watts – gladwell smackdown on diffusion

Popularizers par excellence Duncan Watts and Malcomb Gladwell go mano a mano in a Fast Company article by Clive Thompson over mechanisms of network diffusion.  Watts attacks Gladwell’s key idea that “influentials” are the drivers of viral diffusion effects:

Why didn’t the Influentials wield more power? With 40 times the reach of a normal person, why couldn’t they kick-start a trend every time? Watts believes this is because a trend’s success depends not on the person who starts it, but on how susceptible the society is overall to the trend—not how persuasive the early adopter is, but whether everyone else is easily persuaded.

Ed Keller rides in to defend his (and by extension, Gladwell’s) argument:

“They’re fonts of word of mouth,” Keller insists. And ahead of the curve, too: In the 20 years he has been polling them, Keller has found they began using computers, mobile phones, and the Internet years before the mainstream.

I take Keller’s and Gladwell’s points to be that social space is heterogeneous and the identity (not just the structural position, e.g., being a broker) of an individual matters in explaining the effect that individual has on diffusion within the network.  Indeed, I’ve made that argument myself.

Yet, Watts is on to something here.  Particularly if you are interested in whether a practice or idea or innovation is incorporated into behavior (and I’d guess that most of us who read orgTheory are), and not simply in whether an idea spreads ephemerally.  It suggests parallels with the concept of political opportunity structures in social movement theory or even to the importance of Stinchomb’s “liability of newness”.  People and organizations are embedded, not just in a set of relationships, but in a set of roles and rules governing those relationships.  So, in order for an idea to spread and take root, the roles and rules implicated by an innovation need to be buttered up (or undermined) first. Very few studies have pulled off the dual challenge of examining structural diffusion along side these broader institutional, cultural and political opportunity structures simultaneously.

For instance, take a look at this movie showing the diffusion of the name “Loic” across France in the 20th Century (sorry, the movie can’t be embedded).  The correspondence between geographic diffusion and interpersonal diffusion is not one-to-one and naming a child is more complicated that posting a YouTube video, but it’s close enough to make the point:  If the diffusion was influenced primarily by Influentials, one would think that the tipping point would happen when the name hits Paris.  From there it should spread through the capital’s hold on national media and on status hierarchies.

But that’s not what these data say to me.  To my eye, Paris gets enveloped like the rest of the country in the name’s steady southeasterly spread.  Its plausible to conclude (and I think consistent with Watts’s argument) that naming a kid takes a bit of getting used to; you want to have heard of the name for some time to feel comfortable attaching it to your child for the rest of his or her life.  You want to know that the people he or she interacts with won’t think the name is “weird”.  So even if it happens to be popular in Paris this year, you want to see it closer to home so you can get used to it.  There needs to be an interaction between the network effect (exposure) and the social environment (implementation) to produce the observed pattern of diffusion.

What is that interaction?  Is this merely a matter of density (a pop ecology approach) or is there a more cultural or cognitive or relational dimension that makes people comfortable enough to accept the name?  These are open questions and I, for one, would like to see a study that sorts this out (or maybe I should go for it myself).

Ultimately, the thing I agree with most in the article is Gladwell’s reply to the whole kerfuffle.

In the end, though, I suppose that I feel the same ways about his insights as I do about Steve Levitt’s disagreements with me over the causes of the decline in violent crime in the 1990s. I think that all books like The Tipping Point or articles by academics can ever do is uncover a little piece of the bigger picture, and one day—when we put all those pieces together—maybe we’ll have a shot at the truth.

In other words, this is a difference without a distinction: both can, and indeed are, going on.  The point Gladwell makes is universal though, isn’t it?  In fact, I’d say that sums up my thinking on academia generally.


Written by seansafford

August 13, 2009 at 7:29 pm

18 Responses

Subscribe to comments with RSS.

  1. keller is giving a non sequitur argument. even if it is true that highly connected people tend to be early adopters this doesn’t necessarily indicate a causal connection but could just be a case of what Lieberson calls “riding the curve.”

    one possible benefit of the Watts argument is that maybe marketing surveys will stop including a bunch of leading questions asking if the respondent is an opinion leader.



    August 13, 2009 at 10:12 pm

  2. “Gladwell’s key idea”?? Katz and Lazarsfeld, we hardly knew ye.



    August 13, 2009 at 10:48 pm

  3. Touché. Point taken.


    Sean Safford

    August 13, 2009 at 11:38 pm

  4. Ezra, you’re clearly forgetting that Steve Levitt and Malcolm Gladwell invented most of social science between about 1999 and 2006…as far as the media are concerned.



    August 13, 2009 at 11:50 pm

  5. Thanks for the link to my blog. I’m sorry that the movie can not easily be embedded ! The FLV source is located here :
    (the dataset is “le fichier des prénoms”, INSEE, the maps were created with R and the animation with ImageMagick)
    I think that, with French data on names (“le fichier des prénoms”, INSEE and “l’enquête emploi” INSEE) one could appreciate the variety of diffusions : local influence (the parents learn to like new first names given to local babies), class influence (from 1945-1980, fashionable names came from the managerial classes and trickled down to the workers and the peasants…), “town and country” influences (during the 19th century, innovation came to the country by way of internal migrations : migrants returning from Paris to “la province” came back with new names)…
    Today a series of “new” names in France seem to come from the geographical “margins” : mediterranean names such as Carla or Marius (frequent in the Riviera, they climbed up to Paris), “nep-celtic” names from Brittany, pseudo-basque names from French “Pays Basques” (such as “Maylis”/”Maëlis”…).

    Last point : it would be very difficult to pinpoint individual influences with current datasets.



    August 14, 2009 at 7:18 am

  6. @ Ezra: I will say that Watts’s piece is no less Mertonian than Gladwell’s. He had the whole picture from the get go.

    @ Baptiste: thanks for doing such interesting work! I saw it highlighted on Graphic Sociology and have been chomping at the bit to trot it out on OrgTheory ever since. Merci.


    Sean Safford

    August 14, 2009 at 6:48 pm

  7. Which piece did you mean? If you mean the idea that successful ideas/products/innovations are more about fitting with the zeitgeist than getting endorsement by successful people, I’m not sure I see it in Merton but it certainly is in Lieberson. And anyway, Gladwell is right that Watts is attacking a straw man.

    (BTW, I didn’t say Merton. I said Katz and Lazarsfeld)



    August 14, 2009 at 7:09 pm

  8. Correct again. Anyhow, I was referring to Merton’s version of roles and rules. I’m not sure I get the “zeitgeist” aspect of Watts. I’m more of an institutionalist than that…


    Sean Safford

    August 14, 2009 at 7:13 pm

  9. Well, I’m using the term loosely. Lieberson shows that prominent people (e.g., movie stars) have little influence over which names are popular, and that when you look carefully at cases where they do seem to have influence, what is really driving things are deeper trends– e.g., it is not the fame of Sean Safford that is responsible for the wave of Seans spreading across the land as much as it is the fact that boys’ names starting with ‘Sea’ have rising in popularity, and so we are all primed for ‘Sean’ to take off. So if your name was Ezra Safford, it just wouldn’t take. :-)



    August 14, 2009 at 7:23 pm

  10. […] Malcolm Gladwell, responding to criticism of Duncan Watts: In the end, though, I suppose that I feel the same ways about his insights as I do about Steve Levitt’s disagreements with me over the causes of the decline in violent crime in the 1990s. I think that all books like The Tipping Point or articles by academics can ever do is uncover a little piece of the bigger picture, and one day—when we put all those pieces together—maybe we’ll have a shot at the truth. (via orgtheory) […]


  11. This debate reminds me of Bourdieu’s critique of Weber’s concept of Charisma.

    Contrary to Weber, Bourdieu argued that charisma is not just the attribute of a single individual; rather, the charismatic effect of the few is made possible because of the relations comprising social structure at a point in time. Prophets,for example, emerged in times of war. The charismatic “influential” was not an extraordinary man, but – in his words – a “man of extraordinary situations”.

    I think this is a more nuanced position than Watts. It doesn’t deny the impact of influentials, but it focuses on the social structure and network forms that might explain why these special few are influential in the first place.



    August 15, 2009 at 4:23 am

  12. Ezra: since you are the one with tenure and also the one getting hounded by the media as the voice of economic sociology, wouldn’t that suggest that Sean is the barrier I’ve been facing all my life to orgtheory fame and fortune? I should consider changing my name to the altogether more distinctive sounding Ezra Safford.

    In fact, it would hardly have been implausible. Names are also path dependent and, in fact, Ezra would fit the path pretty well for my family: my ancestors’ names since Saffords arrived in North America: Christopher, Josiah, Thomas, John, John, Elisha, Darius, Burt, Winfield, and John. Ezra would have worked in the New Testament-Old Testament back-and-forth pretty well, right?

    But apparently the actual story is that my parents wanted to name me after my father but didn’t want to call me John Jr. So they thought of naming me after my mother’s grandfather who was named Ivan — a cognate of John — but decided that Ivan would be “tough” (i.e., uncool, lacked legitimacy, not enough groundwork in the zeitgeist). So they chose another cognate: Sean. Why? You guessed it: Sean Connery.


    Sean Safford

    August 15, 2009 at 1:42 pm

  13. Ezra: I think Watts is not attacking a straw man and you misunderstood Watts’ argument.
    Watts’ critique is not about whether influencers matter or not. His point is that large scale diffusion by word of mouth/viral marketing can’t be predicted let alone engineered by, say, targeting certain individuals. This is because what determine a widespread diffusion is not so much about the starters but more about the social interactions that occur as the diffusion progresses.



    August 15, 2009 at 7:52 pm

  14. Great story, Sean. (Or should say “Safford. Sean Safford.”) But, um, I don’t think the media is hounding anyone about economic sociology, least of all me.

    Salam: Is there anyone worth taking seriously who says that “large scale diffusion can be predicted or engineered by targeting certain individuals”? If not, you have yourself a man of straw.



    August 16, 2009 at 1:07 am

  15. On a related note — here’s Andrew Hoffman saying that we should put Malcolm Gladwell out of business:



    August 16, 2009 at 1:31 am

  16. Ezra: as someone who once worked in ad/media i know hundreds of millions of dollars spent each year based on this very idea: targeting certain individuals to trigger large scale diffusion.

    you’re right that most academics don’t take them seriously, but still, they spent real money on something that have no scientific evidence supporting it.

    of course this occurs not only in advertising business. but the point is that something that’s considered as a straw man could be a real business issue and since no one takes it seriously it just then perpetuates itself.



    August 17, 2009 at 3:19 pm

  17. tf – thanks for the link.

    A popularizer like Gladwell definitely simplifies much in order to reach out to the public. This is not necessarily bad, in fact it can be conducive for debate as the Watts-Gladwell incident shows.

    But are there are disincentives in place that prevent academics from become a Gladwell? There probably are some status penalties. I note the academics turned popularizers are usually securely tenured and already have a good reputation.

    In some sense, isn’t this a question on the diffusion of innovation across fields (academia to popular culture)?



    August 18, 2009 at 3:02 am

  18. […] down the status chain. Gladwell also promotes something like that, and it just so happens that OrgTheory is covering a dispute between him and fellow-popularizer Duncan Watts. Watts thinks the […]


Comments are closed.

%d bloggers like this: