NextBurgh, PA          @VannevarB      About      Pittsburgh Murals
February 25, 2009

Homophily aka Birds of a Feather

Homophily, or homo (same) - phily (love), is the love of the same, referring to our tendency to bond and associate with people who are a lot like ourselves. We're comfortable with people who are similar to ourselves, they share our deities and demons, they're invested in the same status quo that we are, and they don't challenge our assumptions.

Recognizing that there's very little new under the sun, let me point out that we learned this in school. Please complete this sentence: "birds of a feather, _______ _______".

Homophily can be expressed geographically, intellectually, spiritually, politically, culturally, socially - it's all over the place. It's a (mal)-function of social networks.

Maybe that's why Pittsburgh BlogFest 17 was so comfortable.

It's Why everyone you know thinks the same as you - you've selected each other.

Homophily is a problem in social media, the new darling of Web 2.0. If you're building a social media space, homophily is both a feature and a bug. It helps build community, but then it limits growth and makes the conversation polarized and shrill.

Homophily is a problem in web information systems, too. If people are free to self-select content, they will select the content and sites that reinforce their homophily. Drudge / Koz, po-tay-to / po-tah-toe.

In an information economy, drawing most of your awareness from sources that comfortably reinforce your own Weltanschauung can lead to isolation, group think, and data corruption through inbred feedback. Bad decisions, bad investments, bad things all follow.

How do we combat homophilia in ourselves? Tryangulation blogs about a homophily self-exam. We can seek multiple sources of information (I read The Atlantic, The Economist, and Utne Reader, and I surf for news just to seek outside perspectives - but I bet most of my friends are doing the same thing.)

How do we combat homophily in the information systems we build? It helps to include affordances that easily extend out of the comfort zone. It also helps to build in a certain amount of serendipity, which Ethan Zuckerman believes is a counter-force to homophily. So, for instance, if you're building a recommendation engine, instead of rejecting outlying points that don't fit the profile, you use them as bridges to a new potential cluster.

It's an interesting question that calls for a dynamic serendipity function in a content management or layout system. Can the site glean clues, serve temptations, build a new normal, then move on? How about a Serendipity-As-A-Service (SerAAS) server? That would truly be a better mousetrap.



Post a Comment

Comments and Feedback? Love that stuff. Please leave your thoughts in the box below--