5 assumptions about social search

A look at the possibilities and issues attached to social search.

As any comp sci major knows, data is nothing more than a pile of facts. It takes meaning to turn data into useful, meaningful, actionable information.

And applying meaning to a pile of data is exactly what’s behind the recent partnership between Facebook and Microsoft’s Bing. The idea is to use social information to select the most relevant search results from the staggering data pile that is the Internet.

Instead of assigning relevance to a given web page the old-fashioned way — because of, say, how many hits it received, or how many times your search phrase occurs in its tags, or how much the page’s owner paid to have it appear at the top of the search result stack — the Facebook-Bing partnership aims to use a new relevance factor: What your friends like.

The reasoning goes like this: Most of us turn to our friends for recommendations when we want to hire a plumber or buy tickets to a play. By tapping into the websites our Facebook friends have surfed to and “Liked,” Facebook and Microsoft hope to be able to serve up search results that are more meaningful. Rating content based on what other people think isn’t new, of course. Amazon has been doing it for years. The thing that makes social search different isn’t just that it attempts to rate content — it’s that it attempts to rate content based on the opinions of people you know and trust.

And therein lies the rub.

Social search makes five assumptions that may or may not turn out to be accurate:

1. That “Liking” something is relevant to all (or at least most) searches

If you’re researching a mobile phone purchase, you might care what your pals think. If you’re looking for the most reliable, comprehensive site for prescription drug interactions, “relevance” probably means something other than popularity.

2. That a single click can convey what someone actually meant when they “Liked” a site

The popularity of the “Like” button (currently on an estimated 2 million web pages) is due in large part to its ease of use. Just a quick click gives people the emotional satisfaction of getting to weigh in on something, of getting to make their voice heard. Trouble is, clicking doesn’t really tell you a whole lot. Did your pal like the site design? The product being hawked on the site? The company? The picture of the celebrity endorser next to the “Like” button? Or did his cat jump on his desk and step on the mouse by accident? There’s no way to know.

3. That you trust the people who clicked “Like” (or even know them)

It’s possible to create multiple Facebook accounts even though they violate the terms of service, and some users are quite indiscriminate about accepting friend requests. As such, there’s a chance that some of your Facebook friends are duplicates — and that others are folks you’ve never met. Both of these factors can diminish the relevance of recommendations.

4. That the people companies say “Liked” something actually did click
a “Like” button

If we’re extending the real-world scenario (“I care what my ex-workout-buddy thinks about that pair of shoes I want to buy”) we could do worse than consider the real-world case of the aluminum-siding salesman who insists that all of your friends and neighbors are already buying (“Come on, you’re the last holdout on your block!”). In other words, it’s not inconceivable that companies who pay a little extra may turn out to be “Liked” just a little more often by your Facebook pals than sites who don’t pony up. After all, how many of us are going to contact all of our friends to confirm?

5. That people won’t get weirded out by having their Facebook and web-surfing worlds collide in such a visible way

It’s been a long time since industry pundits (and Facebook members) were up in arms over Connect, a Facebook technology that shared member information with third-party websites on an opt-out basis. The fact is that Facebook’s partnership with Bing (and other websites) does the same potentially scary thing: It melds what you and your friends do on Facebook with everything else you and your friends do on the web. The jury’s still out on whether the average Facebook member realizes that what he “Likes” today could show up in his boss’ search results tomorrow — and if he’ll care.


Whether or not social search turns out to be useful for the average web user, there’s no question that it will be successful for Facebook — which is perhaps more to the point. Facebook’s deal with Bing virtually guarantees mass adoption of the “Like” button. If getting a good search ranking requires webmasters to get their visitors to click their “Like” button, that’s exactly what they’ll do. And that’s great news for Facebook, whether or not Joe Searcher ends up better off.

Related:


tags: ,
  • Alex Tolley

    If we swap “Like” for “Page Rank”, how would the same questions stack up? Not well I’m sure.

    It is very clear to most users of search, that there is still a long way to go to get the results you want. Just last week, I noticed that “Watson and IBM” flooded the first few pages on Google with references to the Jeopardy match when what I wanted was some information on IBM’s most famous CEO (which eventually resulted in using Wikipedia).

    In the social graph, I speculate that what is needed is something more like a search based on del.icio.us bookmarks. IOW, find people who appear to be experts in certain domains, and use that expertise to flush out good references.

    The Facebook/Twitter approach may be biased towards a “wisdom of my social ditto heads”, which is likely to bias results towards preconceptions, much as we see in political attitudes.

    I’d certainly like to see some actual data on how well different search strategies work on “gold standard” references and more ambiguous searches to understand where the different approaches work best. The probably isn’t any one best way for all search.

  • http://Http://www.facebook.com/byronnewmedia Deano Power – Byron New Media

    Is there any difference between an iframe like button and an sdk implemented like button. I imagine that the sdk like button is more susceptable to gaming that the iframe button from an SEO point of view?