19 January 2013

Facebook Graph Search has potential, but lacks relevance

So this week it finally happened, a move many have expected for years: is getting serious about search! I don’t have access yet, it will be slowly and carefully scaled up from a limited start in the US (you can join the waiting list here, by the way), so I can’t say anything first-hand about the quality of the results, but some criticism comes to mind nevertheless. It’s not the first launch by Facebook that has the tech media crazy with headlines and how it will revolutionize or kill this or that competitor – Messages comes to mind – but in my opinion it’s unlikely to do that for several reasons:


Facebook Graph Search hands-on by The Verge
  • As learned early on, the most important feature of search results is relevance. Web search is a constant struggle against bad SEO practices and bad content and Google regularly tweaks the ranking algorithms to keep them at bay. As it’s standing, Facebook’s data is already dirty and irrelevant; ‘liking’ has become a very nebulous term after years of brand campaigning for users to like them in exchange for access to some exclusive fad, after the ‘Like’ button spreading all over the web in the most disparate of contexts. It remains to be seen if Facebook can somehow weed through these false signals to surface something useful for people.
  • The natural language processing certainly looks promising, but it’s in no way unique, competing with Apple’s Siri and Google Now. While these other options lack the social component, they tap other data sources that could prove more relevant, depending on the situation. Keeping with the dentist example, I prefer a dentist closer to home or to where I work than one recommended by friends, but far away – in this case the social signal doesn’t offer me a better choice.
  • Another weak point is the amount of data available. While it’s true Google can’t access most of the data in Facebook’s walled garden, the same is true for Facebook not being able to crawl the open web – unless, I presume, the site has added some OpenGraph markup. As such, Graph Search can only become a complement to Google search, not a replacement. Despite its ambition to collect all social interactions in a single network, Facebook is far from such a goal and new challengers step up every year, like Instagram and Snapchat. Some data will still come from people logging in to Facebook apps and adding custom activities to the Timeline, but Graph Search is a long way from integrating third party apps.
  • A similar argument can be made about how new data gets indexed. Google crawls the web without any user intervention – other than authors linking to other sites – and is slowly moving to ‘implicit’ search, where results are offered to the user based on past behavior, without the need for explicit queries. On the other hand, Facebook is dependent upon users keeping their interests and social graph up-to-date; that’s unlikely to happen simply because most users will like and friend often, but seldom unlike und unfriend. The Graph just gets bigger and stuffier, but also noisier. There needs to be some sort of weighting of likes according to engagement with the liked object and the time since the last engagement (i.e. between two restaurants I liked on Facebook the one where I checked in more often should be the one I actually like more).
  • Speaking of engagement, this will be another factor in establishing the success of Graph Search. More to the point, I use Google search dozens of times a day, for personal or work-related queries, from the desktop and smartphone, for the most trivial or obscure questions. I simply can’t imagine having the same level of activity on Graph search – maybe in the first days when I get access to try it out, but later it will be more like a query every couple of days. If Facebook plans to put ads next to this search engine, this will make them far less attractive than Google ads.
  • And where’s the mobile strategy? The competitors already have a lead here and Facebook might have a tough time finding a mobile OS partner, other than Microsoft. Google will definitely not give Facebook any extra access to Android phones. And Apple, after the precedent rejecting Google’s attempt to get more data about iOS users from the maps app, will probably reject similar attempts from Facebook and block Graph Search from natively competing with Siri.
  • Last but not least, privacy will remain a concern, despite the official assurances, especially since Facebook decided users can’t opt out of appearing in the search results. Another potential privacy loophole could show up when users change privacy on posts from public to something more restricting; if Facebook keeps an index to deliver results, these updates could show up in search a long time after the authors hid them from the public.

But enough speculation without having used the product myself! We’ll probably get a better picture of this feature in a couple of months, as it rolls out to more people – hopefully me included!

Post a Comment