Serendipity is not randomness. It is unexpected relevance.
I constantly hear the fear that serendipity is among the many things we’re supposedly set to lose as news moves out of newsrooms and off print to online. Serendipity, says The New York Times, is lost in the digital age. Serendipity, it is said, is something we get from that story we happen upon as we flip pages, the story we never would have searched for but find only or best in print. Serendipity, it is also said, is the province and value of editors, who pick the fluky and fortuitous for us. Without serendipity, as I hear it, we’ll be less-well informed (all work, no play, makes Jack a dull boy; all relevance, not serendipity, makes Jill a predictable girl).
A few days ago, a Guardian guy, inspired by Clay Shirky hacked together a serendipity generator: just a random story served up on a click. It wasn’t a serious solution, just fun. But it focused the serendipity question for me.
What is serendipity? It’s not a story from left field. It’s not, I think, “the opposite of what you normally consumed.” There’s a reason we find value in the supposedly serendipitous. When I started Entertainment Weekly, I said that our features had to satisfy a curiosity you didn’t know you had — but you end up having it. When we read a paper and find a good story that we couldn’t have predicted we’d have liked, we think that is serendipity. But there’s some reason we like it, that we find it relevant to us. Maybe that relevance is the unknown but now fed curiosity, maybe it’s enjoyment of good writing or a certain kind of tale, maybe the gift of some interesting fact we want to share and gain social equity for, maybe it’s a challenge to our ideas, maybe an answer to a question that has bugged us. In the end, it has value to us; it’s relevant.
Can that relevance be analyzed and served? Can we still get serendipity online? Of course, we can and do — mostly on Twitter and Facebook. Serendipity comes from friends who find that story and — like an editor — pass it on. If we share their judgment, we may like what they share and call that serendipity. But there’s plenty that passes me by on Twitter that I don’t like; it’s serendipitous by the usual definitions but it doesn’t work for me because it has no value; it’s not relevant.
Can an algorithm serve us serendipity? Maybe, if it has enough signals of what we and people we trust like, what interests us, what we need, our context. It can calculate and predict and try to serve our relevance and serendipity. I think serendipity comes not from one-size-fits-all editing but from better targeting across a larger pool of possibilities. If Google can intuit intent, I think it can also serve surprise and serendipity.