Search is the great triumph of computer science and mathematics. A multi-billion dollar industry was built from a highly technical paper about random walks on the web, which was becoming more obtuse as it grew exponentially.
Google's search breakthrough ensured that the web would not be a victim of its own success.
Now, the social web faces a similar problem. It is enormous, and growing, and central to our lives. There are many successful companies in the social space, just as there were search leaders before Google emerged. Yet so far there is no Google for the social graph.
It's a huge opportunity. But the challenges may even be more daunting than dynamically assigning relevance to any given webpage — as huge an idea as that was when Google re-invented search with PageRank.
There are obvious similarities to the challenge of indexing the social graph, but special problems as well.
There are obvious similarities to the challenge of indexing the social graph, but special problems as well. For one thing, there doesn't appear to be anything to generalize over the entire social graph, so maybe there's no search-level problem that needs to be solved: Perhaps it's a collection of specific problems.
Like the larger web graph — the sum total of all web pages and the hyperlinks — the social graph is people and the connections among them. A non-virtual social graph has always existed: People get married, have children, have friends, are employed, and so on. More and more non-virtual relationships have an analog in the digital realm, like a "real" friend who is also your Facebook friend. But some relationships exist only in the digital realm — poking someone on Facebook or following someone on Twitter.
To be fair, there are specialized applications within specific social networks, such as LinkedIn's "People you may know" feature and Twitter's (grammatically suspect) "Who to follow." Other applications of social web data are often domain-specific: last.fm recommends music that you might like, and Netflix recommends movies that you might like by looking at the preferences of millions of people.
What's needed is something which links up these islands, where we all live part of the time, into a single, contiguous nation.
It won't be easy. I'd like to offer up four challenges that I find important, though undoubtedly there are more:
- What problem are we trying to solve? Search solved the problem of proliferation of web pages that were no longer captured by directories. A good question to ask is: What's the guiding central problem of the social universe?
- A person is the sum of all of their profiles: Identity across social networks must be solved. Linking Facebook, Twitter, Google Reader, LinkedIn, etc. would be invaluable to researchers. Actions across social networks are similar (liking, following/friending, sharing, etc.), so to have a complete list of actions from a single individual across networks would vastly increase the amount of data available from looking at a single social network.
- Every user has her own slice of the social graph: No two social graphs look alike, whereas the web graph looks the same for everyone.
- Let data be free: Many types of social data are not public or are difficult to get. All Twitter data is only accessible to the select few members of the firehose club. Facebook data is available for only a select few users. Search was made possible by web crawlers and a similar accessibility of data must be in place for the social graph. Of course, accessibility of data brings up lots of privacy concerns.
So, cool things are being done with subsets of the social graph, but is there going to be one company to rule them all? Put another way, web graph is to Google as social graph is to … what?
Many new players, including my company, are betting on discovery as the answer. Today, discovery is applied to specific genres — restaurants, movies, books, friends — but to give you recommendations, it needs to harness and use a lot of data from the social graph. In theory, once you've done one genre well you should be able to do the others.
I'm interested in your opinions. Have I missed a company that's using the social graph in a really unique way? Perhaps I'm asking for too much? I'm still optimistic: That there will be a Google-equivalent to the social graph and that company will be the next big thing.
Photo: Connections among Twitter users who mentioned "social graph" on May 18, 2010 scaled by number of followers, Marc Smith/flickr. Used with gratitude via a Creative Commons license.
This entry passed through the Full-Text RSS service — if this is your content and you're reading it on someone else's site, please read the FAQ at fivefilters.org/content-only/faq.php#publishers. Five Filters recommends: Donate to Wikileaks.
No comments:
Post a Comment