What’s doubly disappointing is that both Vanderbei and Newman have the same reaction to being corrected: that they’re important busy people who haven’t got time to chane their color scheme for every eccentric who has a different idea. That’s a silly response because they’ve been happy to spend hours making the originals, and it’s a misguided one because the people correcting them aren’t bums off the street, they’re people who know more about this subject than the political scientists do.

Some good articles and papers on this subject are:
http://www.infovis.net/printFicha.php?rec=revista&num=192&lang=2&palabra=color
http://www.research.ibm.com/people/l/lloydt/color/color.HTM
http://www.research.ibm.com/dx/proceedings/pravda/index.htm

Per these resources, the cure for the poor discrimination of red-purple-blue schems is to abandon the use of pure hue, and instead use a much more limited hue range for labeling only, and use luminance and saturation for continuous discrimination. I made
a red-white-and-blue America map for 2004, they’re no harder than purple maps, and much more rewarding in information. And the symbolism is as good as that of the purple analogy :-)

http://flowingdata.com/2008/11/13/alternative-to-cartograms-using-transparency/

[Greetings from one of your boot camp attendees; what a jampacked and helpful three days that was.]