Severian said:
It is a bit stronger than that. It is the only way to enlarge the symmetry group of space-time. It is the natural extension of the Poincare symmetry, and there is definitely a prejudice that the laws of physics come from symmetries. So having space-time as symmetric as possible is definitely attractive. Even if you don't believe in low energy supersymmtry, it is very hard to live without high energy supersymmetry.
This is the kind of argumentation to which I'm rather insensitive. There are many possible symmetries in nature which simply do not turn out to be there, and there's no "principle of maximum symmetry" as far as I know. If it were the case, we wouldn't have CP violation, or even parity violation ; SU(5) would obviously be a better gauge theory than U(1) x SU(2) x SU(3)...
There are so many "missed occasions" in nature to have a certain symmetry that I don't think that it is justified, just because it is not impossible, to introduce a symmetry, just because it is a possibility. It's not excluded either of course and we should entertain the possibility of its existence.
I'm also not very sensitive to all those hierarchy and fine tuning "problems". To me, there is no qualitative difference between the real numbers between 1+10^(-20) and 1+10^(-15) on one hand, and 1.5 - 50.0 on the other hand. If it is a free parameter, it is a free parameter and it can just as well be part of the former interval as the latter. With a suitable mapping, we can get the former interval on the latter. It is IMO just psychological that we find the former "hard to believe" and the latter "normal free parameters".
That is not a fair criticism. Breaking supersymmetry is actually rather easy, so it is not an unreasonable thing to expect. The reason the models look a mess is simply because we don't know the mechanism of the breaking, so just parameterise it in a general way. Once the mechanism is known, the theory becomes quite predictive, with very few parameters.
Well, the day that there is a serious mechanism proposed, just a few parameters, and hard predictions, I might change my mind

. But for the moment, the non-existence of a precise proposition of its mechanism of breaking allows one to introduce so many "free fit parameters" that one can morph it onto any set of experimental data.
Personally, I have the impression that the main success of supersymmetry is that it allows for easier computations in quantum models, because of the many cancellations that occur. Just make your theory supersymmetric, and you have better chances to have it computable. Whether that is a strong argument for a physical property, I don't know. Classically, integrable systems are also easier to consider. But most classical systems just aren't integrable.
Now, I'm taking some serious risks here, because I might have to eat my hat in a few years, when the LHC spits out superpartners with dozens. But for the moment, I don't bet on it.
EDIT: I must maybe soften my propositions a bit. It is probably because of ignorance that I don't see the compelling reason to almost assume supersymmetry as established, and just waiting for a kind of formal validation by experiment about which one hasn't much doubt. I simply haven't seen this reason yet, and maybe if I were to delve more deeply in its formalism, I might be convinced.