Thank you, Dave Winer

I tried to ignore the term “social graph” when it first started popping up a few weeks ago. For one thing, it sounded like some sort of embarrassing disease; for another, the idea of having to figure out some arcane new Web 2.0 term was depressing. But the term – or is it a meme? – kept on proliferating, so last week I tried to figure out what it meant. The definitions I found, explicit or implicit, seemed to indicate that a social graph was all the connections between people in a community or, alternatively, all the connections that one person in a community had with all other people. Why is this a “graph,” I wondered to myself sheepishly, and what’s the difference between a “social graph” and a “social network”? Clearly, I didn’t “get it.” The shame was intense.

Yesterday, Dave Winer came to the rescue, with a post titled “How to avoid sounding like a monkey”:

before we talked about social graphs we called them social networks, and you know what — they’re exactly the same thing, and social network is a much less confusing term, so why don’t we just stick with it? (Answer: we should, imho.) So if you don’t want to sound like an idiot, call a social graph a social network and stand up for your right to understand technology, and make the techies actually do some useful stuff instead of making simple stuff sound complicated.

Most sensible thing I’ve read in ages.

14 thoughts on “Thank you, Dave Winer

  1. Tom Lord

    This might help:

    A “network,” to a computing systems person, is system for passing along messages. This jives with the ordinary english usage: “Joe needed to identify candidates for the CFO position so he put out a call to his network of chums from college.” Networks can be, and usually are, mostly opaque to most people who use them: you know who you are directly connected to and know some of those you can reach via the network but you don’t know how all strangers on the network are connected.

    So, a network is a thing that people can join and, having joined, they can transmit messages on the network.

    A “graph,” on the other hand, as it is being used here, is simply a picture of formally defined, (usually) asymmetric relations between entities — betwen people in this case.

    In other words, Joe is in my social network because as it so happens I can talk to Mary and Mary can get a message to Joe. On the other hand, Joe is on my “social graph” because in my data I declared “Tom Friend-of Mary” and in her data Mary wrote down “Mary Friend-of Tom” and “Mary Acquaintence-of Joe”.

    Us having written down this data, now someone else — let’s say, Tim, — can come along and write programs that study all of that data. He can write a program that will print out things like “According to this data, people from Harvard don’t like people from Stanford” or “Mick, as it turns out, is the most popular person.”

    “20% of the new designer jeans for anyone who has been ‘friended’ by a friend of the band, or a friend-of-a-friend.”

    “Job hunting tip of the week: Pay attention to your social graph profile! Employers are impressed by candidates who are friended by known performers in the industry but those links from your friends in the neighborhood pub crawl club may come back to haunt you!”

    Obviously a “friend-of” declaration in a database is not a friendship — it is just a piece of data. So what is going on?

    These guys are trying to monetize attention using a new kind of currency (the “link”), then to try to give this currency exchange value by exposing the graph of transactions for global analysis.

    The reason the topic is coming up these recent weeks is because of a technology announcement from Six Apart — they are offering up an open standard for this graph. This is in contrast to the way the graph is implemented at, say, LinkedIn where the graph is closely held by the provider (and in that case, keeping the graph private is one of their main services!) An open graph allows third parties to “spider” graphs, regardless of where those graphs are hosted. Thus, the exchange value of your open social graph links using Six Apart’s stuff is supposed to be higher than you could obtain if you kept your graph locked up.

    So, yeah, these guys have gone off the rail.

    Once upon a time, we built “social networking software” so that we could have human-mediated ways to route messages between people.

    Now we have “social graph software” so that big companies can visualize us in demographic categories and relations and inspire us to voluntarily generate data to help refine their models.

    It’s the most dangerously narcisstic, cyncical, crap-of-negative-utility to come along in quite a while.


  2. Kendall Brookfeld

    This sounds like one of those instances in which people proudly proclaim their ignorance of math or computer science, when they’d never do so for another subject, like reading…or Scientology.

    Graph theory is taught to undergrads in computer science, and probably ought to be shown briefly even to business students, at least at the more pretentious schools, or to people who write about IT.

    Here’s the Wikipedia entry on graphs

  3. Marius

    Nick, I found this pretty interesting, so I posted my on thoughts about it in Smart Mobs. Thanks for inspiring me. Same thanks, indirectly, to Dave Winer. Your comment seems to me very informed, Tom.

  4. Marius

    Yours too, Kendall. This is why expertise in each field is needed. To make good differences between things, so they are not confused or mistaken. Well, at least as much as the progress is some science allows it.

  5. finin

    I’m with Kendall on this. If you (or our software system) is extracting, representing, visualizing, analyzing and/or manipulating date about people and their relationships as abstract graph models or as graph data structures, it seems very reasonable to use the term ‘social graph’ rather than ‘social network’.

  6. Tim

    “Why is this a “graph,” I wondered to myself sheepishly, and what’s the difference between a “social graph” and a “social network”? Clearly, I didn’t “get it.” The shame was intense.”

    I tend to agree with the comment about “I don’t know anything about maths, me!”

    I’m all in favour of getting rid of stupid words like blogosphere and crowd-sourcing, but graph is a perfectly acceptable word (with a long history) to use to describe the phenomenon being discussed. Claiming that people who use correct terminology are somehow trying to confuse you or make you feel stupid seems, well, depressing, actually.

    Plus, you broke the Internet Rule. Never thank Dave Winer. It just encourages him.

  7. Tom Morris

    I don’t agree that it’s willful ignorance of mathematics. The fact is that the phrase ‘social network data’ describes everything that ‘social graph’ does without having to explain graph theory to those who don’t understand it.

    I find it ridiculous that people are banging on about the social graph stuff like it’s new.

    It’s not like nobody figured out that this was a problem back in 2000. Not like anyone came up with a decent enough standard to describe a solution (FOAF). I’m betting Dan and Libby will have their history rewritten, and it’ll all be a magical 2007 discovery.

    Welcome to the party, folks. I wonder what new (that is to say, not very new at all) stuff there is for the Web 2.0 crowd to discover. Because it’s all about innovation.

  8. Dave


    dave winer is the cheerleader for simplifying geek talk & standards?

    um… kettle, black.

    and while we’re at it, maybe he should stop saying RSS and just call it XML too:

    if anyone should have a fit about using “social graph” instead of “social network”, it should be Reid Hoffman (or maybe Adrian Scott). but Reid doesn’t seem to have a problem with it. so why do you guys?

    seriously, get over yourselves already.

    and while you’re at it, come to my conference… titled, ahem:

    – dave “you say toMAYto, i say toMAHto” mcclure

  9. Mike

    I’m not a terrific fan of Winer’s but I hope that someone has the balls to point out that it’s How to avoid sounding like “A” monkey not “an” monkey.

    Second, I see 1 social network as “Facebook” or “MySpace”. I don’t think of my social networks as networked together. When I consider what is my (sum total) Social Network, I don’t calculate “MY” social network as MySpace + LinkedIn + Ryze + FB + whatever; I think of them as separate social networks. I have a social network in MySpace. I have a (maybe somewhat overlapping w/ the same people) but disctinct social network in FB, LinkedIn, etc.

    Thus I consider the social graph more of a layout of relationships between the social networks (size, relativity, usage, importance, etc – all very subjective of course) but the graphs showing MySpace and FB in the center with Ning and whatever on the outskirts seems reasonable to me.

    Why would Adrian Scott care about this McClure?

  10. Sergey Schetinin

    It’s ridiculous how many comments such a throwaway post generates, while some of the most insightful ones barely get any comments at all. Clearly an illustration to the fact that people prefer discussing things that don’t matter.

    I know math (was a math major in uni) and I too think the term “social graph” doesn’t bring anything to the table.

    Regarding the mentioned Six Apart announcement, FOAF spec was around for a long time, but not a single useful application based on it. I guess there’s inherent flaw in thinking that social graph / network data is useful at all. Just look at the examples Tom gives.

  11. Charles Cameron (hipbone)

    I imagine we each bring to this discussion what we’d one and thought before, and that’s what makes the comments — and the divergences — worthwhile.

    I used graph in the first sense (wavy or jagged line plotted against axes, etc) for years, but was introduced to the second sense (nodes with lines connecting them, which I learned to call edges) when I dreamed up a family of games that involved playing “ideas” on nodes and asserting analogical or other links between them along the edges.

    Now I am intrigued by the wide variety of networks of ideas represented by graphical nodes and edges in a wide variety of fields, and compare a few of them at:

    For me, the similarity between, say, argumentation maps, social networks and my games is a feature of note, since I think we are moving from linearity in thinking towards thinking in ways that model complex networks and interactions — because that’s the basic ‘shape’ of the problems we face.

    And “graph”, in the second (nodes and edges) sense, is the proper word for that *when diagrammed*, as “network” is when conceptualized.

    So let’s use “social graphs” to mean diagrams of people connections, and “social networks” for the

    kinds of linkage between people which they map.

  12. Phil

    Why is this a “graph,” I wondered to myself sheepishly

    BT, DT. Me, I’ve spent 12 years in higher education and another 12 working in IT. I read four languages, I know pi to 75 places, I’ve been published in the national press and my entity relationship modelling skillz are l337. However, I’ve never studied computer science, I gave up maths at the age of 16, and my IT career never involved responsibility for networking. So when I hear ‘graph’ I almost always think of one of these, not one of these. (Incidentally, the first page of the Google images search brings back one nodes-and-vertices graph and 19 examples of plots and charts (what ignorant people like me and Nick call ‘graphs’).)

    Tom L – ouch. That should be a blog post.

  13. Lisa

    Not sure I agree.

    “Social network” has several meanings so is not quite so useful mathematically.

    “Social graph” means exactly that and anyone familiar with graph theory can understand it instantly with the precision the author/speaker wants.

  14. tlrobinson

    At it’s heart, a “social network” IS a graph.

    So, you say I can’t call them “social graphs”? Fine, I say you must strip “social networks” of all their aspects relating to graph theory. Whoops, no more social network.

Comments are closed.