Diffusecasting: The New Model for Mass Influence

The BrainYard - Where collaborative minds congregate.

Venkatesh Rao, Contributor

March 25, 2009

19 Min Read

Drip some ink on a piece of fabric and watch what happens. Depending on the type of fabric, the blot spreads at different speeds along the warp and woof. The pattern that appears reveals as much about the fabric as it does about the ink. What does this have to do with social media? Here is a picture of a chain email diffusing through the social fabric, created by Cornell researcher Jon Kleinberg (picture taken from a Cornell University news article).kleinbergcloseup


As I write, a Presidential news conference is going on, a broadcast event that I, like many of you, would have treated as 'unmissable' 10 years ago. Yet, today, I am happy to keep twhirl in my peripheral vision, trusting that if anything truly important is said, tweets or emails will come my way. I have let a vast, trusted crowdsourced filter descend over my eyes. My changed behavior is just one symptom of the waning of broadcasting and the waxing of diffusecasting (I hereby claim credit for the term) as the central process in mass communications. Virality and word-of-mouth are just surface characteristics. Here is a deeper X-Ray view. Mass persuaders, read this if you value your future in your profession.Broadcasting vs. Diffusecasting in HistoryLet's get one thing out of the way: word-of-mouth, viral memes and 1:1 influence are not unique to our age. Neither is it merely a matter of increased speed and degree. By way of proof, consider two pieces of evidence.Exhibit A: Viral diffusion is not new. Innovations, historically, have mostly spread by diffusion rather than by broadcast. There are hundreds of examples of innovations diffusing through an economy as fast as communication allowed. Students of innovation will be familiar with the Plank road, a flawed innovation that spread like wildfire in the 1840s. Back then, communication was expensive enough that only high-value news could go viral (like the idea of the Plank road, or news of gold in California), but the dynamics were the same.Exhibit B: Word-of-mouth is also not new. Consider this excerpt from William Whyte's Organization Man (fast becoming my management bible, easily displacing Drucker's writings), about how the close-knit suburban word-of-mouth networks were eroding marketers' influence:

In this process, merchanidsers have been comparatively passive. In the check I made of air-conditioner ownership in Philadelphia, I found only two cases where the original suggestion to buy had been made by a salesman; in almost all cases the initiative in the purchase had been taken by the consumer himself. To a surprising degree, retailers -- and most manufacturers, for that matter -- fail to appreciate the power of these word-of-mouth networks. Few tend to use "outside" salesmen to speed up the process, and those who do tend to have the salesmen scatter their calls over a wide area rather than work on natural neighborhood groupings... [a footnote continues] Many manufacturers talk about the cut-price problem as if it were due largely to the machinations of discount houses...What has been happening is that the consumer has been taking over part of the selling burden historically alloted to the retailer...The real selling job, in short, is done before the customer comes into the store. Guided by the group, the customer already has determined almost everything about the purchase -- including the fact that he will make it-- except the price and a few minor options; he is earning the price cut, and whether manufacturers like it or not, he is going to get it.

This surprisingly modern analysis, which could apply to IKEA, remember, was written in the 195os, when TV was in its infancy. Howard Armstrong, the iconoclastic inventor of FM radio, had just committed suicide (1954) after losing his legal battles with RCA. Brownie Wise was just inventing the Tupperware party. Whyte describes how the minority of smart WOM marketers operated back then:

With mingled admiration and horror, I heard a door-to-door selling expert explain how a smart merchandiser could exploit the group contagion. First, he said, he'd make a special effort, even if it involved a slight loss, to place air conditioners in the key homes in the neighborhood. Then, after the rest had succummed, he'd leave them alone for a while. Just about the time they would be feeling guilty that there wasn't a conditioner in the childern's room, he would return to trigger the next round of purchases. "All you do now," he said, "is pull the trap. Here's the way I would do it. I'd go in to 'check' the first conditioner, and while I was about it I'd mention to the couple that my wife and I had just bought a second conditioner for the little ones. I'd pause and let that hang in the air for a while. Then, very quitely, I'd say, 'You know, Mr. and Mrs. Jones, now my wife and I really sleep nights.'"

So much for Malcolm Gladwell and his apparently new theory of the tipping point. Whyte was chronicling the same phenomena 50 years ago, as canny salesmen figured out how to leverage WOM and predictable groupthink to drive adoption with domino models and surgically precise nudges.So what stopped the movement in its tracks? Why did WOM and virality need rediscovery in the 2000s? What happened was the few-decades long age of broadcast, a form of cheap mass influence that could be used with so little marketing creativity that it derived its name from the inefficient model of pre-Industrial age agriculture. To "broadcast" is to sow seeds by randomly scattering them in a field. Jethro Tull (1674-1741) perfected the seed-drill, a far more efficient way of sowing seeds (tellingly, at the time, it aroused fears of job losses among farm workers). Mass influencers were not unaware of this waste. John Wannamaker famously noted, well before the Golden Age of broadcast, "Half the money I spend on advertising is wasted; the trouble is I don't know which half."The reason they could be this wasteful of that most precious commodity, human attention, is a brittle construct nearing the end of its design life, known as the 'market segment.' Broadcasting relied, for its effectiveness, not just on the technologies of delivery, but on the social structure that allowed the 'market segment' to exist as a meaningful idea.The Rise and Fall of the Segment, and the Legibility of MarketsWhat happened in the Golden Age of broadcast between the 60s and the 90s (a term I'll loosely use to include mass circulation print, not just TV and Radio), was that three social forces temporarily domesticated WOM and viral dynamics, and conspired to make marketing easier than it naturally is. The Tupperware party, the ancestor of modern 2.0 marketing, was reduced to a marketing curiosity, meant for a fringe type of product, by the Super Bowl commercial.

  • The first force was what I call the direct-line-of-sight. Broadcast allowed mass persuaders to directly get a message to pretty much everyone in the middle class American (and to a lesser extent, global) population. This was because people reliably tuned in to only a few channels that served all attention. Exactly what I am not doing right now for Obama's press conference. Nobody was solely at the mercy of his neighbor's second-hand opinion about anything. Society was all surface area, zero volume.

  • The second force was what I call the homogeneity effect. When Whyte observed suburbia, socio-geographic mobility was just beginning to take off. It was just strong enough that people could move into neighborhoods of superficially similar social class and cultural inclinations, and succumb to the homogenizing Organization Man forces. But it was not enough to create today's culture of global digital nomadism and neo-urbanism (I know nobody in my apartment building; most of my social network is probably an average distance of 1500 miles from me). This meant that highly cohesive, easily visible, and geographically localized groups were forming. Differences were being smothered by the norms of inconpicuous consumption (what Whyte calls "keeping down with the Joneses"). The intricate societies of the teeming pre-suburbia cities were sorting themselves out through sprawl.

  • The third force was local, subsuming peer-influence. Word-of-mouth effects were largely contained within the geographically localized and relatively homogeneous flocks, and peer influence only came from one subsuming peer group that entirely enveloped the individual.

These forces fed on each other. For example TV and Hollywood creating one-size-fits-all grand narratives that further homogeneized communities, making them even more susceptible to broadcast. This virtuous cycle created the stable echo chamber known as the market segment. The idea of an identifiable, localizable, homogeneous group of consumers, who enforced and policed the homogeneity so strongly that WOM/viral effects became predictable and contained. Not only did everybody see the same commercial on TV, but they talked about it within a predictable group of people like themselves and created en bloc responses. The result was that the only possible effects were wholsesale amplification through echoes, followed by adoption, or wholesale rejection. 1 or 0. An entire complex culture of buying reduced to a geographic map that could be broken down into segments marked 1 or 0. For a few precious decades, the fog of human complexity lifted, and human buying behaviors organized themselves in ways that were, in the words of James Scott (Seeing Like a State) "legible" to the coarse eye of the large organization. The market became legible as a very coarsely pixellated black-and-white geographic map. Something even the primitive visual apparatus of the organizational brain could parse.Think about it. This actually makes no sense at an absolute level. What exactly is a market segment independent of the product that induces a segmentation? Why should constructs like "suburban Midwestern white female" mean anything in relation to a car model?Demographic and socio-psychographic profiling have been serendipitously effective tools for several decades. Tools that have enabled marketers and other persuaders to not think about the problem of mass persuasion in all its richness. This effectiveness has been a function of these 3 factors, all of which are in full-reversal today. The direct line of sight is vanishing; society is trading surface area for volume. The largest mass audience is much smaller than it used to be, which means messages that do get through to everybody do so indirectly, soaking in from the listening 'surface'. Second, socio-geographic mobility has accelerated to the point that the homogeneous echo chambers are gone. You cannot predict individual reactions to messages by locating the individual on a map and in a neighborhood. Making the map more fine-grained isn't going to help. Worse, it isn't just that these homogeneous "neighborhoods" have gone virtual. They just don't exist, period. This is because each of us now belongs to dozens of overlapping virtual networks rather than one all-subsuming geographic or virtual community. The fact that I share a specific group context with somebody on Facebook, prima facie, tells you precisely nothing about how I will react to a message, because that affiliation accounts for just 1% of my identity, not 100%. Further analysis may yield some intelligence, but you can't hope to just use that data bit in simple ways.Goodbye Segment, Hello InkblotTaken to its logical extreme, the logic of 'segmentation' gets us to the somewhat ridiculous idea of mass personalization/customization, the idea that the market can be profiled, sliced, diced all the way down to market segments of size 1. The idea of mass personalization is actually applicable only to a small group of power users of a small group of products and services. The vast majority of economic activity will remain aggregated in larger chunks. The problem isn't that the chunks of the market are getting tinier. It is that they are becoming less visible and are bleeding into each other.Think about it. Is everything you own customized? How about 90%? 10%? Yup, that's right. Very little is actually customized. I buy commodity rice and ball-point pens, branded, but mass-produced cameras and cellphones, and so forth. What has changed is that it has become very very hard to predict what total basket of goods and services any of us will consume. In terms of total buying profile, we have all become very unique.Some marketers are getting the idea. The authors of "Trading Up" get that it isn't about mass customization/personalization everywhere. It is about selecting where to go personal/customized. Our 'trade-up' choices, instead of bonding us to a single cultish group, pull us centrifugally towards many, and our loyalty to each is very limited and contextual.So what replaces the idea of the relatively product-independent market segment? It is the highly product-dependent Inkblot. Conceptually, this is the social grouping created by just letting WOM operate naturally and watch where your product or service actually goes, and to what use it is actually put.Practically, however, you cannot operate this way. You cannot find your product/service inkblot by a priori analysis or by the risky strategy of letting your product/service "find its market" after launch (witness the Segway). But you do have to discover and size your market before investing too much in developing it. This then, is the real reason viral, social media marketing is valuable. They are not tools of cheaper-than-Superbowl-ad promotion and persuasion. They are tools of Inkblot discovery, like radioactive tracers. Which brings me to diffusecasting, the process of discovering and mass-influencing illegible, foggy markets that cannot be found on a map (geographic or virtual), predicted through psychographic profiles or inferred reliably from the intent of product/service design. At least, not for much longer.The Seven Laws of DiffusecastingA lot of this section are my analyses of empirical observations conveyed to me by Jeremy Epstein. I am totally an armchair guy myself. So a major hat-tip due to to @jer979

  1. Virality is neither a cause nor an effect, it is a market-tracer: For a year now, I have been listening to superior-sounding 2.0 marketers preaching that virality is not a cause, but an effect. True and false. True, it is not strictly a cause (duh!). You cannot make a 'viral video.' But you can follow some design principles and create something with a high probability of going viral. Infectiousness can be engineered with predictable probability, but contagion for the sake of contagion is pointless. Only college kids with too much time on their hands can afford to waste it making random things go viral (not very hard, not very useful). Virality is only useful if subsequent analysis shows that the virality correlates to interest in an attribute of the product or service riding the virus. The reason people don't get this is that they confuse diffusecasting virality with broadcasting memorability. In broadcast, recall was enough for influence (since all you needed was to get suburban housewives talking over coffee). But a YouTube video about burgers that goes viral because it shows Paris Hilton in a swimsuit washing a car tells you very little about the market for the burger, and a whole lot you already knew about the market for Paris Hilton videos. Yes, the by-now well understood attributes of virality (a good introduction is in Made to Stick) are necessary, but they are not sufficient. The only way to truly leverage viral marketing models is to make a large and diverse set of messages go viral, and hope that post-hoc analysis of at least one of them will reveal the contours of the hidden market you are trying to size, access and sell to. Once you've had that one 'useful' viral, you go in, do the detective work around the path of that tracer, and define your market. Which means by the way, that you have to instrument your viral experiments to death, and get all the analytics you can. Otherwise you are just playing.

  2. Drip Irrigation, not Waterfall: Release early and often ain't just for open source. Thanks to Jeremy Epstein, I found the right agricultural metaphor for social media: drip irrigation. Why do you want to have an extended series of small messages going out rather than one big one? Same reason you want agile programming and public betas for product development, instead of waterfall (curious that the same 'waterfall' metaphor works for the antithesis of 2.0 in both programming and marketing). You want to maximize the number of diffusion experiments and learn from feedback to refine future diffusion experiments to reveal structure even more. You do not want to merely maximize impressions. Having the most popular Paris Hilton video on your marketing resume is not much good if you were trying to sell industrial cement mixers or get McCain elected.

  3. Be the fuel, not the fire: Again thanks to Jeremy Epstein, who got it from I-don't-know-who. While release-early-and-often plays to the role of virality as primarily a tracer, it is still pretty darn expensive. You have to be creative and create a diverse set of viral experiments and hope that one of them finds your market (aside: too many viral marketers just look for the first hit, and then use a 'sequels' strategy. Dumb unless your first viral hit is also the right viral hit.) What better way to do that than to crowdsource your tracer experiments? Scan for proto-viral messages that might have the same tracer effects that you want, and just pour fuel on them. Find your Jared. You need to build an early warning system of course. Joe the Plumber isn't as easy to find as Michael Jordan.

  4. If you can't find the echo chamber, create one: This I believe is Jeremy's own strategy, inspired by Seth Godin. For Dan Pink's book, which he was marketing (The Adventures of Johnny Bunko), Jeremy had the insight that they needed to get readers of the book talking to each other rather than to Dan. So he came up with the idea of catalyzing "Bunko Breakfasts." I have no idea how successful that was, but this is the right direction of thinking. If you can't see your market, and seeing isn't much use anyway because of the non-subsumption effect, why not just provide the constructs that help structure it? 60s and 70s markters found readymade communities in the suburbs, you just need to create the crucibles of peer-reinforcement yourself. And no, this does NOT mean creating a community around your product or service. It means creating a community around whatever set of values might serve as glue (as discovered in viral tracer experiments) for people who like your products. As P&G (I think) discovered by accident, the community glue for young girls who might buy tampons isn't "tampon," it is "girl talk." Again, what came readymade in the Organization Man suburbs needs to be discovered and foregrounded to form communities that serve your messaging needs. And remember, you are dealing with individualists here, not tribes, so your glue will be temporary and for a limited context. Which brings me to...

  5. Say it with me once more, NO TRIBES: I am now getting truly tired of the 'raving fans' and 'tribe' metaphors. Besides Kevin Kelly and Seth Godin, we now have Tribal Leadership by Dave Logan. There's some good thinking in all this writing, and empirically valid principles, but I strongly believe the analysis in terms of the construct of 'tribe' is wrong. Not only is the implied collectivist culture being displaced by individualist ones, but the "many overlapping communities and no 100% subsumption" principle makes 'tribe' the wrong idea. The defensible definition of tribe is "[closed and all subsuming] community where everybody knows each other," roughly what the primate ethologist Desmond Morris used as his definition in The Naked Ape and The Human Zoo. Primate ethology also gives us the limits of 'tribal' organization in the idea of the Dunbar Number and the Monkeysphere. 'Tribe' is an extraordinarily misleading term. What look like tribal dynamics are situational, temporary, 'aspect' dynamics. The Organization Man suburb is the tribe. The groups we are seeing today are more like the temporary communities of airport waiting areas. Your job is to create the equivalent of flight delays: a reason for people to start talking. Hopefully about you. Even the God of Media, Marshall McLuhan, fell victim to the 'tribe' idea when he talked about 'retribalization' (again, good ideas, wrong metaphor).

  6. The Long Tail is more important than you think, but not for the reason you think: Most discussions of Chris Anderson's "Long Tail" concepts have focused on the aggregate volumes of many small niches. But remember, the very idea of segment (of which 'niche' is a variety) is suspect now. Framed this way, the "Long Tail" seems important only for the mass-personalized/customized part of the market. WRONG! It is vastly more important than Anderson says, because the Long Tail actually is the market research arm of every product and service, mass-customized/personalized or not. Here is why: we've talked about various degrees of laziness in finding your once-easily-legible markets. Viral methods put tracers through it. Drip irrigation spreads your risks and increases the probability of discovering your market boundaries. Fuel-not-fire helps you crowdsource the creativity required for diverse drip irrigation. Why not outsource the whole darn thing? With so many little, temporary, non-subsuming cults forming around all sorts of things, and forming natural boundaries through customization choices, maybe one of them is the one for you? Maybe everybody who voted Jericho back is the market for a new Sci-fi book? Maybe everyone who buys non-white iPod headphones is your customer ('design conscious, but under-the-radar'). I'd pay as much for a service that tracks a select group of long-tail product launches as I would for a good segmentation database.

  7. Down with 'Authenticity', Up with 'Ironic Authenticity': Coming in a close second to 'tribe' in terms of 'concepts that make me want to hurl' is 'authenticity.' The idea of genuine, high-trust, no-mind-games discourse as the norm, and an age of 'genuine' conversation as mass persuasion is downright silly. I mean, ask yourself, do you really want your soap sales pitch to reach the same level of rawness you had when you bared your soul, after Drink #6, to your college buddy (which you were so embarrassed about after, you never spoke to him again?). When it comes to marketing, we want, not 'authenticity' but a return to open, fish-market style joint-narrative-construction bargaining. There, we merely have to go from the bad acting of broadcasting to the method acting of diffusecasting. Authenticity does have a role, with the adjective "ironic," in a truly solid notion of that overused idea, the personal brand. But that's a subject for another polemic, another day.

As always, I am just trying to stir the pot here to invite debate. This isn't a complete or rigorous analysis by any means. Heck, I am not even a marketer. But this is just the beginning. A great deal more needs to discovered about diffusecasting. I am personally fascinated by the idea of a 'diffusion editorial' -- the discourse that emerges around a single viral probe into a market. I might blog about that sometime.Venkatesh G. Rao writes a blog on business and innovation at www.ribbonfarm.com, and is a Web technology researcher at Xerox. The views expressed in this blog are his personal ones and do not represent the views of his employer.

About the Author(s)

Never Miss a Beat: Get a snapshot of the issues affecting the IT industry straight to your inbox.

You May Also Like

More Insights