The most significant factor in this issue is scale. The simple term that is used in the engineering work is called k/C, or a non-dimensional roughness parameter. "K" is the measured height, perpendicular to the airfoil surface, of the ice shape. "C" is the chord length. This term is has a major role in the aerodynamics of the contaminated wing.
Any given icing environment will yield a range of ice shape dimensions and features. However, it is more or less intuitive that a C-5 passing through Cloud X will encounter the same droplet size, liquid water content and droplet distribution that a Citation will encounter in Cloud X. The C-5 will accrete quite a bit more mass, due to the area swept by the wing, but the nature of the ice shape is unlikely to be greater in measured height than the shape found on the Citation.
Therefore, the "k" term will be somewhat comparable, for the sake of the argument at least. There are other factors at play, but they can be set aside for the moment. If the "k" term is the same, but the "C' term is so radically different...well, you get the idea. The k/C ration for the Citation is much, much larger than the k/C for the C-5.
The C-5 will also push a lot more droplets just plain out of the way, due to the wing size and the pressure wave ahead of it. Smaller airfoils arrive at the cloud with less "warning" to the droplets, and more are accreted. This is why the tail often ices when the wing doesn't.
Thus, no C-5's or 747's dropping out of the sky. Those of us in this part of the business have contemplated for years the possibility of an engineering standard reflecting scale, but we just don't know enough to really define one. Instead, the manufacturer works this out on a case-by-case basis. Airbus knew they wouldn't need much protection, so they designed accordingly and were easily able to certificate that design.
Unfortunately, many, many pilots live under the impression that their airplane can "handle" a lot of ice. As I have said in other threads, this is truly a myth. Trunov and others showed as far back as the seventies that only a few thousandths of an inch of roughness was required for substantial degradations. Douglas pointed this out in their many ground deicing cases as well...look up the work done by Ralph Brumby. In the icing accident database that I maintain for the FAA, the average ice accretion either found afterward or reported by the pilot is between one quarter and one half inch. Often, one eighth is sufficient...particularly for the Citation, by the way.
The problem is, you can't identify the critical parameters, such as horn angle, horn height, roughness, etc. from the cockpit. And you have no idea how close to the modified Cl max you are at any point. The only real solution is to get rid of the ice...which is where good, hot, thermal systems work best.