Search engines are proving more effective by the day at turning up high quality results for our queries. Google’s indexing and ranking of content has ballooned to a level approaching totality faster than most of us would have imagined when PageRank debuted merely a decade and a half ago. They are no longer indexing just web pages, but books, video content, photos and the layout of the entire globe, complete with images. Google and its competitors are getting better every day at showing us what we think we want to know, but what about what we didn’t know we wanted to know, or weren’t sure we wanted to know?
These kind of tangential almost random connections between ideas are the basis for human creativity and innovation. Realizing connections where none had been noticed previously is such a mysterious and powerful force for the human experience it is often chalked up to the divine. This is the domain Google is trying to enter with its new project known as Knowledge Graph. As reported by Tim Adams in this Guardian article (http://www.guardian.co.uk/technology/2013/jan/19/google-search-knowledge-graph-singhal-interview) Knowledge Graph is in many ways like Facebook’s Graph search. Both seek to quantify links between topics as a way of issuing better recommendations for the searcher.
There are two differences discussed in the way Google approaches it. Firstly they are using an algorithim which gives weight to what they call long-clicks and short-clicks, basically a classification of the usefulness (or at the very least interestingness) of a page to a searcher. Long-clicks are the pages which users remain on for a substantial time; short-clicks have users clicking the back button to try again. I feel the strength of this is allowing for a more human weighing of what types of pages pique interest related to certain searches, though this in itself isn’t terribly different from the way Google gives results, just further refined. The other feature has the potential to be a real game changer.
Though Knowledge Graph will still return the results it thinks most desired first, further down the list an element of randomization comes into play. The developer doesn’t get into much detail of how this would function, but I think it could really change things up and seems to run contrary to the increasingly insular way Google is functioning. Perhaps this would be a way to really deliver on the promise of access to all the ideas in the world, not just the ones we thought we needed at a time. Maybe in a few years people will be fondly recalling inspiring Google searches, the random results of which moved their life in a way they never expected. Or maybe it will just fill my head with even more tangential trivia than Wikipedia has. Either way I think this is an idea unseen in the other engines of search and it will be fascinating to see how it is developed and implemented. Any thoughts?