A recent New York Times article titled “The Human Touch That May Loosen Google’s Grip” has sparked some fresh discussions about adding more people-power into the search relevance equation — a subject near and dear to our collective hearts here at Collarity. Primarily a reaction to the entry of Mahalo, a new search engine site that enlists the aid of human “Guides”, the article points to the trend of balancing machine-based algorithms with “editorialized feedback”.
Google’s Matt Cutts was interviewed in the article and offers his own “The role of humans in Google search” post on his blog. Matt says Google’s current methods include plenty of human input. The article was also covered by Eric Enge in “Human Input and Algorithmic Search” at Search Engine Watch Blog and by Barry Schwartz in “Google’s Human Touch” at Search Engine Land.
Our take? We still believe better results will come from implicit people-power — anonymous behavioral data used to distill search relevance that is not dependent on “explicit” user feedback. Think of John Battelle’s (his take on the NYT article here) “database of intentions” and the emerging concept of “the implicit web“.
A much richer user-generated dataset can be developed by avoiding the pitfalls of low participation rates (people won’t do the extra work) and of survey bias errors (partial or false information). Early implicit people-power successes will be applications related to site search and content discovery for specific websites. Later, as more data is collected, we will see improved relevance for web-wide index searches using implicit user feedback loops.