![]()
Recently I edited Building Web Reputation Systems by F. Randall Farmer and Bryce Glass, to be published by O’Reilly Media and Yahoo! Press this March. Think of a star rating for your favorite consumer product — that’s a familiar example of a reputation system at work. But star ratings are only a small part of the story that a reputation system can tell.
Through artful construction of algorithms that process combinations of data types, it turns out that you can uncover surprising and powerful information about users’ motivations and intentions and the quality of user-contributed content on a social media site.
As an example, suppose that you’re managing an online community for parents, and you notice that in the last few days, you’ve had to delete a growing number of posts that are just advertisements or rants unrelated to the topic threads. The posts come from a number of different users, and you’ve banned some of them (the technical term for them is “trolls”
), but the problem persists.
By itself, the user ID — one type of data about people using your website — probably won’t tell you much about what’s going on. But suppose that in addition to the user IDs of the trolls, you also know the IP addresses of the computers where users are signing up for the site. When you look at the IP addresses alongside the user IDs, lo and behold, you discover… the trolls are all using the same computer! You shut down all access to the site from that computer. The problem goes away.
That’s a simple example. Through iterative design and testing, it’s possible to design complex, powerful reputation systems to handle many kinds of business problems. Depending on their purpose, these systems may have obvious, public user interfaces, or they may sort, prioritize, and act behind the scenes, unnoticed by the average user.
Author Randy Farmer, one of my colleagues at MSB Associates,
coinvented many of the basic structures for both virtual worlds and social software. Bryce Glass works on Internet community products and platforms with well-known brands. Both were on the team that developed Yahoo!’s reputation platform. Read chapters of the book at buildingreputationsystems.com. ![]()
