Welcome to the bloggy home of Noah Brier. I'm the co-founder of Percolate and general internet tinkerer. This site is about media, culture, technology, and randomness. It's been around since 2004 (I'm pretty sure). Feel free to get in touch. Get in touch.

You can subscribe to this site via RSS (the humanity!) or .

How Influential are Influentials?

I don’t often do book reports around here, but I just got through Duncan Watts and Peter Sheridan Dodds’ paper, “Influentials, Networks, and Public Opinion Formation” [PDF] and thought it might be worth sharing some quotes and thoughts (especially since it’s 36 pages of fairly dense material).

As I wrote recently their basic thesis is that so-called “influentials” are not all they’re cut out to be (especially by people like Gladwell and Keller). Though as they explain in the conclusion, “Our main point, in fact, is not so much that the influentials hypothesis is either right or wrong, but that it’s micro-foundations, by which we mean the details of who influences whom and how, require very careful articulation in order for its validity to be meaningfully assessed.” While Watts and Dodds’ own work leaves me with some questions, this seems like a hard assertion to argue with. To come up with a true theory of influence, the details of influence need to be universally defined and understood.

In fact, I don’t know that Watts and Dodds go far enough themselves, mainly because influence is so hard to pin down. Observationally, who influences whom and how can change on a daily basis and greatly depends on things like topic & relationship (as well, I’d argue, on outside factors like how busy the recipient is at time of influence). Watts and Dodds do acknowledge these factors, however, suggesting that “large scale changes in public opinion are not driven by highly influential people who influence everyone else, but rather by easily influenced people, influencing other easily influenced people.”

This, in and of itself, doesn’t seem particularly controversial. If you go with the idea that 10% of the population is influential, that leaves 90% of the population that’s not. Then if you assume that, especially in the current media/advertising landscape, the influential 10% is hardest to reach because they are the most overexposed (and thus have their attention stretched the thinnest), it seems that your effort may be much better spent thinking about options. What’s more, according to Watts and Dodds’ research, while “influentials have a greater than average chance of triggering critical mass, when it exists … [their effect is] only modestly greater, and usually not even proportional to the number of people they influence directly.”

As they explain in their conclusion, the simplest way to understand this is to look at natural analogues such as forest fires:

Some forest fires, for examples, are many times larger than average; yet no-one would claim that the size of a forest fire can be in any way attributed to the exceptional properties of the spark that ignited it, or the size of the tree that was the first to burn. Major forest fires require a conspiracy of wind, temperature, low humidity, and combustible fuel that extends over large tracts of land. Just as for large cascades in social influence networks, when the right global combination of conditions exists, any spark will do; and when it does not, none will suffice.

Upon reading that I was immediately brought back to something I wrote about last year. My thesis in that entry was that the marketing paradigm of leading with a single message was outdated and your better bet was to create a huge array of messages (sparks) hoping that one would ignite a cascade effect (forest fire). Especially in a digital context, where message production costs are significantly lowered, why not throw lots against a wall and see what sticks (after all, measurement and fast iteration are possible).

With all that said, there is one major issue I have with Watts and Dodds work, which they admit to in the paper: They are examining interpersonal influence, not media influence. While they admit that the distinction is a bit blurry, especially in the eyes of things like blogs, they continue on with the assumption (which doesn’t seem to be grounded in any research) that “the influence of the blogger seems closer to that of a traditional newspaper columnist or professional critic, than that of a trusted confidant, or a even casual acquaintance.” Now I don’t want to harp on bloggers, but I don’t know that I agree with this thesis.

Part of what makes blogs such a fascinating communications medium is the combination weak and strong ties that can constitute a readership. While large readership blogs (like BoingBoing for example) most likely reflect a more journalistic relationship, smaller blogs like this one act much differently. Of the thousand-plus readers who frequent this site I would guess that a significant portion constitute what I would consider a weak tie (we have emailed back and forth) and a smaller portion constitute strong ties (family and close friends). This, I would assume, is significantly different than the average “newspaper columnist or professional critic” who tend to live in another realm. In other words, the availability of bloggers may change how and when their influence functions.

This, of course, is a major critique I have of most communications theory. As my sister, who is getting her undergraduate degree in communications can attest to, I get incredibly upset when interpersonal communications disregards mediated communications. In our current age, the boundaries between interpersonal and mediated communications is hard to pin down. That’s because the same technologies (email, blogging and even text messaging) can be used for both broadcasting and interpersonal communications. Therefore, it’s left up to the recipient to decide whether the communications is interpersonal or not. Prior to that, interpersonal communications was done entirely via one-to-one media (things like face-to-face and phone). While I’m not sure how to resolve this, it does create a major issue in all influential research because it leaves the researcher with an incredible amount of variables to contend with.

Finally, I think a discussion of engagement is probably relevant as I think it’s directly correlated to influence (and when combined with reach may change things slightly). When we launched Street Mining we got two links from largeish sites, one with a very large, but more casual readership and one with a smaller, more dedicated one. While the larger site drove more clicks, the smaller site drove more signups. This, I believe, is the simplest explanation of engagement/influence I have seen: Clearly the smaller site’s readers were a better audience for the message than the larger site. (Of course the lack of control in this experiment means that it’s impossible to say whether it was these factors that led to additional sign-ups.)

This is an interesting paradox that I think relates to this whole influential debate and is often a stumbling block. Influence and reach are two entirely different things. While the two can be related (and historically have been), on the internet they’re not necessarily. For example, we’ve all heard about the “Digg effect” when I site gets to the front page of Digg and is hit with a deluge of traffic. What’s interesting about this traffic is that it often doesn’t result in much additional long-term interest, as the audience is a large and varied one. Therefore, while the site may be considered influential from a pure mass perspective, it’s influence seems much more superficial (I don’t have data to back this up, but have read many discussions on the subject). I would say that while Digg has a large reach and high influence (causing the influx of visitors), the engagement of that audience is low (meaning that they visit the dugg site once and don’t return). (Engagement is a bad word for this, but I’m having trouble thinking of another at the moment. If someone has a better way to describe it, please let me know.)

My argument would be that on many smaller sites the influence is deeper since those relationships tend to be stronger. This creates an interesting dynamic. While I don’t know that it’s statistically relevant, I do think it’s worth exploring some more. Blogs and other associated media do allow people to amplify their voices to more strong and weak ties than ever before, allowing people to have journalistic-sized audiences with relationships that more reflect interpersonal communication.

I think that’s about it. I hope I haven’t bored you to death (I can only imagine if you’ve actually made it to the bottom that I haven’t). Would love to hear your thoughts and feedback (on both Watts and Dodds’ paper and my thoughts). Thanks for reading.

January 28, 2008