The Importance of “People like Me” features

People like me features are one of the most promising ways to help people find content that is interesting to them.

Jason Kottke points to a study in which researchers found evidence that the brain reacts differently to people who seem like us.

This isn’t surprising, of course. We do tend to react differently when we feel like we’re around a like-minded person.

But how can this help inform design?

We already see many features which take advantage of this, such as grouping features, demographic filters, “viewers like you”, and many others. Folks mentioned around the idea of Circles of Relationships.

In addition, many recommendation engines are built on what’s called “person-based collaborative filtering” (see Wikipedia: collaborative filtering). When Netflix figures out what movies to recommend to you, what they’re really doing is assuming that people who have rated like you in the past are the best predictors of future ratings. You can get a sense of this from the “similarity” number that shows up in your friends pages on the site.

However, Amazon’s “people who shopped for this also shopped for” isn’t actually person-based collaborative filtering. They use “item-based” instead, meaning that they collaboratively filter based on what items are purchased at the same time, regardless of who purchased them. (see David’s comment below, he also talked about the negative reaction “people like me” sometimes gets)

In my own research I’ve found that many people read comments, reviews, and other online material and make some judgement of how similar they are to a person. If they are similar, they’ll weigh that person’s opinion much more than others. If they aren’t similar, if the person doesn’t seem to have the same values or appreciations, then we give them less weight.

Do you know of any great examples of a “people like me” feature?