This series of posts was very clear, comprehensive, and informative - I suppose a general question of mine is why isn't this taught more widely? The color exercises we went through in foundation year certainly didn't address the important issues of "lumpy" color perception, the difficulty of equalization when you can't have a dark yellow, and other related concepts.
Interestingly, it seems that many of the palettes that follow the perceptual and data-representation rules Simmon sets out also happen to be aesthetically pleasing. Could it be that some of the things that translate to "pleasing" in our minds share principles with hat which is legible?
How about discrete datasets with only two or three categories? I feel like it's very difficult not to indicate value judgments under these circumstances, and using similar saturations often leads to a clashing map.