Graphing Likes

With big data comes great responsibility. This is what Spider-Man* might have said if he was working for Facebook these days. Facebook has the personal information, photos and likes of over a billion people with more than a trillion connections. That’s a lot of data, so no wonder that privacy concerns are mounting since Facebook announced its new Graph Search a couple of weeks ago, which uses everything we have been sharing on Facebook to let us look up people, places or photos, packaging it into structured information we can actually use.

This is like Google but instead of giving you links to external sites, results are Facebook pages for the people, places or things that match your query, making the search more personal than a traditional search engine. For example, you could search ‘restaurants my friends who live in Dublin like’ to get a list of places to eat in Dublin your friends ‘liked’, look for ‘friends that work at my company who like karaoke’ if you’re looking for a singing partner in crime or ‘new movies my friends like’ if you don’t know what to watch. This all sounds very exciting, but it can also be a bit scary, as Tom Scott’s Tumblr Actual Facebook Graph Searches shows.

Scott has put together some examples of Graph Searches, showing how it provides personal information for all sorts of search queries. Some of these examples are very funny, like ‘Current employees of Tesco who like horses”, but some others  are actually quite disturbing, like “Islamic men interested in men who live in Tehran, Iran”, where homosexuality is considered a crime. Beyond the funny examples, this highlights how Facebook is continuing to fulfil its mission statement of ‘making the world more open and connected’, with everything, positive and negative, that entails.

As with any new Facebook feature, one of the main questions that comes to mind is what Graph Search means for brands and marketers. One of the sources of information that Graph Search uses are brand ‘likes’, e.g. shops my friends like. This will encourage brands to make more engaging pages and get as many fans as possible, aiming to come up first in people’s searches (sort of a Google SEO strategy)  However,  just because you like a brand, that does not mean that you engage with them or have bought / will buy their products.

If I look at my ‘likes’, I would say that my overall public profile seems quite accurate: I ‘like’ books, travelling, fancy cocktails and Asian restaurants. But if you examine my ‘likes’ closely, you’ll find that there’s some ‘dirty’ data in there.  I ‘like’ The Kooples , Whole Foods and Ketel One but also Heinz Ketchup, Glen’s Vodka and even a Walmart in Kodiak, Alaska, which obviously I don’t really ‘like’, have been or think I’d ever visit. The only reason I ‘liked’ it was to send rapper PitBull to the most remote Walmart in the US, as part of an online prank, and the reason for liking Heinz was to take part in a competition if I remember it well.

Like me, many people ‘like’ things because there is an offer or a competition. And more people are now deleting their likes from their timelines after getting what they wanted. So these ‘likes’ are not a ‘recommendation’ and we cannot compare them to people giving a book 5 stars on Amazon after reading it or adding a positive review on Foursquare or Qype. However, this does not mean that Graph Search won’t work.

In the same way that Google does not rely on only one piece of information to determine if a link would come first or fiftieth, Facebook could combine ‘likes’ with other metrics to produce a more accurate ranking. For example, it could use ‘check ins’ or geotagged pictures to confirm that you had a drink at the bar you liked, and being a bit more ambitious, maybe even the comments on pictures and check-ins (e.g: ‘had a great night in here’).

I do believe Facebook Graph search has a lot of potential, but to become a successful recommendation tool, it needs to reassure users on privacy and also encourage people to ‘like’ places in the ‘right way’, so when you search for ‘restaurants my friends like in Dublin’ you do not miss all those lovely places your friends would recommend to you but couldn’t be bothered to ‘like’. On the first, Facebook has already tried to show that it takes privacy concerns seriously, with an explanatory Q+A post  and a video.

The second, shifting behaviours, might require some new ideas to encourage users to give up information (like Tesco does with Clubcard). Paraphrasing from Spider-Man again, if there are things in this world that you have to offer, things you do better than anyone else, things that could help people, then it is your responsibility to do those things. Whether Facebook takes this opportunity (indeed whether it has the ability) remains to be seen. Let’s hope they do – as I’m eager to see the search results of ‘People who let Spider-Man references guide their life’.

Ines Nadal is Head of Trends & Futures at Ipsos MORI. Follow her on Twitter @inesnadal

*actually, it was Uncle Ben who said that, thanks Liam

Campaign Jobs