Situating Social in Search

There were two interesting announcements this morning, one from Microsoft and one from Google, about how the companies would be revising the role of ‘social’ within their search products. The first was Microsoft’s announcement of So.cl, which has been billed as a “social networking service”, but appears to be more along the lines of a cross between Pinterest and a collaborative search system. The second is Google’s announcement of a button which would let you toggle whether search results are personalized or not.

So.cl: While So.cl is not yet available for public use, Microsoft has posted some information on the domain. On the FAQ, they explain that the new service is ‘focused on exploring the possibilities of social search for the purpose of learning’. Though definitions of social search vary widely, what they describe seems to be more specifically akin to ‘collaborative search’ (for a good review of collaborative search, see [1]) where users are explicitly sharing search results with each other (‘in the way students do when they work together’, as the FAQ continues). I am happy to see that Microsoft does not intend to build ‘yet another social network’, and the landing page currently prominently features a Facebook sign-in button, which is very encouraging. I am eager to see how this plays out – this could turn into a very useful tool, as collaborative search tasks (such as research, travel planning, etc.) are still not very well supported by common systems. For a more in-depth analysis, check out the Technology Review post on the topic.

Google: At first glance, Google’s blog post about the personalization toggle makes it sound as if you are gaining the ability to ‘turn on’ personalization of your search results. Much of the post is dedicated to the fact that ‘social’ results will now appear in your Google search results. In fact, the real implications are quite the opposite, as Google started tailoring search results to individuals several years ago (see Sep Kamvar’s blog post from 2007 on personalization in search results). This has fueled worries about what Eli Pariser has called the ‘Filter Bubble‘, the very real phenomenon where increased personalization by web sites such as Google and Facebook actually serve to restrict our knowledge by reducing the diversity of what we see on the Internet. More concretely, in the worst case, this could mean that two searchers, perhaps one conservative and one liberal, could receive two different sets of results for the same query, each conforming to their own preferences and opinions; they may each be happier with the search results, but we can’t argue that they are more informed, especially when this filtering occurs silently behind the scenes. Google’s introduction of a toggle to turn off personalization, however, appears to be an important step in remedying this problem both by allowing users to turn it off and by signaling that it is happening at all.

The first steps in social search were, perhaps, obvious ones: searching status updates (like Twitter Search) or explicit Q/A (like Yahoo! Answers or Quora). Together, I think that these two announcements are a signal that ‘social search’ is starting to come out of its infancy and companies are starting to adopt a more nuanced approach to using local (friend) information for filtering global content and using global information to contextualize local content.

[1] Paul, S.A. and Morris, M.R. (2011). Sensemaking in Collaborative Web Search. In Human Computer Interaction Special Issue on Sensemaking (Special Issue Eds. Daniel Russell and Peter Pirolli). 26(1), 38-71. (PDF)