- Apr 17, 2011
Would doing a low amount of research (small amount of bookmarks, not even necessarily the exact same everywhere) and then simply weighted-averaging them be more professional and a good solution?Because it's not professional. As long as we have start-at-any-date as a feature we need to support all major features through-out the game. When that's not feasible - like Victoria II or HoI3 - we have to stick to a small amount of bookmarks.
For example, let's say there are two bookmarks, A and B.
At A, the composition is as follows: (100%; 0%; 0%)*
At B, the composition is as follows: (0%; 67%; 33%)*
Halfway between A and B, the composition is calculated to be the the following, and rounded as necessary: (50%; 33%; 17%)
At the third of the interval, closer to A: (67%; 22%; 11%), rounded as required.
At the third of the interval, closer to B: (33%; 45%; 22%) and approximated.
I guess it is good enough to approximate any changes occurring, is professional enough and doesn't require so much painstaking research, only a fraction.
*: assuming that these values can be represented accurately. If not, then obviously the actual values would be different, in order to be represented by the system's resolution.
Edit: I think it's sane to assume that any change between any two adjacent bookmarks was monotonic, or at least approximately monotonic enough that it didn't leave the interval between the representations of the adjacent bookmarks.
Thus the above solution to the problem couldn't cause severe reality errors, while eliminating both the problem of unprofessionality and the problem of research overload.
Furthermore, if in some regions and times the actual value does deviate enough that it does leave the interval between the two bookmarks, then the addition of another between these should not be too hard (possibly even to modders). That way the problem could be handled once more with the addition of just another single bookmark for the single exceptional province. But I assume that this would be a rare occurrence as a deviation of well over 11% is unlikely.