The leisurely pace of this winter break has finally given me a chance to read "The Wisdom of Crowds," James Surowiecki’s 2004 bestseller. Naturally, I can’t help wondering whether Surowiecki’s theories about crowd wisdom can be applied to my favorite tech topic: wikis.
Wise crowds, in a nutshell
Like many books in this pop-business category, "The Wisdom of Crowds" consists of a bullet-point theory and a slew of supporting but anecdotal examples. As such, it is best summarized through a single example (Surowiecki, pg 6): an experiment by physicist Norman L. Johnson, in which the subjects were sent through a maze, individually, two times. The first time, as expected, they wandered about inefficiently, averaging 34.3 decisions to reach the exit. The second time, as expected, they improved to a mere 12.8 decisions.
The somewhat surprising result, though, is what you find when you combine their decisions in the second run. By solving the maze using the average decision (i.e., at any given decision point, you choose the more popular choice), the exit is reached in a mere 9 steps, which was also the fastest possible route through the maze. The takeaway? By aggregating the diverse, independent, and decentralized information in the minds of individual subjects, a better route was revealed than the individual subjects had discovered.
This is the core of Surowiecki’s argument–that crowds are more wise than individual experts when they have diversity of opinion, independence, decentralization, and a mechanism for aggregation. On that note, what better place than google to start my inquiry on the wisdom of wikis?
Wisdom of Wikis on the Web
The results, I found, were a bit mixed.
Curiously, the first significant hit I saw when I googled "Wisdom of Wikis" was this article from Friday’s issue of the Missoulian, the local paper in my hometown, where I’m writing this post. Unfortunately, it turns out to be the typical soft-news coverage of technology: an overview of Wikipedia and Wikiscanner, accompanied by a subtle misuse of the word "crowdsourcing," with almost zero depth of analysis.
Similarly, CatherineThirty blogs on the topic, but provides little more than a chearleading session for wikis, based on their decentralized but systematic input mechanism.
Ross Mayfield (founder of SocialText) has a fantastic post that challenges the link between Surowiecki’s work and wikis, which is in dialogue with another post from Julian Harris. The crux of their discussion is that, where Suroweicki emphasizes the independant decisions of individuals, a wiki is inherently collaborative, so the Wisdom of Wikis and the Wisdom of Crowds cannot be equated. Mayfield describes this as a distinction between collaborative and collective intelligence, respectively.
Of Surowiecki’s four criteria, three can quite obviously be applied to wikis (or, more accurately, the crowd that uses a particular wiki). A wiki with a broad usership certainly has a diversity of opinion. It also clearly uses a decentralized model. As for aggregation… the wiki itself is the mechanism.
All that remains is the question of independence. "Diversity and independence are important because the best collective decisions are the product of disagreement and contest, not consensus or compromise." Is this notion compatible with the wiki? Recall Stephen Colbert’s supposed definition of wikiality: "truth by consensus."
Overlooking that Colbert only spoke those words in wikiality (not in reality) the points stands: as Mayfield suggests, the prominent role of consensus in wikis is in conflict with the independence that Surowiecki requires, suggesting that Surowiecki’s crowd model can’t be applied to a wiki.
There are at least four possible conclusions that one could draw from this analysis.
- As Mayfield suggests, these two models are simply incompatible, and reconciling them is unnecessary.
- Surowiecki’s model is incomplete, and should be refined so that it accomodates crowds of wiki users.
- Wikis are limited, in some way, by their failure to take advantage of the crowd.
- Our understanding of wikis is limited.
I’d like to explore this final conclusion a bit further. For most people, a wiki and Wikipedia are indistinguishible: Wikipedia is the only wiki they’ve used, and they haven’t conceptualized a wiki beyond that limited model. Some characteristics of that model:
- Heavy editing of individual pages by many users.
- Consensus and congruence both within single pages and among multiple pages.
- Consensus minimizes the role of opinion.
However, we also know that wikis are being used in a variety of other contexts, including academic and corporate. Some wikis are open, like Wikipedia, others are closed (and those that are open are very susceptible to corporate whitewashing or other forms of PR manipulation, whereas those that are closed are not).
The model of wiki use plays a huge role in how it relates to decision making. In a corporate setting, for instance, a wiki might be used to store meeting notes and link them to different meeting notes, to product specifications, to administrative documents, to how-to guides, or to any number of other types of information.
Much of the focus on consensus in wikis is based on Wikipedia and, as the three characteristics above suggest, the emphasis in Wikipedia is on emulating a traditional encyclopedia, both in terms of accuracy and organizational structure. Linking is merely a bonus. In other wiki settings, however, the linking and aggregation of information takes precedence over consensus.
Here, Surowiecki’s example of Johnson’s maze experiment can be helpful, as mapping a maze and mapping a network of information share a great deal in common. In Johnson’s experiment, each subject creates his own map of the maze, and the ideal map emerges from their individual results.
On Wikipedia, obviously, this is not the process at work. Instead, the entire community navigates the maze together, debating and negotiating at each turn. On another wiki, however, it might be the process at work, as individuals create their own pages for similar topics, network information in individual ways, and from these individual processes better solutions may emerge.
In this sense, a wiki is really no different than the Internet as a whole, which Surowiecki considers (thanks to Google’s PageRank system), to be an extremely wise crowd.
I’ve meandered a bit, I know, and ended that last section in a way that is far from conclusive. The point I ended on (that a wiki isn’t much different than the rest of the Internet) is a big one, over-simplified as it may be. So is my general assertion that wikis are grossly misunderstood, because of the prevelance of one atypical wiki: Wikipedia.
Perhaps this is a bit pointless, trying to fuse together the wisdom of crowds with the wisdom of wikis — clearly, there is evidence that suggests tremendous strengths and possibilities for both and, equally clearly, fusing them is rather tricky. Is there any value to a grand unified theory of collective/collaborative wisdom? Exercise
Regardless, I think the is helpful, at the very least, because of its potential to initiate discussion about collective vs collaborative wisdom, their respective advantages and disadvantages, and their relationship to contemporary media technologies. Exercise