I’m here at the #ODR2014 conference. ODR2014 is the annual meeting of the Online Dispute Resolution Forum, an international assembly of lawyers, mediators, technologists, and others who care about technology and dispute resolution.
The livestream is here. What follows is a best-efforts summary of the indicated session.
Next up is Eric Goldman, a Professor of Law at Santa Clara University School of Law and director of the school’s High Tech Law Institute.
Eric begins by defining “reputational information” as “information about an actor’s past performance that helps predict the actor’s future ability to perform and satisfy preferences,” which he describes a very “Bayesian” approach to the concept. Eric was trained as an economist, so he approaches reputation as a problem of the marketplace, where the goal is to ensure that good actors are rewarded (i.e., those who are truthful) and bad actors (i.e. those who misrepresent themselves) are punished. Reputation, Eric says, can be understood as the “invisible hand of the invisible hand,” guiding consumers into participating (and through it constituting) the market.
Eric distinguishes between what he calls “unmediated” and “mediated” reputational systems. In his model, examples of the former include word of mouth and job references, while examples of the latter include credit scores and consumer reviews. He runs through a litany of online customer review sites and what reputational function they provide. Eric notes that there has been a recent expansion in the ‘consumer review’ space to include reviews of other people, e.g. DontDateHimGirl.com.
Eric argues that disputes regarding online reputation are often characterized by asymmetries. The effect of a bad review regarding a potential suitor, for example, exceeds the cost to the person leaving the review; the same obtains for buyers vs sellers on a site like eBay. These asymmetries are intensified, Eric notes, when reviewers are anonymous or pseudonymous and the reviewed are identifiable. The Streisand Effect is another powerful dynamic in the space, since those disputing reviews risk calling greater attention to it.
Eric pinpoints Google as the center of reputational disputes. Whether the review is ‘located’ on Yelp or TripAdvisor or some other site, the greatest effect is felt when it is indexed by Google. Google “casts a shadow” over the reputational space, says Eric, “because that’s where the action is.”
Returning to the economic frame, Eric notes that the most important aspect of online reputation is the reputational trend, and not any one given review. He cites research which suggests that consumers mostly care about the aggregate reputational score of a service provider as opposed to the most recent review (a claim somewhat, though not entirely, in tension with some points made by the eBay team from yesterday).
Eric points to some design practices that certain review companies have implemented in order to help to correct for some information asymmetries while retaining the reputational utility of reviews. They include reviewer authentication/verification, reviewer reputational and community scores, and a prominent right of reply for reviewees. Some try to implement pre-publication or post-publication review by a customer support representative (CSR), but, as Eric notes, this tends to be very costly (in terms of staff time and resources) and often unsatisfying (both because it takes longer and because most CSRs are not well-trained or supported in best practices for dispute resolution).
The biggest path forward for ODR professionals, Eric says, are helping platforms think through these kinds of strategic design questions: what remedies might be employed, what incentives are at play, and what kind of transparency might be brought to the process.