SmoothSpan Blog

For Executives, Entrepreneurs, and other Digerati who need to know about SaaS and Web 2.0.

Tipping Points, Personality Styles, and Email: A-List or Spam?

Posted by Bob Warfield on January 29, 2008

Guy Kawasaki recently wrote that you could “Forget the A-List After All.”  The post was based on a Fast Company article on Duncan Watts, a Yahoo researcher and academic who believes Malcolm Gladwell’s celebrated “Tipping Point” concept is wrong.  Both articles are a fascinating read, and really rocked my world thinking about them.  I was surprised not to find more chatter in the blogosphere about it, because if Watts is right, it has huge impacts for marketing and social networking.  It’s on Techmeme, so I guess people are still pondering what it means. 

The short form says that Watts has concluded that key influencers (what Kawasaki refers to as the “A-List”) actually have very little impact on how fast ideas spread at all.  Rather, Watts feels its more a matter of whether the world is ready to accept whatever the message may be.  If it is, then anyone can spread that message.

Pause for a moment and let that sink in:

There are no key influencers.  Nobody has undue influence over others.  Anyone can spread the word if the word is ready to be spread.

While we’re at it, there are a whole host of corrollaries:

– Facebook shouldn’t bother with Beacon, what your friends do doesn’t matter to  you.  Just as long as you are able to hear from anyone about something, that’s good enough.

– Markets shouldn’t bother trying to reach the A-List of influencers.  It’s more important to make a lot of impressions than to make the right impressions.  Even more important is to tune your message, because if all impressions are equal, that’s all you have to work with. 

– Spam works if the world is ready to receive your message.

Do you agree or disagree?

Mathew Ingram, for example, seemed almost to say there are influencers, you just never know which one will turn out to be the “patient zero” for your meme.  I don’t think that’s what Watts is saying.  In fact, he specifically feels there were no individual influencers who were markedly more effective in spreading the memes.  What he says is not that anyone could be an influencer, but that he just doesn’t believe there are influencers.  As Watts put it to Fast Company, “A rare bunch of cool people just don’t have that power. And when you test the way marketers say the world works, it falls apart. There’s no there there.”

As for Watts, I find I seriously disagree, but after thinking about it for a while, I can see ways in which Watts may be partially right.  Or to put it another way, I can see how to value what Watts is seeing, but think he hasn’t seen the whole picture.  Watts, for his own part, admits he may be wrong, but he’s quick to add that he thinks he’s the only one who clearly says what he’s talking about and has real data and models to back it up.  Basically, he has experimented with how fast word of mouth spreads via email, and also with computer simulations where he can change various parameters and watch their impact on the simulation. 

Let’s get that data out on the table right away, at least as much as we can understand based on the Fast Company article.  The email test involved signing up a very large group to see how fast they could propogate a message to a desired audience they didn’t know directly.  Watts reaffirmed the old “six degrees of separation” rule of thumb when he mapped out the paths the mail took, but he also discovered little evidence of the key influencers.  Yes, there were some, but quite a lot of the messages got through without recourse to influencers.  This is in contrast to Gladwell’s experiment which involved written letters and was heavily impacted by the influencers.  Half of the letters that got through came via just a few of these influencers.  It is interesting to keep in mind that the other half didn’t need the influencers.

The second thing Watts did was to create a computer simulation of a network of communicators.  The simulated communicators were given various characteristics.  For example, simulated Influencers were about 10% of his simulated population, but they had 4x the influence of mere mortals.  The sim started when one of the population got an idea and tried to pass it along to others.  Watts found that more often than not, even when he upped the Influencers to 10x the influence, the viral wave was due to non-influencers.

What does these experiments mean?  I think there are several interesting and intertwined phenomena potentially at work here. 

  • Message Context Matters for Trust
  • Message Medium Matters for Acceptance
  • The Online World Changes How Tipping Points Work

Message Context Matters for Trust

What message are we spreading?  Probably the easiest message to spread is humor or something interesting.  A great LOL Catz picture can come from anywhere.  It doesn’t need Influencers to get out there.  There is very little cost or risk associated with accepting such a message.  Perhaps the next up the hierarchy would be news.  News may cause us to act, but since we generally don’t associate that action with the person delivering the news, we will still take news from a broader range of sources than perhaps even we should.  Witness cases where bloggers break news or news appears on Twitter.  Another step up is a message that tries to get us to take action that has a cost, but that apparently does not benefit the sender.  Last step, just like the former, the message wants us to take action that has a cost, and taking that action benefits the sender.  We are most suspicious of the latter.  Seth Godin puts it well when he says the old saw about it’s who you know is wrong, it’s really more about who trusts you.

Neither of the experiments attributed to Watts appear to consider this issue at all.  Until experiments like these are carried out with messages that have real costs, such as purchasing a product or service and potentially being unhappy with them, I’m not sure how relevant they are.  Certainly I wouldn’t just take them entirely as gospel.  Watts can talk about how the evidence of others is anecdotal, but it is a near universal anecdote that people have varying levels of trust for the person delivering the message, and nearly all of us know of individuals we rely on for advice.

Message Medium Matters for Acceptance

I’m fascinated by the choice of email in Watts’ experiment versus written letters for Gladwell.  Today, there are many more media formats available to us.  What’s the best one?  Do they behave differently?  Is a random email really a good test for influence, or would we call that Spam?  Do we need richer media to communicate strong influence?  What sorts of messages generate the most influence with which media?  Is it simply the news that a person you respect has made a purchase?  Or do you need to see their purchase and experience their enthusiasm?  Or perhaps you need to see other people getting enthusiastic about your friend’s new purchase? 

This is all extremely subtle stuff, and is again not reflected in Watts’ experiments so far as I can see.  Let me bring up another issue with respect to media.  I’ve written before that different people have different learning styles, and this leads to a sort of “Myers Briggs” effect in terms of how we like to receive communication on the web.  I like to use this chart:

Web 2.0 Personalities

This sort of thing accounts for the love/hate polarization that people have for things like Twitter or Video: its a matter of your preferred learning style.  I have no idea which styles are more popular, it’s even likely that it’s too early to tell–not enough data for many services.  But I’ll bet they have a huge effect on all this.  In fact, one of the things Influencers may instinctively do well is to know what style of communication and messaging works best for the person they’re communicating with.

The Online World Changes How Tipping Points Work

I liked Scott Karp’s article on this, because he argues that influence still exists, just that it works differently on the web.  He postulates you can use Social Networking, for example, to make yourself an Influencer.  This is a theme I’ve written about before.  See for example my post, “The Psychology of SaaS and Web 2.0 Persuasion (and Selling).”  In it I argue that the roles envisioned by Gladwell’s Tipping Point (Connectors, Mavens, and Salesmen) all exist on the web, but they aren’t always people:

Interestingly, these three types of people may not necessarily even be people in the Web 2.0 world.  Consider Facebook and their widget API.  If you create a widget for Facebook, you are using Facebook as a Connector (because it has a huge social network), a Maven (because it’s the authority in social networking right now), and a Salesman (because it’s “hip” to be on Facebook).  Don’t miss the opportunity to harness entire herds like this to help create your own herd.

The online world provides so many new ways that tipping points can come into being.  They also provide a lot more interesting ways to help try to get the word out.  For the first time are crude measures of the degree of connectedness various individuals have.  For example, by the number of friends they have on Facebook or the size of their LinkedIn network.  Perhaps the next stage would be to think of ways to measure the efficacy of these relationships.  Lastly, the web enables sheer brute force.  If all else fails, it is cheap to create tons of impressions on the web than by purchasing Super Bowl Ads. 

I still think you’re better off thinking about who to impress.  If it didn’t matter, as Watts seems to say, the “Digg” effect would be more valuable.  If you’ve never heard of it, it’s the phenomena where an article is linked into Digg, Techmeme, Techcrunch, or whatever your favorite attention concentrator du jour may be.  This will drive huge traffic.  The trouble is, the traffic disappears again almost immediately.  OTOH, folks like Andrew Chen have written that when Robert Scoble links to  you, your subscriber list of repeat visitors is permanently enhanced.

It sure sounds like Scoble is a Gladwell-style Influencer and Watts missed the point on that one.

Related Article

Sweet irony:  David Armano writes that he had the Fast Company article in hand and never read it until influenced to do so by Guy Kawasaki.

6 Responses to “Tipping Points, Personality Styles, and Email: A-List or Spam?”

  1. huperniketes said

    Leave it to an engineer to see the flaws in Watts’ reasoning!

    The Tipping Point is about signal propagation, although he might not have realized it. He includes the importance of signal and transmission medium in his model, which most people ignore in reading his book (if they’ve actually read it). And while you’re the only blogger I’ve read on this topic so far who remembers the three types of influencers in Gladwell’s model, I think you’ve still followed them in getting caught up in the one-size-fits-all A-listers are influencers misunderstanding behind Watts’ work.

    But your criticism of Watts’ experiments are spot-on. They lacked any analysis of message (signal) or medium, both of which are incorporated into Gladwell’s model as stickiness and context respectively. How are multiple conflicting messages handled? How are noise and signal degradation treated? An analysis of an end-to-end effect of a single signal passing through a multiplier filter does not a complete marketing plan make.

    I’m glad I followed your comment from Logic+Emotion: your analysis on this and the multi-core problem indicate to me that you’re one who merits attention.

  2. snp1962 said

    Thanks for this, an excellent perspective, for a different take on this you might find Duncan Brown’s new book – an insightful read.

  3. smoothspan said

    Huperniketes, I’m sorely tempted to code up the simulation Watts describes in his paper. It’s not all that hard. It would be entertaining to play with if nothing else. But there are some assumptions in the fabric of how the decisions get made in the simulation that are troubling with respect to all of these issues.

    Snp1962, thanks for the reference. Looks like a great book!



  4. […] Smooth Span […]

  5. […] Tipping Points, Personality Styles, and Email: A-List or Spam? […]

  6. I’ve heard of this study a few times now. Actually, it doesn’t surprise me that the concept of influencer’s is actually wrong. However, the book made light of some very important points. Simply stated; ideas that are sticky cause tipping points.

    His analysis of why, however, was ultimately flawed.

    Building on the concept of stickiness a new book was just published called “Made to Stick”. I have found this to be spot on in explaining why tipping points occur. The crux of the concept is that ideas must be sticky to tip. The key is understanding the core factors of stickiness.

    Andrew J Durstewitz

Leave a Reply

%d bloggers like this: