Excellent article by Jon Evans of TechCrunch. He first sums up well what are Facebook’s strengths as of today:

First, of course, there’s the network effect; with a billion active users, everyone goes there because that’s where everyone is. (Of course, there are individuals and subcultures who avoid FB — perhaps quoting Yogi Berra, “Nobody goes there any more, it’s too crowded” — but whether they like it or not, they’re a tiny minority among the networked of the world.)

Second, their Open Graph strategy means that Facebook isn’t just Facebook; they’re woven into the bones, and in some cases arguably the DNA, of millions of other apps and web sites. Every “Log In with Facebook,” “Share to Facebook,” and “Like” button across the Web and within every app store is Facebook territory. They don’t just own all the end users: Developers, executives, marketers, startups — they all want access to the Facebook empire, which means, to a considerable extent, becoming part of the Facebook empire.

He then gives some of the factors that could accelerate Facebook’s certain demise such as users turning away for another service or that users are getting bored of social networks in general. His last point is of interest:

So the only thing that I think will actually kill Facebook, in the end, is the same thing that’s slowly killing Microsoft today: a platform shift. Not the shift to mobile; they’re riding that one out nicely. The next platform shift. When it happens, maybe Facebook’s death spasms will finally trigger a long-awaited spate of innovation and creativity from them, as seems to have happened in Redmond of late. If I were a betting man I’d put my money on wearable computing as the platform of the future — something like Google Glass plus Kinect, bearing in mind that each of those is but the Altair II of its field, and I’m talking about their descendants a decade or more hence.

Wearable computing is then the next platform shift that will kill Facebook. But what if Facebook had already planned a few things to enable people to “like” objects in the real world?