What reason in current theories explains order of magnitude difference in fermion masses? If you look at the masses of leptons and quarks in the standard model, you will see that the masses are not very similar but are orders of magnitude apart.
e.g. Top quark mass is about 40 times bottom quark mass which is about 40 times strange quark mass.
Likewise the lepton masses are orders of magnitude apart in each generation.
If, for example, the known fermions were made of smaller particles you would expect to see the masses (squared) follow more of an arithmetic progression. So this probably rules out this idea.
In a theory like string theory, the masses depend on how certain extra dimensions are curled up which in turn affects how much the Higgs boson interacts with each fermion. But would we expect to see orders of magnitude mass differences in most cases?
Likewise should we expect the neutrino masses to be orders of magnitude different from each other. And WHY would we expect this? (What physical principle?)
 A: The masses come from, and are proportional to, the yukawa couplings to the Higgs field. So part of the question is, how to obtain yukawas that differ by so many orders of magnitude?
I may have overlooked something, but all the examples I can think of, fall into one of two categories. Either the yukawa is an exponential function of something else - in which case the "something else" only has to vary by a small factor, in order to produce large variation among the yukawas; or, the yukawa is obtained from Feynman diagrams in which different numbers of some vertex appear, giving rise to different powers of the associated coupling.
The examples of exponential dependence best known to me, involve localization of the SM fermions at different points in compact extra dimensions. The distance apart in the extra dimensions is the small quantity, on which the yukawa is then exponentially dependent.
As for the other option, the Wikipedia page on family symmetries gives several examples. Typically, the three generations have a different "family charge" under some new U(1) symmetry, which then determines the number of vertices in the underlying process which gives rise to the yukawa coupling. For example see figure 6, page 33, in these lectures.
Something else to remember is that the yukawas actually come in 3x3 matrices. So in order to obtain the hierarchy, one will consider different properties for these matrices - their "texture" (number and location of zero elements), their "rank" (number of linearly independent columns) - that approximate the observed spectrum of masses, and then look for symmetries etc that will give the yukawa matrices that structure.
I will add that this is my favorite paper on the mass hierarchy, but the extra structure it observes is apparently not something that any known mechanism can naturally generate.
A: Here is a long comment  on masses in the standard model from a experimental point of view:
All masses were supposed to be equal before symmetry breaking as the available energy in the expanding  cosmos was distributed to larger volumes. There were only the group structures, example ., which remained unchanged.
After symmetry breaking and the coming to existence of the Higgs field, the particles acquire mass, but the group structure is unchanged.
Thus it is the way that the Higgs field is functionally implemented that gives mathematically the experimental   different masses to different particles in the group structure, to fit the data. At present  the vacuum expectation value and the Higgs mass are fitted from the data.
It may be that  in  a different future theory  these could be theoretically predicted.
