The Wall Street Journal has written an article entitled You’re a Nobody Unless Your Name Googles Well. This front-page article has been syndicated to many newspapers around the country, causing substantial buzz. It reflects somewhat of a new reality – that your “search identity” is a resume of sorts. In fact, many employers place a large amount of credibility in search engine results, which is troubling as you don’t have presentational control over your search identity (or what people write about you).
There are a number of companies struggling with this problem currently, and ClaimID is one of them. Techcrunch profiled a number of identity search engines, and services like Reputation Defender can be retained to monitor your identity online. Put simply, there are a number of companies addressing the problem of identity from a number of different perspectives, and the market has not spoken in any particular direction.
Identity is one of the internet’s last great unsolved problems. The reason this is the case is not because of lack of effort, but of the sheer complexity of the problem. Identity search, for example, attempts to disambiguate name-based entities. Essentially, it attempts to tell one John Smith from another, an extremely difficult challenge (many Ph.D.’s have been written on this challenge), and the models we have work best in constrained environments, not the internet.
Beyond the problem of name-based entities, there’s another big problem with the net – the stuff you’ve done but your name doesn’t show up on. Sure, a search algorithm can do basic cocitation analysis to guess at stuff “related” to you, but it won’t find a good deal of the stuff you’ve done, nor will it understand the relationship. The company you worked at, the project that you worked on that was written up in the newspaper that doesn’t mention your name…all of these things present identity search some very serious, potentially unsolvable problems. A computer would need to pass a turing test to fully address this problem – Bayesian models can only take us so far.
The approach that seems to be popular in identity search is a hybrid of search + claiming. Knowing that models will never fully disambiguate or find any one individual, the search engines allow individuals to claim related results, creating a dossier of sorts. Of course, this is the approach we’ve always taken in ClaimID – you know yourself, and we’re not going to try to design an algorithm that knows you better than you do.
Of course, part of me wants to believe that these companies can do it better. I want to see a company come along with an approach that is revolutionary, that promises real results. I believe that the challenges of managing search identity present the information sciences one of its greatest challenges over the next ten years. People need these solutions, and the market is not going to get smaller. But what exactly are the solutions people need?
Outside of the magic laser beam that erases links you hate and raises your favorite links to the top, I think we’ve got to take a reality based approach. Research and work on name-based entities will continue making the models better. Standards and open-source approaches are a must, as identity simply cannot be centralized. The market has proven this again and again. Identity must be decentralized. Finally, we must accept some realities. Largely, people will have their identity searches be mediated by Google. (Google sends a tremendous amount of identity search to ClaimID, with Yahoo search owning a very small part of our traffic.) People will also need trasportable, web-wide solutions. The idea of fixing identity in one place is fine, but what about the rest of the internet? For many, Google is the internet, so we’re just playing in their playground.
As you can see, there are some tremedous challenges in this sector. However, that’s what makes this sector exciting and interesting – its one of the last places on the net you can make real change that will make people’s lives better. And is there a better goal than that? It has certainly kept us motivated here at ClaimID. Ultimately, identity is a solvable problem. Major vendors like Google, IBM and Microsoft might have to start paying better attention, and upstarts certainly will contribute to the discussion. I look forward to the progress to be made in this sector…it will be interesting to watch over the next ten years.