Not all centipawns are created equal

I would like to hear your thoughts on a thing that has been confusing me about centipawns. It seems to me that there are games where the centipawn count is less than usual. This happens in games where there are maybe half a dozen or more moves that all have almost the same evaluation e.g. +.50 +.49 +.49 +.48 +.48 +.47. In such a game there are many moves that have evaluations that are close to the best value and so choosing anyone of them results in a very small centipawn loss. Games with this characteristic, usually very balanced positions where there are few tactic e.g. a closed Stonewall setup, result in much smaller centipawn losses than more unbalanced dynamic positions which could/should have larger centipawn losses. So maybe for the centipawn loss to be a more accurate measure of performance in a game it should be modified with a measure of the variance of evaluations at each move. OK! This is not really an important topic and I am offering NO solution to a problem most would not even consider a problem but for me it is interesting and I have not seen any discussion on it. Any thoughts?