How about because we don't have any knowledge of it?
If the constants have to be the way they are (in other words, the probability of them being as they are is high), we don't know why that is. If they can vary widely, then we don't know why ares are the values they are. IN either case, there's a problem to be explained: some underlying theory that can tell us why the contstants take on the values they do...or what possible values they could take.
No one is saying that "we have to explain it because it's improbable". They're saying "we have to explain it". Since we have no info on anything that would contraint the values of these constants, the assumption (and thus most of the speculation) is that they are uncontrained. On the other hand, a lot of the work toawards G.U.T.s is in the hope that one will provide the constraints for the constant values, and the values for these constants will "fall out" of the equations (shwoing them to be dependent on some other, simpler value...such as one or two constants determing the value of all others).