Dave McClure Is Wrong, Continued: Social Graph v Social Network

Dave McClure responded to my recent post about the redundant and unhelpful 'Social Graph' meme. I maintain that the 'Social Network' term -- which is in wide use and well-understood -- is fine. Dave doesn't agree:

hey stu -

of course you're entitled to your opinion, however i'd still suggest that 'social graph' is more specific & nuanced than 'social network' -- specifically the idea of a graph (from mathematics graph theory) is a series of nodes connnected by directed vertices with weights/attributes. this is not the typical definition of a social network, which altho similar, is usually just described as my list of friends.

i made the analogy with XML & RSS, as i believe this is a similar situation -- RSS is a more specific & nuanced implementation of XML, that emphasizes a standard for content publishing (rather than a more generic structured data standard).

so i disagree the 'social graph' term is more confusing -- in fact, quite the opposite: it's more specific.

i can understand why some folks feel it's the same thing, and prefer to use the earlier 'social networking' term instead. that's fine & i don't have any problem there... it's still accurate, just not quite as specific in some cases.

that said, i don't agree with Winer that using the term 'social graph' is somehow incorrect or confusing, or that we should do away with the term... any moreso than we should do away with using RSS.

my .02,

- dave

[Note: the 'Stu' thing is a small joke, since I introduced him as 'Doug McClure' at the Office 2.0 panel session I chaired. Apologies for that, again, Dave.]

The notion that 'Social Graph' is more 'nuanced' that 'Social Network' because the former term comes from the study of mathematics is just fluff. The term social network comes from decades of research -- in anthropology, sociology, and related fields -- and is supported by a corpus of articles, books, and a rich collection of mathematically-defined terms. The study of social networks has been going on for over a century, and the term has been in use since the '50s. The work of dozens of social network theorists, like Mark Gronovetter, Anatol Rapoport, Stanley Wasserman, and Duncan Watts has supported the richness of our growing understanding of social networks and network theory.

The biggest weakness of McClure's comments is that a Social Graph is "a series of nodes connnected by directed vertices with weights/attributes. this is not the typical definition of a social network, which altho similar, is usually just described as my list of friends." However, the basis of social network study is all about the relationships between individuals, modeled as nodes in a network (a form of a graph), with specific sorts of attributes, many of which are weighted in various ways. For example, a few terms that I love for their poetry and precision:

[from Wikipedia entry on Social Network]

Betweenness

Degree an individual lies between other individuals in the network; the extent to which a node is directly connected only to those other nodes that are not directly connected to each other; an intermediary; liaisons; bridges. Therefore, it's the number of people who a person is connected to indirectly through their direct links.

Closeness

The degree an individual is near all other individuals in a network (directly or indirectly). It reflects the ability to access information through the "grapevine" of network members. Thus, closeness is the inverse of the sum of the shortest distances between each individual and every other person in the network.

Centrality Degree

The count of the number of ties to other actors in the network. See also degree (graph theory).

I liked Centrality as a term so much, I used it as the name of a blog I worked on for several years, sponsored by Visible Path.

It makes me think that Dave hasn't read any of the basic works on social networks, and maybe not even the Wikipedia entry. Or maybe he was confused, and typed in www.wikipedia.com/social_graph which directs you to the entry on Social Network, since there is no entry for Social Graph.

I agree that social network theory is related to the mathematical study of graphs, which is a support for what goes on in human social networks. However, these can be used to define a wider range of phenomena, such as the behavior of computer networks, ecological relationships of organisms, and problems posed by topology.

The most generous interpretation of his argument -- that social network has become conflated with a buddylist -- may have a speck of merit. But I think that the generally accepted and understood use of social network includes a great deal more richness. People understand the six degrees of separation concept, and the Kevin Bacon game is a commonplace. I don't buy it.

Finally, it is just wrong to advocate dropping the use of a richly described and rigorously researched term like social network for a less constrained and not particularly relevant term like social graph. It is just nuts, without some strong rationale. If someone could make a case for the application of a specific insight from graph theory to social networks -- like Cayley's work in theoretical chemistry, or the Four Color Problem in topology -- would at least create some rational for considering a new term, or redefining the old one. Or arguing that the body of research that has been called social network theory lacks some critical elements that only a new study, social graph theory, explains some real-world phenomena better would at least be worthy of discussion.

However, simply stating that the 'social graph' meme holds some nebulous mystical significance is not science. It's not even good logic. It's just some kind of sketchy marketing mumbo-jumbo that benefits no one except the advocates who might derive some ripple of attention while the term is passed around.

I am no respecter of persons, as those who have read my work can attest, so I am not making this case for the sake of a bunch of academics. I am an advocate for clarity and simplicity in discourse, however, so I will continue to make the case that 'Social Graph' is at best a synonym for 'Social Network' and at the worst is a marketing ploy or a publicity stunt.

In this fissile age, where a ideas can become money in next to no time, we have seen several extremely profitable and beneficial conferences rise through the introduction of models (and associated terms) that describe new agencies and forces at work in the tech landscape. The stunningly successful Web 2.0 conferences, that generate millions now every year, and the younger Office 2.0 and Enterprise 2.0 events are all good examples. However, the memes that motivate the shows actually help us to understand things better, because metaphors can be powerful tools for understanding the world. (Making a case for Social Networks 2.0 could have been a possibility, but one that was rejected or didn't occur to the advocates of the Social Graph meme.)

On the other hand, metaphors are not magic. They can be unhelpful or just dumb. While building a conference around 'Social Graph', as Dave has done, does not doom the event to failure just because the metaphor is unhelpful. On the contrary, I think the upcoming conference -- focused specifically on Facebook's growing ecology -- is a great idea. I will be attending. Likewise, I plan to attend the next Social Graph event, schedule for 2008, now an O'Reilly event. But the success will come without the metaphorical help that Web 2.0 and its derivatives have offered us.

I recommend that Dave and O'Reilly consider a debate on the term, at one or both of the eponymous events, since they now are the term's leading proponents. I would be happy to be one of the voices on the other side, and I am sure that others would be glad to join in.

Comments

[Dare Obasanjo]

This post was for people like you and Dave - see http://www.25hoursaday.com/weblog/2007/08/30/TheDifferenceBetweenASocialNetworkSiteASocialGraphApplicationAndASocialOS.aspx

PS: It's interesting to see how there seems to be no overlap between your circle of bloggers and the ones I read given that you think the term social graph is new.

[Dave]

Stowe -

(minor nit: if you use my comment as the lead in your post, i'd appreciate if you could also approve the comment *in context* on the earlier post where i submitted it, so that we can continue that conversation. i almost missed this followup this morning)

ok: so while i may be less informed -- or hell, plain wrong -- about the historical body of work on the term 'social network', i doubt that is the notional understanding 98% of folks have when they use it. i'd still maintain most people think of the term to simply describe their friends list / buddy list, and that using 'graph' to distinguish something different or more specific is a useful artifice, but perhaps i'm incorrect. why don't you ask 5 people (who aren't uber geeks) what they think of the term, and see what the reaction is... i'll do the same, and we can compare notes. i think you & dave winer are being a little too ivory tower about terms that have entered standard parlance, and might have a different meaning on the street.

in any case, even if in the most narrow & pessimistic reading of the term -- suppose that Facebook *IS* simply trying to re-brand the term 'Social Graph' to suit their own purposes -- then is that really such a huge sin? how many times before has that happened? and as i mentioned, when RSS first came out, i sort of thought to myself... "so what's so different about this and XML"? over time, i understood the distinction, but at first it seemed like Dave was just trying to rebrand someone else's earlier definition.

this reminds me a lot of the discussion around the term 'Web 2.0' that began a few years ago... everyone was arguing so hard that it didn't mean a damn thing, that you had to believe it DID mean something if only because so many people were wasting their breath.... as we are now.

so whether or not i'm a complete idiot on the 'official' definition -- i'll widely admit i'm an idiot on many subjects, it should be no surprise on this one as well -- does it really matter, if enough people decide to say it often enough with a similar sense?

anyway, i've probably already wasted more brain cells on this than i have to spare. happy to continue the conversation at the conference, and if at the end you've convinced me... well then, i'll just have to rename my conference Networking Social Patterns ;)

- 'doug' mcclure

[Bob Warfield]

Wow, Stowe, lots of words, lots of emotion, you're fired up! That's good, passion is helpful in most cases, even if it can be off putting to some. I think, though, that you're reading a lot into McClure's note that I don't see there.

First, he isn't the one seeking to eradicate one particular way of talking about the concept. He specifically says its fine to use either form, he just likes one better. I have heard that position from all the Social Graph proponents I've read so far. It seems to be the Social Networkers that want to eradicate the Graph, but I can't for the life of me see why.

Second, it is ironic, but you guys have exactly the same confusion about each other's terms and the background surrounding those terms. McClure finds Social Network to be too imprecise because most people just see it as a list of friends and "real scientists" (my term) think about graphs. In this case, McClure is familiar with the literature of computer science, where in fact he is right and we should be talking about graphs not networks when speaking in the context of algorithms. That crowd will titter just as much about the "network" term as your crowd titters at "graph". It is the laypeople who McClure sees as using the term "Network" and doing so with less precision, hence his preference.

Your position is the mirror image, and perhaps its why this is igniting such a debate. You find Social Graph to be imprecise because it ignores your decades of research from the social sciences wherein the term Social Network was preferred. For you, the laypeople are those who haven't read this literature, and they would choose "Graph" out of ignorance and out of a desire to glorify their views with "sketchy mumbo jumbo". See what I mean about the mirror image?

But Stowe, there is no sketchy mumbo jumbo here in using the term "Graph". It's quite the opposite. I have a far easier time finding the concrete examples you crave (the equivalent of the four color problem) around the term Social Graph. The reason? Because of all the discussion about how to create an Open Social Graph API. Now surely Computer Science's nomenclature should have some standing when we're designing API's, no? Don't we need to discuss how we're representing a Social Network (using your term and view of the term) inside the computer? Wouldn't that representation be a Social Graph? And isn't that how McClure and the others use the term? That's why they're prepared to let the two stand together.

What about all that Social Network science? That's the one that seems to be still struggling to catch up. Where are all those learned articles with respect to what's going on here on the web right now? They're potentially useful, but how have they informed the design of this software to date? They seem to be almost absent. The exception I see is some writing by Andrew McAfee. Now you can make a good case that perhaps these works must be consulted, but it's just not there so far. Perhaps if we could see more of those writings it would be easier to get fired up about all this, but for now, it seems a tempest in a tea cup to many of us watching the great diatribe.

Tell us not what you want us to call it, but what it is that we may finally see it clearly. Leave aside the name calling and the appeal to decades of scholarship. Show us the result of those fruits. Bring that knowledge to the table with as much passion.

Don't you agree that's a far more interesting discussion?

[Jeff McNeill]

Aloha Stowe, I have to agree, especially as someone who has waded through Rogers & Kincaid's "Communication Networks" as well as Granovetter, Wasserman, etc.

The main thing for the use of a new term, is does it illuminate a new phenomenon. Social Graph doesn't seem to do that, as far as I can tell.

As far as "representing social networks" inside of computers, that doesn't need graph theory other than in computing connections and visualization.

I have to say that the comment about the communication network research not informing current design is like saying "Well, Plato doesn't mention MySpace, how relevant can he be?" It's a matter of picking up these resources and (possibly) gaining illumination thereby.

Cheers,
Jeff McNeill

[Joel Helbling]

It seems to me that the term "social graph," as it is being used these days, is an abstraction for the social network. When I say "abstraction" I mean it in the programmer's sense of the word: when designing programs, the programmer must model real-world systems with a logical system which is meant to be an abstraction (or representation) of the real-world system.

It is important for the programmer (or any other system modeler) to distinguish the abstraction from the real thing. Social networks happen everywhere, with or without the Internet. Social graphs exist within systems designed to model those social networks, systems whose purpose may be to enhance (or even spawn) real-world social networks.