It's not true that "Google Doesn't Do Social"

September 10, 2010 Dennis D. McDonald

Reading about the new Google Instant search feature reminded me of the fundamentally “social” nature of Google and why saying “Google Doesn’t Do Social” isn’t really an accurate statement.

It’s the search and linking behavior of millions of people using the Web that provide the foundation for Google’s search algorithms. An individual user’s search behavior is tracked along with what others are doing. Google stores and processes data about this group behavior. Extremely clever programming and system engineering then allows the Google Instant feature to be instantly responsive not just to the user’s search but to what Google already knows about what the searcher — and others like the searcher — are doing.

The fact that there’s no “like” button or “thumbs up/thumbs down” icon associated with Google Instant doesn’t mean that social or group behavior doesn’t figure into the process. Just the act of selecting links — or Google supplied ads — provides both individual and group data Google then feeds back to its algorithmic infrastructure for future processing.

So, yes, conducting a search is something that an individual initiates, as is the selection of the links to follow that are displayed from that search. But what what goes into Google’s presentation of intermediate search results, now available instantly, incorporates much information based on group behavior. That’s why I say it’s not entirely accurate to say the “Google Doesn’t Do Social.”

Given this view, it’s interesting to speculate how Google could add social elements more explicitly or transparently to its search offerings. For example, what if Google provided a dashboard that let users set a “Search persona” to guide the presentation of intermediate results? For example, consider the different reasons for why a particular search might be conducted, e.g., to support a task at work, to support an activity with family or friends, to support a household item purchase, etc. If you could easily specify this type of search goal or application, then Google could use that information along with information supplied by others performing the same role — to further “tune” the intermediate and final results in a search. Basically, you’d be asking Google to “…show me what people like me are searching for.”

Of course, searching for “what people similar to me” are searching for is already a well-established function in many different online venues that take advantage of various data segmentation practices to slice and dice behaviors, transactions, and groups. Some require explicit user behavior such as Netflix and its movie ratings as input to recommendations, while others are more passive as with Amazon’s book recommendations. Joining or following a group or individual in Facebook or Ping can also provides segmentation information based on self-selection that defines a population of like-minded individuals from which focused information and recommendations can emerge.

What Google is doing with Google Instant is beginning to expose some of the inner workings of its internal processes in real time. Making the fine tuning of this process more “social” and also responsive to realtime user tuning could be next. This might be one way that Google Wave based technology finds its way into Search.