Tuesday 7 April 2009

Of Market Analogies and Ultimatum Games: the myth of web utopianism

Web utopianism is the idea that the web is somehow fundamentally or essentially a positive force. It’s not just that the web is more important, socially “impactful” or different than other media. Rather, it’s the claim that the web is, by its nature, for the greater good. Many futurists and web pundits seem to push utopianism as the web’s “brand,” effectively the set of concepts, assumptions, implications and preferences making up its popular conception. But if we buy into this brand, this idea that the web is in some way essentially good, we greatly reduce our responsibilities as designers and users. Once we take the web to be simply good in itself, we no longer really have to consider the potentially bad consequences of our creations or actions, by whatever standard.

Often utopianists frame the web as the ultimate market, where great ideas, unique voices, vital information and compelling products rise to meet the true and internally generated needs and desires of fully autonomous, choice-empowered, creative consumers. From this freedom of access and choice and the ease with which we can create, share and elevate content, they draw positive democratic and communitarian conclusions. As the story goes, the web is about cultural disintermediation on a grand scale: anybody can create the next video craze through YouYube or help build a surprisingly accurate reference work like Wikipedia. Powerful cultural gatekeepers, who for years perpetuated pernicious and self-serving social and informational hierarchies, are suddenly irrelevant. What’s more, nobody will miss them. As it turns out, claim the utopianists, markets are essentially better at arriving at truth, quality and beauty than experts ever could be. And markets coordinate on these desirable ends through the individual, undirected, autonomous choices of consumer-creators. Thus, the utopianists conclude, the web is a technically aided manifestation of Democratic ideals. And this is obviously good.

Perfect Markets Meet Imperfect Web

Unfortunately, this view is wrong in the same way superstitions are wrong: its evaluations and claims are based on shaky assumptions about mechanisms. I want to avoid the cultural politics of the issue and look at some of the utopianists assumptions. First of all, the positive evaluation of the web as a market seems to follow only by analogy to a specific sort of competitive market, what’s called a perfectly competitive market. This is an idealization of the conditions under which the price arrived at by interacting consumers and suppliers will match the “true value” of a commodity. The idealized conditions are pretty stringent and numerous. For example, you must have a large number of suppliers; no barriers to entry; everyone gets the same complete information; no one turns a profit; no one advertises or markets; each supplier’s output is pretty much individually negligible to the ultimate price; and each supplier’s output is intersubstitutable for any other’s. If these conditions are met, then prices will reflect the “true” value of the product.

The utopianists’ idea seems to be that the web is like a perfectly competitive market because it appears to meet somewhat analogous conditions.
  • Anyone can contribute and huge numbers actually do.
  • Consumers can access any information at will.
  • They can find exactly what they want regardless of how niche it might be and can costlessly choose between options.
  • They can state their opinions, create whatever they want or share their knowledge truthfully or “authentically” without the biasing influence of social pressure.
  • There are few recognizable extrinsic incentives to contribute, which avoids the skewing of contributions associated with profits.
  • Finally, any one contribution is as good as any other and doesn’t really affect the final information. In other words, no one opinion can skew the final collective result because there are so many and there’s always someone willing to refute anyone.
Obviously, it’s a strained analogy. Prices and evaluations are "true" in very different and hard to express ways. Settling true evaluations from masses of tastes and opinions isn’t much like setting prices from masses of tastes and desires. Still, something very close to this analogy is a pretty implicit assumption among many web utopianists (and almost explicit in Wikipedia’s “Jimbo” Wales’s public veneration of F.A. Hayek, father of modern equilibrium theory). To many utopianists, the web is simply structurally conducive to settling on true information or elevating true quality in much the same way perfectly competitive markets are conducive to settling on “true” prices.

But the web just doesn’t work this way. Perfectly competitive conditions don’t hold together on the web. There are many reasons the web's not a perfect market, but I’ll just look at one big one: the false assumption that people online aren’t subject to social pressure or influence that might skew the collective result. It’s based on the idea that people’s visible actions aren’t informational signals themselves; that people’s choices follow an ideal of narrowly rational, autonomous reflection. Duncan Watts’s much discussed recent research shows that social influence in cultural markets, e.g. an opinion market like a movie review site, actually leads to radically unpredictable (i.e. quality doesn’t predict success), highly unequal (i.e. huge difference between very famous and slightly famous) distributions for cultural items. In a nutshell, famous things get more famous and this is a contingent, path-dependent process which has very limited correlation with actual quality. In other words, the same items might show very different success or evaluation patterns depending on chance uptake events at earlier stages of the process.

Clearly, this suggests that the conditions for perfect competition aren’t met in markets with significant signaling, like the web. Markets like this won’t necessarily result in the “best” or "truest" rising to the top automatically because the assumption of autonomous decisions just doesn’t hold. As Clay Shirky has noted, the web makes socialization, communication and coordinated activity "ridiculously easy." I'd say it's more than easy, it's pervasive and inescapable. Stretching the economic metaphor a bit, whenever coordination and communication are cheap, information "cartels" form reflexively and not just when people are positively motivated to alter the competitive landscape to their advantage. People pool and align their opinions (and thus productions) whenever communication is possible, skewing the market. They look to others' actions – not just to their own fact based judgments of quality – to help them make decisions. If something starts to take off, this is “social proof” of value, an assumed signal of quality, and others follow suit. Seeing what does well, more of the same is produced. The social web is all about such signaling – sharing your actions and choices – and thus it's rife with social influence. It really can’t be a perfectly competitive market. (This effect is similar to, but distinct from, so called herd behavior and informational cascades, which also make perfect competition difficult.)

Also, the “Democratic ideal” utopianists claim the web promises is the political analog of the perfectly competitive market: the “best” or “truest” result arises when no special interests have undue influence, when tastes, knowledge and needs can aggregate, free from the distorting interests of profit, influence and pooled power. Utopianists point to knowledge aggregators like Wikipedia as prime practical examples of this idea. But Wikipedia is clearly not a perfectly competitive knowledge market or democracy. Rather it’s an oligopoly, or its political analog an oligarchy. A core group of contributors, the editors, has considerably more power and wields considerably more social influence (i.e. their actions are neither negligible nor intersubstitutable) than all other contributors. “Jimbo” Wales’s admiration of Hayek notwithstanding, Wikipedia is not a pure democracy. It’s an oligarchy. And oligarchies do not structurally result in the “truth” the way democracies or perfect markets supposedly do. Wikipedia is successful (to the degree that it is) not because of the nature of wikis or the web, but rather because of the oligarchy’s ability to manage the negative effects of social influence, informational cascades and bad behavior. And this is arguably the result of social influence used for positive ends, not just the structurally positive force of a perfect "knowledge market" or informational democracy (added April 8: the importance of "community" to the success of wikis is made very clearly by Clay Shirky in Here Comes Everybody and by Cass Sunnstein in Infotopia).

Ultimatum Games and Two Types of Prosociality

Well, the utopianists respond, the relative success of Wikipedia’s oligopoly/-garchy suggests that maybe the web isn’t structurally good in the sense that truth, originality or the “best” stuff succeeds solely in virtue of the implementation. Maybe the web just helps people’s native goodness, prosociality and urge for reciprocal interaction to flourish. The web makes it easier for folks to coordinate, create and realize their intrinsic desire to be, not just social, but prosocial. So the necessary goodness of the web is its ability to amplify and enable user's intrinsic prosocial motivations.

However, this weakened, indirect argument – if used for utopianist ends – founders on something like social influence just as the stronger perfect competition argument does. When people are susceptible to social influence – informational cascades, herd behavior, Watts’s social signaling and conformity or esteem effects – bad behavior or lock-in of less than optimal norms often results. After all, there are two distinct ways to understand “prosociality”: as a desire to share, strive and be nice or as a desire to simply do as others expect you to do. Both ways are about deference to your social group. But only the former is automatically positive in the way web utopianists seem to assume. On the latter, if all of your friends are jerks, you think they expect jerkiness of their peers, and you want them to like and esteem you, then you’ll probably be jerky, too. You’re deferring to what you take to be the expectations of those you hang out with.

Against this less than estimable version of “prosociality” the utopianists often claim that research shows we’ve a hard-wired preference for good prosociality, that we're intrinsically motivated to be prosocial in the positive sense. In particular, they often point to results from Ultimatum games, which purport to show we’ve an evolved preference for flat fairness, what the behavioral economist Herbert Gintis calls “strong reciprocity.” Ultimatum games have two players, Proposer and Responder, and a set sum of “money,” say, $4. The Proposer can offer any sum to the Responder. If the Responder accepts, they both get their cuts, but if he rejects, neither get anything. Research indicates that Responders often reject what we would consider unfair offers and Proposers often start with close to a 50-50 split. In other words, Responders are willing to sacrifice potential gain if the Proposer is unfair and many Proposers seem to immediately offer the fair split. Supposedly this shows that people have a simple preference for fairness, which they argue is a hard-wired, positive-prosocial drive.

Obviously, we’re social animals and thus have an evolved knack for some sort of prosociality. But we have to be clear what’s actually exhibited by the Ultimatum game. Relying on a literature review and critique by Cristina Bicchieri and some ideas of Ken Binmore, some Ultimatum game experiments suggest that these results may be more an effect of the perception of the situation or context than a simple preference for fairness. People are conditionally fair depending on their expectation of other’s expectations and often on what they think others will accept given these expectations. It’s not a matter of fairness, per se, as much as it's a matter of what others expect and thus what you can get away with.

For instance, cross-cultural studies [pdf] of Ultimatum games suggest that people’s rejection rate is culturally determined. In some societies, rejection is very rare regardless of the offer. This suggests that people have a preference for meeting situationally relevant expectations rather than a preference for simple fairness. It's the norms in play that matter rather than some absolute or hard-wired standard of fairness, and this influences both what Proposers expect to be acceptable and what Responders expect to be offered. Similarly, Ultimatum games with asymmetric information suggest that Proposers are generally more interested in appearing to meet salient fairness norms – thus decreasing likelihood of rejection – than actually being fair. Consider the case in which the Proposer knows that the chips used in the experiment are worth 3 times more to him than to the Responder and he knows that the Responder doesn’t know this. If we all simply preferred being fair, as opposed to appearing fair to hedge our bets going into interactions, the Proposer should most often offer 75% of the money. That’s the fair value split. As it is, Proposers in this scenario offer slightly less than 50% of the chips on average. Obviously, we don’t necessarily just prefer fairness for its own sake, which is what’s assumed by the positive-prosiciality idea. Rather we prefer to follow whatever norms we take to be expected by others – the less than estimable prosociality.

So if it’s the case that we’re not so much driven by simple, positive prosociality as we are by the desire to do as our peers expect of us – as the situation-relevant norms suggest we should act – then the prospects for this weakened form of web utopianism aren’t all that great either. That is, if the utopianists argument is that the web is essentially good because it allows the intrinsically positive-prosocial motivations of agents to flourish unhindered by the overhead of real world socialization, then it’s founded on a mistake. Our prosociality isn’t as normatively rosy as they assume. It’s a desire to do as we think others expect of us, which is not necessarily good in the way utopianists need it to be for their argument to work.

No ‘Topias

In this post, though we avoided the political versions of web utopianism, we’ve discussed two of the most interesting non-political strains. They clearly don’t hold water. The web isn’t the wholly positive boon to humanity the utopianists want it to be. But neither is it the great destroyer of culture the panicky, proudly paternalist web dystopianists take it to be (aside: avoid Lee Siegel’s Against the Machine. It’s one of the few books I’ve read that actually deserves to be called a rant or a screed. It’s like the web insulted his mother or something). The web is just a massively influential tool – or fact – with impossible to predict social and cultural impact. Some of the impact will be positive, some negative and some both on different time scales. We just don’t know. But utopianism of the sorts considered above assume that the web just is positive. Taken seriously, this assumption mitigates our responsibility as designers, creators, sponsors and consumers of content and experiences on the web. In reality, we don’t know what the real social and cultural results of our actions will be. But we need to act as if they could be negative so that we feel compelled to strive for – not just expect – the positive in the long term. Doing anything else is irresponsible no matter how nice or progressive it would be to believe the web a simply positive force.

No comments: