New Scientist has up a great article on the status of grand unified theories. It's very critical of superstring theory. Unfortunately the article in full is available only to subscribers. But Not Even Wrong has a nice writeup about it. Be aware though that he's fairly anti-string. Still, there are some good points made in the article.
The big problem with superstring approaches is that it simply explains too much. On the one hand that might be natural. Consider evolution. We don't want to say that there is some necessity of a duck evolving. Why should we assume that the basic laws of physics must break symmetries in a fashion that produces something like this universe of necessity? It's certainly true that this has been the historic bias in physics. The old necessitism of Newton and the Rationalists seems to raise its head.
My personal opinion is that the final theory will end up being more akin to evolution than perhaps some might like. i.e. it is intrisically open regarding possible kinds of universes and forces.
The big problem with superstring is that one has to assume space and time. It doesn't really explain them. (Everytime I get my hopes up regarding some form that does, they are dashed) In a way Loop Quantum Gravity does do this. One of the reasons that LQG folks like it better than Superstring theories. But then there's a feud between those folks. (grin)
Somewhat along those lines Lee Smolin is the only one who suggest that the lack of success, especially empirical success, suggests anything. He says, "If you look back over the last 200 years, every decade or two there's a dramatic advance, people always understand something new that couples theory and experiment... I suspect there is some right question that we're not asking." Probably a good point, although I suspect that the energy levels are simply such that it doesn't mean much. There are some experiments coming over the next 10 years that will probably tell us a lot.
Overall the article is worth reading even if you think it exaggerated.