I wonder if this algorithm would still present biases given the fact that large cities tend to vote differently from less densely populated areas. My hypothesis is that this algorithm would favor city voters over rural voters, for the following reasons:
1. Successfully gerrymandering in favor of one party generally involves stuffing the other party's voters into one district, while splitting your party's voters into multiple districts.
2. The proposed algorithm favors straight-line divisions through regions such that about half the population of the region is on each side of the division.
3. Thus, chances are, the straight lines will run through the middles of cities, cutting cities into multiple districts.
I'd be interested to see someone test this out--it's likely that my hypothesis is wrong, but it also seems pretty likely to me that there will be at least some urban/rural bias, one way or another.
1. Successfully gerrymandering in favor of one party generally involves stuffing the other party's voters into one district, while splitting your party's voters into multiple districts.
2. The proposed algorithm favors straight-line divisions through regions such that about half the population of the region is on each side of the division.
3. Thus, chances are, the straight lines will run through the middles of cities, cutting cities into multiple districts.
I'd be interested to see someone test this out--it's likely that my hypothesis is wrong, but it also seems pretty likely to me that there will be at least some urban/rural bias, one way or another.