Visualizing global democracy
I saw George Monbiot speak at Duke University a couple years back, and I was impressed by the passion of his talk. I even bought a copy of his book, Manifesto for a New World Order, which I almost never do after such seminars. I recently reread the book, to sort through what in it I agree with and what in it I do not. I take issue with some of his specific solutions. I’m also more of a moderate politically than Monbiot, so it makes for awkward reading in spots. Nevertheless, the book is very important just for pointing out how manifestly undemocratic the current global system of governance is. Furthermore, Monbiot vigorously defends the need to strive for something more democratic than the current global system of governance: “If you consider this distribution of power acceptable, that is your choice, but please do not call yourself a democrat. If you consider yourself a democrat, you must surely acknowledge the need for radical change.”
The proposal in the book that most grabbed my imagination was for a global parliament, as a compliment to or replacement for the United Nations and its one-nation-one-vote paradigm. Monbiot points out that this proposal is repulsive to many in the developed world, simply because of the overwhelming dominance in the system of the developing countries, due to their large populations. As a landscape ecologist and a geographer, I was interested in how this would look on a map. Below is a map of 400 voting blocks of approximately equal population, based on the excellent global grid of population in 1995 available from the Columbia Earth Institute. First of all, let me say what this map is NOT: It is not an attempt to put forth a reasonable set of voting districts (which would be rather arrogant of me, as such things are always the outcome of a political process), nor is it an endorsement necessarily of Monbiot’s scheme for a world parliament. The idea is to get people thinking about how political power would be distributed if every person on earth had equal voting power. While national boundaries are shown on the map to help orient the reader, they were not used at all in the creation of these voting districts, and so the districts freely span national boundaries when population densities require that. Note the high density of small regions in southeastern Asia, particularly India and China- this is simply due to the high population density in these places.
Click on the thumbnail to the left to view a global map and a close-up of southeast Asia. Each color is a region with around 15 million people.
Now, technical details on how this was made: Population data were taken from the Center for International Earth Science Information Network (CIESIN), Columbia University; International Food Policy Research Institute (IFPRI); and World Resources Institute (WRI). 2000. Gridded Population of the World (GPW), Version 2. Palisades, NY: CIESIN, Columbia University. Available at http://sedac.ciesin.columbia.edu/plue/gpw. This gives population estimates in latitude/longitude for cells that are about 5km on a side near the equator. To perform my calculations in a projection system that is more equal-area than geographic, the grid was projected into a Robinson projection, which is reasonably equal-area between 45 and -45 latitude, and doesn’t have the high distortion of shape and distance at high latitudes that a true equal-are projection (e.g., sinusoidal projection) would. The current algorithm for dividing the world’s population into geographic blocks is very crude (this is why it’s so boxy), and is literally just what I could do last night when I had a couple hours of spare time.
1. Global population, excluding any already assigned persons, is summed in an X by X resolution grid for the whole globe, starting with small values of X like 50km.
2. If any cells in this grid are within the target population range (10 million to 20 million), they are saved as one voting district.
3. The global population grid is recalculated to exclude folks assigned in #2.
4. Increase the size of X slightly and repeat steps 1-4 until all areas of the globe are classified.
Now, there is an algorithm that’s quite similar to the algorithm used for spatially-constrained clustering that could be used with this data, where individual cells are agglomerated until the target size is reached. This algorithm would give much smoother looking voting districts, but I think I’d have to spend a couple days writing custom code to do it (there are, after all, some 24 million cells in the population grid), and I just don’t have the time!
