Originally Posted by
PraisetheSun
I've actually been wondering (primarily directed at Theck) whether, since we cannot realistically gear around discrete breakpoints any more without likely over- or undershooting the mark to quite some degree, some measure of total contribution towards TMI reduction would be a useful metric for comparative analysis?
Assume we have a number of stat-dependent TMI functions - y(mastery), y(haste), y(multistrike), etc - and that some of these functions behave in a more or less linear matter while others (say, haste), behave erratically and have both major peaks and valleys. Would comparative analysis of the integrals of these functions over a given stat range not give us a rough estimate of what "stacking" a certain stat does for TMI compared to another, or am I completely mistaken?