Crowdsourcing and The Wisdom of the Crowds

When my review copy of Jeff Howe’s Crowdsourcing arrived in the mail, I figured I’d use the opportunity to finally finish my half-read copy of James Surowiecki’s The Wisdom of Crowds (2004) and do a two-in-one review. It took a good deal longer to write than I expected, mainly because I kept getting distracted by other connections I wanted to explore, which threatened to turn this post into an extended and discursive riff on all sorts of other subjects. I finally firmly trimmed off the wayward thoughts and put them into other post drafts, so luckily for you, this is a relatively focused two-for-one book review.

Crowdsourcing vs. The Wisdom of Crowds

On the surface, the two books have a lot in common, not just with each other, but with other books like Wikinomics, Groundswell and Open Innovation. The last three though, tackle the notion of crowd indirectly, being primarily concerned with the antithesis of the crowd, the corporation. The usual suspects (think Innocentive, Ronald Coase, Dell’s Ideastorm) appear in many of these books, as do concepts like prediction markets and user-generated content. Apparently, we have a long way to go, before we hit the Southwest-Airlines level of diminishing returns from this set of exhibits.

Here’s how you keep all these books apart: Wikinomics tells corporations what to do with the crowd, Groundswell tells them how, Open Innovation tells them why.

About turn; march into the crowd itself (either personally, or with your corporate mask on). Put on your New Yorker academic-voyeur hat and go ecstatic about individual and collective biases and behavioral-economics irrationalities, as revealed through too-clever experiments, and you get The Wisdom of Crowds. Cross over those tracks, simultaneously stepping from gown to town and from monolithic, Main-Street culture to the alarmingly fractalized world of Internet-enabled subcultures. Put on your urban ethonographer hat. You get Howe’s Wired model of subcultural techno-voyeurism. Let’s look at both in order.

The Wisdom of Crowds

I have to admit, the New Yorker style of self-assured, faux-authoritative rhetoric annoys me in general, and particularly in this case. Surowiecki is much too confident about his ability to represent and interpret academic work. I lost my faith in his inferences when I read the bit, somewhere in the middle of the book, about traffic behavior and automated highway systems, a subject I studied at a fairly detailed and technical level during my PhD, which was on the related problem of formation flight. His treatment of this particular theme is somewhere between simple-minded and plain wrong. I am no expert in the wide range of other domains he talks about, but I suspect a good approach to take towards the book is to accept most of his conclusions as being only roughly, directionally correct. Compared to his intellectual bete noire at the New Yorker (and defender of experts over crowds), Malcolm Gladwell, Surowiecki’s ideas are fundamentally much sounder. Gladwell, though he has an eye for unusual angles on important themes, just keeps getting things fundamentally wrong in deep ways (both of his bestsellers, The Tipping Point and Blink suffer from that failing). But that’s a whole other story.

My personal pet peeves aside, WoC is overall a good starter book on the theme of crowds. It is broad, and anchored by a reasonably robust conceptual model. Surowiecki’s model has two pieces. One is his implicit definition of a crowd (or, at least, a wise one): a collection of diverse agents whose members individually have access to different sorts of private knowledge about the subject at hand. These agents must be sufficiently non-interacting to enable individuals to form private judgments, but sufficiently coordinated (possibly or even typically, by a few individual benevolent dictators) to enable coherent and useful aggregation of the results. The second piece of his model involves classifying problems loosely into cognition, coordination and cooperation problems. If the crowd interacts too much internally, pathological group dynamics can drive the group towards homogeneity, at which point it loses its collective wisdom and, at best, becomes as smart as the smartest individual. His rough-and-ready classification of crowd-solvable problems also works quite well.

Through this warp and woof of crowd and problem characteristics, WoC weaves a variety of interesting stories, mostly about rather academic behavioral economics experiments, but also about real-world examples (the bit about the Columbia disaster is particularly good). You get everything from classroom experiments involving colored balls in urns, to starling flocks in Africa and hunts for sunken submarines off the coast of Florida. And of course you get the classic problems of crowds guessing the weights of animals or estimating the number of jellybeans in a jar.

Perhaps the most valuable contribution of WoC is its characterization of entities related to the wise crowd, like the dumb mob, the moronic small group and the blinkered choir of experts. Read the book to understand how, where and when those entities appear and mess things up.

Crowdsourcing

Though it tackles many of the same themes and examples, Crowdsourcing is a very different book. The title suggests a process that corporations might follow to “crowdsource,” but the book covers nothing of the sort (Groundswell does that reasonably well). Don’t look for a master plan to set up a crowdsourcing procurement process within your purchasing department.

Howe clearly stumbled upon this theme via his passion for obscure subcultures. This is an urban ethnographer playing management consultant. When I started reading the book, my first, rather uncharitable thought was, this is just a rehash of Wikinomics and The Wisdom of Crowds, with a couple of newer examples thrown in. But somewhere in the middle of the first chapter, I managed to get my mind around the book’s novel perspective. Howe’s disarmingly honest statement of his own biases sets the tone:

Technology itself consists of wires, chips and abstruse operating manuals. Worse yet, for a writer, it’s boring. Far more important and interesting are the human behaviors technology engenders.

Since I’d probably call myself a technologist-writer, and one who routinely pokes fun at the holier-than-thou earnestness of ethnographers at that, I heartily disagree with this sentiment. But this ethnographer perspective is what makes Crowdsourcing an interesting book. Howe is obviously a very different sort of person from me. The subcultural minutia which clearly fascinate Howe would bore me to tears if I tried to observe them myself. So the diversity of this particular two-person reader-writer crowd made for the construction of some interesting meaning.

Crowdsourcing is, effectively, a series of very quick thumb-nail sketches of the subcultures which emerge around different social technologies that use rich Internet applications to allow some mix of user-generated content and online community building, the staples of social media, to emerge. Howe looks over the shoulder of Web-video sitcom directors and iStockphoto photographers. He rides along with obscure bands and goes to concerts that I’d have to be dragged to at gunpoint. He follows the rise to fame of viral videos and t-shirts. Where other authors drill down into the economics and return-on-investment of innovation intermediary companies like Innocentive (where free-agent researchers wins bounties for solving problems posed by corporations), Howe decides to go and track down individual winners and listen to their stories and motivations. Where books like Groundswell present formulas like POST (People-Objectives-Strategy-Technology, an acronym for what Li and Bernoff, the authors, believe is the right decision-making sequence for building online communities), Howe picks out the conversational subtleties of engaging crowds.

One anecdote in particular, struck me strongly as representing the heart of the book. It is about the Cincinnati Inquirer’s attempt to attract user-generated content from the masses. Howe highlights the evolution of an apparently trivial user-experience design decision:

“[The link] used to read, ‘Be a Citizen Journalist,'” Parker [an editor of the newspaper] says. “And no one ever clicked on it. Then we said, “Tell Us Your Story,’ and still nothing. For some reason, ‘Get Published’ were the magic words.”

The reason this is not a trivial anecdote, and something worth highlighting, is that online communities — crowd containers so to speak — seem to be scarily sensitive to the quality of such tiny decisions, especially early in their lives. I could write a whole 1500 word speculative analysis of this particular anecdote (who is “Us?” in the second phrase and why does it matter? How is the state of being published a different incentive than the state of being a (clearly-second-class) ‘citizen journalist’?)

I learned the importance of this sort of stuff first-hand through my own handful of experiences (both successful and unsuccessful). My own most memorable experience of the gossamer-like delicacy of successful community building occurred when I went from being a rather vocal community opinion leader at Sulekha.com to being an employee. That was rather tricky, navigating the accusations of “selling out” to “management” and re-establishing my personal social capital. Prosumer crowds are rather like unions that way.

Besides the newspaper example, another thumbnail sketch stuck a poignant note in my mind, this time for personal reasons. Howe provides a detailed account of the MATLAB programming contests. While I haven’t participated in these, I have spent much of the last 10 years of my life working with MATLAB in very similar situations (oddly enough, coding up models and simulations of teams and crowds), to the point where I am effectively ruined for programming in any other environment. Most of the dynamics of the contest that Howe observes (surprisingly acutely, since I think many non-programmers would miss much of this) are present in situations that are not explicitly crowd-like. So maybe the lessons here are applicable well beyond modern digital crowds.

So if you are in a mood for Chicken Soup for the Crowdsourcer’s Soul, as Chronicled by Howe the Digital Ethnographer, go ahead, read the book. It might be particularly valuable if you are actually involved, hands-on in any effort to manage a crowd-like community and happen to have, like me, a yeah yeah, it’s all a rich tapestry attitude towards ethnography.