• If you think you received an unfair feedback rating, you can start an appeal by posting a statement of up to 100 words and uploading up to three photos as evidence.
• The person who left the negative feedback has 14 days to justify the comment and post a counter-statement and photos, respectively.
• Once the response is in, you have another two days for commenting on the response.
• The case is then randomly allocated to 100 ebay jury members to be voted upon: if at least 70 think the feedback should be removed, it will. Otherwise, it remains in your profile and becomes part of your reputation.

The idea of a community court is not exactly new. Having a group of citizens as “finders of the fact” is at the heart of the jury trial as especially Anglo-Saxon lawyers (and courtroom thrillers) will tell you. What is new, however, is the scaling of the jury in a digitally networked environment. With no deliberation or even knowledge of each other’s identities, 100 experienced ebay members are randomly selected from a pool of volunteers and decide individually by mouse click on the fairness of a given piece of feedback. The result is a large-scale collaborative group judgment based on the evidence provided by the parties. Justice, one could say, is being “peer-produced” from a commons of normativity – thanks to the low-cost, bi-directional architecture of the Internet.

Will it work in practice?

It seems that the success of the ebay community courts will hinge on whether the designers manage to strike a balance between efficiency and (perceived) fairness. To judge from the outside, efficiency seems to be taken good care of: an automated procedure with clearly defined time frames and responsibilities for parties and jurors, immediate enforcement of the judgment through technology, and no options for further appeal (at least within the ebay community).

As far as fairness is concerned, time will show whether the courts meet the normative expectations of the community. Some speculation: active ebay members seem to be well suited to judge the fairness of a comment as they have first-hand experience of what is considered “good” and “bad” in their community (which may well differ from what is considered “good” or “bad” in other contexts). They also know about the consequences of negative ratings for a seller’s reputation. Apart from this “practical knowledge,” ebay’s designers seem to have tried to tackle the problem of bias. Not only are cases assigned randomly to members, who volunteered for jury service. The rather large number of “jurors” also suggests that potential outliers and extreme views are likely to be buried in the mainstream of the crowd. By design, individual judgments are made independently from one another, which may help the community courts avoid the well-known pitfalls of group decision-making like informational cascades, herding, or groupthink. This looks very much like the Condorcet Jury Theorem in action. Still, the procedure remains a very crude one and is likely to be criticized.

It will be interesting to see how these large-scale collaborative group judgments fare in the community – and what we may learn from this new kind of digital institution for other contexts.

The Patent and Trademark Office is starting a pilot project that will not only post patent applications on the Web and invite comments but also use a community rating system designed to push the most respected comments to the top of the file, for serious consideration by the agency’s examiners. A first for the federal government, the system resembles the one used by Wikipedia, the popular user-created online encyclopedia.

As far as I know, this is one of the first cases in which a government makes targeted use of the Internet’s new collaborative technologies. It seems to be a promising approach. For a long time, scholars and policy makers have complained about the increasing knowledge gap on part of government agencies on a rather abstract level. Now there is an active attempt at leveraging the (in many people’s opinion already rather buzzwordy) “wisdom of crowds” for better regulation.

It will be interesting to see how the decentralized review fares in practice: Will a critical mass of people participate? How susceptible is the process to gaming? And in a government context most importantly, will the system meet the normative benchmarks of the legal system, such as due process and accountability?

Beth Noveck, one of the masterminds behind the plan, addressed some of these issues in a recent paper for the U.S. context.

I wonder what other applications of recommendation systems in public policy one can imagine, especially in the European context. One idea would be to think about the problem of how to classify content that is potentially harmful to minors, a big issue in most Western European countries. A collaborative filter that generates community-specific ratings based on the reputation of raters may be just one idea to think about. It would bypass the problem of rather obscure procedures in overworked and understaffed content rating commissions and add transparency as well as – arguably – legitimacy to the process.

What is interesting from the perspective of regulation is that there seem to be quite different mechanisms for coordinating traffic in different parts of the world. Two examples from my own experience: In my home country Germany, the default still seems to be coordination by traffic lights. The state sets rules as to who is allowed to cross the intersection at what time in what direction. Red light, stop. Green light, go. This is regulation by law, and most people abide by it – even if they are the only ones waiting at a deserted intersection in the middle of the night.

In southern Italy, my experience has been different. Traffic lights exist, but people do not seem to pay too much attention to them. Rather, they tend to slowly roll into the intersection and wait for reactions of other drivers. If the other driver slows down, this is taken as permission to cross. If not, people wait. If one driver notes that another driver cannot see her approaching, she honks. This is not regulation by law, but regulation by eye contact and honking.

Does one work better than the other?

I am not aware of any studies, but it would be interesting to know on what conditions each system depends. Isn’t the Italian way of coordinating traffic a real-world example of what others have called “peer production of governance” on the Internet? Under what circumstances does a decentralized social signaling system (eye contact and honking) trump a centralized law-based regime (traffic lights)? May it work particularly well in communications networks? And could the law do anything to foster it?

As to the latter question, Jonathan Zittrain mentioned after one of his recent lectures that Germany is gradually introducing roundabouts. As research (German only) indicates, this promises to be a big success – and an example of the law facilitating regulation by honking.