I was contacted by a dutch journalist who’s writing an article on the merits of social interaction versus search engines. She read a paper of mine and emailed me with two questions. I thought it’d be useful to post my reply publicly:

First, do you think search engines making use of social networks will improve search results and thus make our daily life a bit easier?

Yes! A lot of fact finding and information discovery already comes from friends, colleagues, and even acquaintances. Online social networks organize our personal relationships in a way that reduces the barriers to information exchange. At the moment, social networking sites aren’t set up for search per se—Facebook and Twitter want to get into search, but with Facebook, they can only reveal what your friends have OK’d to be public, and with Twitter, there’s enough noise, spam, and unknown people making claims that the results can be hard to trust. But, there’s already evidence that people are turning to these online social networking sites to ask their friends questions. On Twitter, these have been dubbed “lazy tweets”:

“Wondering where Bowtie saves it’s themes… Anyone? #LazyTweet” –@iphone360

“Looking for social media trends in the healthcare industry. Anyone out there have resources they can share? #lazytweet #healthcare” –@chrismevans

“Anyone know if there are Brocade SAN and Cisco MDS simulators? #lazytweet #healthcare” –@sloane

The real value for social search is making the search experience more personalized. If search engines can make use of existing ties, relationships, and data coming from social networks, they can use that data to bubble up results that come from a trusted friend network that may be as relevant (if not more relevant due to the trust factor) as traditional search results. The risk, however, is that our personal networks are narrow and not every search we perform may have counterpart results from social networks. This is why search algorithms will continue to play a large role in search, regardless of how “social” it gets.

Another benefit of using social network information in conjunction with search is that a services can begin to “learn” which of your friends have expertise or knowledge about certain topics. Then when you search for those topics, people from your network who may have relevant knowledge could be made available to you. It’s still unknown how visible searchers want other people to be in the search interface. People may only appear as a search result listing, linking to their profile or email address; or they could appear as a direct contact, like through an instant messaging window on the same page as the search results. Either way, the point is that direct person-to-person conversations can greatly supplement an information discovery process (as we pointed out in the “Do your friends make you smarter?” paper).

Another interesting area for social networking support during search is for searches that present difficulties. Anytime you can’t find what you’re looking for on the first try, or you rework your query over and over again — these are use cases that could benefit from asking a friend a question, pinging your social network, or finding a colleague/acquaintance who may have experience with this particular problem.

And second, do you expect these social search engines to become available in any near future?

Yes and no. Yes, in that there are a number of services that provide human answering — but these mostly do not include a search algorithm component, meaning that results are only driven by a direct human contribution. Such services include Aardvark, Cha Cha, Hunch, Quora, and Mahalo…

Another popular class of social search services only makes use of aggregated social data from large networks. I call this “collective social search” since it’s like the wisdom of crowds effect, in that you can see trends from the collective that might be useful in guiding your search. Google Search Suggest is an example of this — it shows you the common search phrases for a given few words. Twitter’s Trending Topics and OneRiot are similar. I think the popularity of this approach is that it’s algorithmic, so you can throw more programmers at it and hopefully improve the results. But it’s quite limiting in its utility since searchers will trust people they know or people who can be vouched for, whereas trends across an entire network have no intrinsic relationship to the searcher. Such results may help in the early stages of search when you’re still trying to formulate an ill-formed query, but won’t be as useful when you want to narrow down to a specific answer to your question.

Thus, the kind of “social” component I want to see in search will require combining both of the approaches I mentioned above. This is not trivial, and there are a lot of unknowns about how people will respond to a service that does this. How will people react if their search results are shared with their social network? Will it be different if we see how valuable it is when our networks’ search results are shared with us (the reverse case)? How will reputation and obligation come into play? How will reactions differ by personalities? By location? By past history? By the political climate?

If Facebook’s Beacon experiment taught us anything, it’s that “social” can’t be solved by an algorithm. We’re still a ways off from really solving social search.