Three emails:

Jonah,
Yes, dumber than rocks.
You need to calculate a death RATE by COUNTY, so that you compare
county-based map to county-based map, and so that you show the
contributions of each county based on their ability to give. You
could color code this map like a weather map showing varying degrees
of storm intensity. I think even the existing icasualty map shows
rural counties have given proportionally more than counties with
major cities. I think you’d find rural counties would be the “hot”
counties on my proposed county-based map.
The data are available for some enterprising blogger.
###
Jonah:
Debunking the map is simply a case of apple and oranges. The Blue vs Red State map is comparing percentages of the voting population whereas the death map is looking at discreet points, i.e. individual deaths. The clustering of the points gives a similar appearance but does not display the same trends or patterns.
Byron assumption was right that a portrayal of Iraqi war deaths as a percentage of each state’s military personnel would most likely reveal a more uniform pattern with no correlation whatsoever to voting patterns. If I was to map out homicides, I am pretty sure I would come up with a map that disproportionally clusters such deaths in the areas that voted heavily for Kerry yet I would not be correct in assuming that Kerry voters are more likely to commit murder.
###
As a visual conveyer of information, the death map is horrendous.
For starters, the dots are disproportionately sized. To be more accurate,
the dots would either (a) have to be sized in inverse proportion of the
density of the state in which it would be placed. In other words, dots in
Montana should be bigger than dots in Massachusetts (but then this would
convey other wrong information). Using dots is a terrible way to convey
*any* information other than which regions had deaths (not how many).
Secondly — and this is related to the problem of the dots’ size — is
that overlapping dots appear to be more dots than thay actually are, which
leads to the seemingly overwhelming density of deaths in high-density
population centers. The dots in rural areas don’t overlap as much, and so
appear to be fewer in number.
Thirdly, as was already pointed out, the dots on the map mean stasisticly
next to nothing unless you know how each dot voted (which is now
impossible).
The red-blue map has its own problems, but on the whole is much more
accurate relative to the information intended to be understood from it.