.
Dr. Phil Hendrix, immr and GigaOm Pro analyst

simon hamer

I still feel that you cannot track perception particularly well, just by links, as one other comment mentioned consideration needs to be given to the comment attached.

Also Google will downgrade too many links to one site where it is seen to be a collective, so the rankings of those sites if great can work against the social graph.

I wish Facebook the best of luck with it.
It will need it.

Chethan Kashyap

This is interesting, but not surprising. Its happening quite as expected by most tech-savvy people. Ever since Facebook became the face of social networking community, it has been trying hard to take-over virtual world and posing stiff competition to Google.

Joe Lazarus

I’d argue that Google’s business is based almost entirely on a “graph” that Chris describes. Google’s search engine ranks results based on links from one site to another. Those links represent the largest “endorsement” graph in the world. If I link to your site from mine, I’ve cast a vote for your site in the Google index. The text I use for that link provides additional insight around the nature of my endorsement.

Mathew Ingram

I agree, Joe — and Chris mentions that the web is the original graph. But it’s not an explicitly “social” graph in that sense, I don’t think, although it starts to become one when it is filtered through apps and services like Digg and Twitter.

Steve Ardire

all this fretting about Facebook ( massive head start on owning virtually every one of these sub-graphs ) is alarmist because they are overrated.


All (well most) of the social graphs are “self” organizing.

Google build their system on top of an existing graph as an analyzing system. Hence their mistake with buzz, trying to force a graph on people(we know how your relations are organized,we did an analysis, uuups), instead of allowing it to form.

It will be easier for Facebook to build an analysis system for their graph[s], than for Google to build a system to allow for self organization[no data, no grasp how it should work].

Mathew Ingram

That’s a really good point, Ronald — Chris made the same point, that Google tried to create a graph from people’s email contacts and that didn’t work. I think most successful graphs emerge from people’s relationships naturally, rather than being imposed on them. But Google isn’t built around relationships.


Facebook is trying to find anyway possible to increase interaction on their website


Face book in my opinion is just a trap for marketers to get you data. When they say both sides must agree it means to read the 20 page terms and conditions that reads like a thesis or dissertation for rocket science. I am not a facebook fan and even myspace is a spam party



you just tricked me into commenting on your comment. are you a mass push marketer?

Comments are closed.