Discovery & the Social Graph

“Discover” is a fundamentally flawed idea in social networks. People discover by interacting with their graph, not in an explicit section.

I don’t care what *everyone* is discovering, I only care what people I trust are discovering. People in aggregate have bad taste.

Dustin was likely tweeting in response to the top-level #discover tab in the recently-shipped Twitter products. Whatever you think of the new Twitter, I’m primarily concerned with how broken #discover is in its current implementation. The discovery opportunities that exist for Twitter as a product and for me as a user are vastly under-served.

For example, what in my four years of interaction on Twitter would lead any sane person to think I care about Fantasy Baseball, #CosasAburridas, or Scott Baio? I’m not just listing random stuff; I have the opportunity to “discover” each of those are stories/trends right now in the #discover tab. But I don’t want to, because none of them are relevant to me.

If relevancy isn’t a priority, #discover actually is a list of random stuff. Or more specifically, it’s a list of random stuff I’m being shown by a service who has access to years of my documented behavior (content, RTs, favorites, frequency of replies, etc.) and seemingly doesn’t care because I should care about Scott Baio because…why? Other people on Twitter do? Or it’s easier to selling advertising to advertisers around Scott Baio or [insert trend here] instead of something I actually might engage with? 

The #discover results could be almost any randomized content and it would be exactly as relevant to me as the current offering. How do you sit on that much data about your users, ignore it, and serve up Chachi instead? How is that not a massive, wasted opportunity? Why do you ship a product that doesn’t address it?

I don’t want to discover anything in the #discover tab. My 6000 tweets + favs/RTs should be enough data to serve up customized content. I don’t care what everyone wants to discover, or even what everyone is discovering, I only care about content that’s relevant to me based on my previous and current behavior and/or the behavior and content of people I trust. The aggregate stories/trends aren’t useful to me because people in aggregate have bad taste (or at least different enough tastes that serving the same content to all of them is silly). Surely somewhere in 6000+ pieces of user-generated content there is enough information to make #discover useful to me.

If you’re going to force “discovery” on users, at least give them content they might want to find. Isn’t that in the best interests of your users and your bottomline?