Skip to content

The vicious cycle of reputation systems

The Total Economic Impact™ of Guaranteed Fraud Protection™

Get case study: “The Total Economic Impact™ of Guaranteed Fraud Protection™”

Cover of the Forrester TEI for Signifyd's Guaranteed Fraud Protection

One of the interesting things we noticed when we started looking at social interaction abuse was how reputation systems were built and evolved. A need for some kind of reputation score usually appears when your site becomes popular and you want to help users find quality content and interesting users quicker; but some destinations (such as marketplaces) use them to actually facilitate interaction. Either way, very soon it becomes
very lucrative to be ranked high – whether it brings you more sales, better SEO or just respect and a sense of self fulfillment.

But there is a trap with reputation scores, that you only discover after the fact: paraphrasing on a few meetings with Randy Farmer discussing Yahoo! Answers and other projects, once you build your score you discover that placing all users on the same scale is super, super hard. Good and bad users present varied behaviors, and once you set up basic scoring functionality you discover two things: one is that too often you get unpredictable, changing scores that furstrate your good users, and the other is that it’s very easy to game your rules to get a better score. So you improve and you change and eventually end up with a mish-mash of rules that no-one can effectively follow, and an army of moderators trying to enforce them.

We think that reputation systems should be straight forward and rather easy to understand, and protection mechanism against gaming the system need to be built separately. This is why we built Signifyd – it does not replace your reputation system. What Signifyd does do is that it tracks patterns, signatures, of good and bad users and the way they go through your product – based on their actions and the matches we find to historical patterns we (completely anonymously) harvest from multiple systems we work with. The system works pretty much like an anti virus to detect those behaviors and bring them to your attention – and improve when you give it feedback. This is a tool that is complementary to reputation and gamification systems and allows them to function properly while you grow your community.

Do you have war stories about reputation systems? Want to tell us more about the ways users tried to trick you to get ahead? Share with us in the comments!

Signifyd

Signifyd

Signifyd provides an end-to-end Commerce Protection Platform that leverages its Commerce Network to maximize conversion, automate customer experience and eliminate fraud and customer abuse for retailers.