purple county cartograms

at last!

Tags: , , ,

7 Responses:

  1. macguyver says:

    Now can we get one with exit poll results :P

  2. en_ki says:

    America's Wang has a chubby.

  3. conradkilroy says:

    I showed my expert friend this and his response was,

    "Among us, I think their cartograms are crappy. They overgeneralize the polygons,
    i.e. take out many intermediate points along the boundaries to make the computation faster. Our group wrote a competing article today with pictures that use our method:


    If you are a blogger feel free to link to this site."

    Michael Gastner, Ph.D. Student. Physics Department, UofMich

    • jwz says:

      Well, I think the superiority of one of those cartogram projections over another can only be seen by people who write cartogram code. The differences in shape don't look very interesting to me.

      But, I think your friend's notion of what "purple" means is a bit off. It looks like they only used 4 different colors or something: a county should only be solid red or solid blue if 100% of the voters went that way, and your friend's map makes it look like most did that -- which is contradicted by the bell curve right below.

      • it's more than 4 colors. but I can't replicate their results. they color Multnomah OR as 100% blue, but the official tally is 72% blue. so I'm suspicious of the bar chart too.

      • gen_witt says:

        I think programmers/researchers (mostly researchers) have a skewed view of reality, wherein if somethings not bleeding edge it's vastly inferior. It could be elitism, or it could be that when you've worked on a problem for years you tend to notice the smallest errors and improvements.

        Case in point, when I was looking through your Lava Lite xscreensaver source, I noticed your marching cubes implementation (specifically the table) was hiddeously passe. I was going to update it and send you a patch but I don't have a table that can be GPLed. And truthfully no one is going to notice/care about the very ocasional geometric errors (there is a theoretical posibility of a hole in the surface of a blob).

    • ok, I just fetched and parsed the data myself from USA today, and I see only 3 districts/counties that are 90% for Kerry, and 6 districts/counties with no data. I'm guessing that when he fetched the data, there were hundreds of counties with no data yet, and his program marked them as 100% for Kerry.