Reputation Systems and the Mobile Web

Athens.
For the first round of paper sessions here at WebSci '09, I've chosen a session on trust and distrust. Having just watched people juggle USB drives for the best part of 15 minutes, we finally start with a presentation by Dave Karpf. His interest is in the Web's impact on collective action for Internet-mediated organisations - and he suggests that the emergent mobile Web wll be of particular importance in this context.

Mobile Web-enabled devices enable new forms of collective action; rating and reputation systems attach track record data to individual participants - when the two meet, this has potentially radical implications for what uses become possible. Reputation in this context refers to complex, context-dependent community assessments; it plays a crucial role in solving collective action problems, and introduces what Axelrod has called a 'Shadow of the Future': they lead people to do well for others as they make visible the contributions of each participant (and introduce possible future repercussions for those who fail to put in). This is visible for example in communities like eBay or Slashdot, which both promote positive and sanction negative contributions through their reputation systems. Even Google's PageRank can be understood as a reputation system: PageRank measures, indirectly, reputation.

For any such system, there are a set of designers working with the community; they need to identify proxies, measurable signfiers of the actions they hope to encourage. Such proxies may be few or many, may come in small or large sets (depending on how difficult it is for the average participant to provide proxy data on their evaluation of others' performance) and may be mobilised in more or less effective (non-gameable) ways - resulting in more or less effective and reliable reputation systems. Today there is an abundance of data (including, increasingly, from mobile devices), and such data may be able to be captured more effectively on the fly, reducing transaction costs for users, and potentially resulting in better quality proxy data for reputation systems.

MoveOn is a good example for such developments: it is working hard to track the quality of offline events organised by its members, in order to focus on those participants who organise the most effective offline events and move away from those who are highly active but very ineffective organisers. Mobile devices are crucial here, and are likely to see the rollout of 'kudos' systems. This may be able to avoid the 'tyranny of the annoying' phenomenon, which holds that political organisations tend to be dominated by those who have sufficient time to devote to organising, rather than by those who are best at it.

There are four reasons for concern here, however: continuing digital divides including class- and education-based cleavages, the use of such technologies by terror or hate groups, a rapidly changing privacy environment, and a broader shift in social practices towards a society that is constantly judging each other - reputation systems may be very exciting, but also very problematic in this regard.

Technorati : Web science, WebSci '09, community, mobile, recommendation, reputation, trust
Del.icio.us : Web science, WebSci '09, community, mobile, recommendation, reputation, trust