powdr, my "opinion" is just reciting some mathematical relationships others have put forth. I never said you couldn't use the same powder to good effect!
Frankly, there is no such thing as a "3% slower powder." Such a statement means only that the optimum powder for the larger case has to be somewhat slower in gas generation ("burn rate") if you plan to fully fill the case. Filling the case is usually the route to maximum velocity, for it maximizes the chemical energy available to each grain of bullet mass. Whether you fully utilize that potential chemical energy depends on the burning characteristics of the powder, and burning characteristics simply cannot be reduced to a single number.
Keep in mind that 3% is a small change, less than the difference between 4831 and 4350. One might well be running the parent case nicely with a powder that's a wee bit slow for it; using it again in the blown out case might leave you with a powder a smidge too fast. Neither would be optimum, but neither would be bad. Further, it's quite possible there is no other powder that would do better in either case. Trying to distinguish such a fine difference without good quality pressure measurement tools is nearly impossible.
As for your observation, how exactly have you proven to yourself that you have the optimum powder for your blown out cases, or for the parent case, or that you've been operating each at the same peak pressure? As you point out, there is a mind numbing number of powders to try out. When it comes to optimum loads, "knowledge" is very difficult to establish.
Regardless, to get to tcp's specific question, you can bump the charge roughly 6%--but there's no guarantee. The various ways to estimate this are not in agreement. There is no "rule" for this.
