Facebook yesterday demonstrated its new service, Graph Search [2]. The company is proposing to let people search their "social graph" to find new, useful information. (The social graph is the network of relationships between you, your friends, and their friends. Sadly, it is not a search engine for graphs.)

The company is positioning the search service as a competitor to, well, everything. Search for "music my friends like" – graph search is designed to take natural language input – and you've got something which can take on Last.fm [3] or This is My Jam [4]. Search for "restaurants in London my friends have been to" and you've got a rival to Foursquare [5]. "Friends who work for PWC" could fill the same niche as LinkedIn [6], and "Photos taken before 1995" offers something which only Flickr [7] does half as well.

The potential is huge, and, judging by Steven Levy's exhaustive behind-the-scenes account [8] (impressive too for the total absence of leaks it resulted in – the man can keep a secret), Facebook is betting the farm on it.

But there're two potential speed-bumps ahead for the company. The first is that perennial Facebook bugbear: privacy. The company is careful to emphasise that only things which are public or shared with you will show up when searched for — but that relies on users understanding how privacy settings actually work, which has historically not been the case. That's not Facebook's fault per se, but it also won't save them from a user back-lash. And as the company has learned before, while it recognises a binary "public/private" divide, most users don't think in such black-and-white terms. The launch of the News Feed, way back in 2006, was widely opposed by existing users, because despite merely aggregating content which was already visible elsewhere, it felt like an infringement of privacy.

Consider: Person A rejects friend requests from Person B who is a creepy stalkerish character. They nonetheless have several mutual friends. Can B search for "Posts by A which friends have commented on"? (Those posts would be visible to B now, but not aggregated in any one place). Similarly, someone who checks into a specific location on a regular routine might not appreciate that suddenly being aggregated together, making the routine clear to all.

Where privacy is emblematic of Facebook's past concerns, the other problem Graph Search faces strikes at the heart of where it's future problems lie. The usefulness of the service is directly tied to people using Facebook the way Facebook wants them to. That means liking a lot of things; filling in all your personal information, and keeping it up to date; checking in every time you go out; and making all of that public, or at least softening your privacy settings.

For many, Facebook has become a glorified PA: it's a way to contact friends whose other details you have lost, and a way to bulk-invite people to social events, but as a social network, its utility is fading. Graph search doesn't seem to do anything to reverse that trend, because it doesn't offer any incentives to change the data you put in to Facebook — just change how you get other people's data out.

Of course, hovering unspoken during the launch is the key question: will this make more companies want to advertise on Facebook, or increase the amount the company can charge for space? The technology underpinning the search will almost certainly help the company provide better services to advertisers, but being useful — to admen or end users — doesn't necessarily translate into revenue.

What the service does demonstrate is the foresightedness of Twitter's broadsides against Instagram [9], the Facebook-owned photo network. While Instagram was only using its Twitter connection to enable people to export their relationships to the service, Graph Search reveals that the knowledge of those links — the literal social graph — can have intrinsic value. By limiting Facebook's access to Twitter's information, the latter has guaranteed that the former will have to try that little bit harder to get useful results from infrequent users — as well as reserved the possibility that Twitter can launch their own version.