String! Theory! What is it good for ?!?

I'll try to answer this.
Back in the 1970s people were very interested in using symmetries to figure out what theories might describe nature. Symmetries mean that some quantity in the theory is conserved, so I will talk about them that way. Symmetries come in two forms, internal and external.
Internal symmetries mean that things like electric charge, color (American spelling intentional) for quarks are conserved. Internal properties of the particle.
External symmetries mean things like momentum and energy are conserved. They relate to things external to the particle, like space and time.I should also mention that there are two kinds of particles, bosons which can exist in the same state as each other at the same time (they can pile up in one place) and fermions which will not exist in the same state as each other (they can't pile up in one place).In the 1970s physicists wondered if you could find an enormous symmetry containing both internal and external symmetries. Two physicists, Coleman and Mandula, showed that this was impossible assuming the algebra in quantum field theory obeyed certain conditions. Later people showed it didn't need to satisfy such conditions and a bigger symmetry was possible, Supersymmetry.
Supersymmetry basically says the theory is exactly the same when you turn fermions into bosons and vice versa. Basically the probability for any collection of particles to turn into any other collection of particles is exactly the same if I switched the type of particle around. Example:
Probability for two bosons and one fermion to turn into three fermions
equals
Probability for two fermions and one boson to turn into three bosons.Finally the Haag—Lopuszanski—Sohnius theorem was proved which shows that Supersymmetry is the biggest symmetry you can have if you want the theory to be a quantum field theory. Any bigger and it would have to be something else.

Although I agree that if reality is like this, then the correct theory would have to model it, I'm sure you can appreciated that there would need to be serious thinking as to how one could go about testing it.

Of course this is incorrect, if we eliminated theories by the fact that they can't yet be tested we would have lost general relativity.Although on the other hand I find the whole issue ridiculous. Fair enough if there was a big disagreement over QCD or the general framework of QFT, aspects of QM or GR. In that case there really would be a serious issue.
However I find these "battlegrounds" concerning theories which have never been tested (String, Loop Quantum Gravity(LQG), e.t.c.) to be a bit silly. Let's say somebody has strong opinions against LQG, what does it matter? If the theory is correct, they will eventually been shown to be wrong. There were people who thought GR and QFT were a load of arse and to be fair we needed those people, they're part of the scientific method.
I understand that these theories sometimes meet with unfair criticism, but it shouldn't really be a divisive issue like it has become in some people's minds. (Of course I'm not talking about you.)Also sometimes the criticism can be helpful, I know a few string theorists who were actually surprised to hear that there was no proof that String Theories caculations were finite beyond second order, they'd just assumed Mandelstam's paper demonstrated it. Now there is more interest in the work of D'Hoker and Phong, who recently showed it at second order in this series of four papers:
http://arxiv.org/abs/hep-th/0110247
http://arxiv.org/abs/hep-th/0110283
http://arxiv.org/abs/hep-th/0111016
http://arxiv.org/abs/hep-th/0111040
Introduction here: http://arxiv.org/abs/hep-th/0211111
(I'm way behind, maybe they've now extended it to higher orders, I don't know.)Of course I understand the frustration that some researches have on this being turned into a public issue by Smolin and Woit bypassing usual academic channels.Although thankfully this was a mid-2000s issue, (almost) everybody has become more sensible now.

Hey, I don't know if you've heard but N=8 Supergravity may be renormalizable. Again I know next to nothing about this stuff (which is pretty bad because I consider myself quite well read on renormalization in general, particularly rigorous results.), a link:
http://arxiv.org/abs/hep-th/0611086

Hey Straggler,
I have some idea of how to answer your question, but to be honest I never really learnt much String Theory beyond introductory books. I see cavediver is answering the question, he'd be the person to listen to as he worked professionally in String Theory.