3 ms·

> There's no natural or fair way to draw districts, it doesn't work that way. All maps may be biased, but not all maps are equally biased, so there's still roo

> There's no natural or fair way to draw districts, it doesn't work that way. All maps may be biased, but not all maps are equally biased, so there's still room for refinement. You could define "nothing up my sleeve" measures, similar to the way some cryptographers pick arbitrary constants in ways that are possibly biased, but unlikely to be selected in advance according to that bias. https://www.schneier.com/blog/archives/2006/10/a_million_rando.html https://www.schneier.com/blog/archives/2006/10/a_million_ran... So for example, there are some systems that automatically divide states into districts, though a looser version would be to accept proposals and choose the one that most effectively satisfies certain reasonable constraints. https://handsoffredistricting.net/opra-explained/ https://handsoffredistricting.net/opra-explained/ http://bdistricting.com/about.html#other http://bdistricting.com/about.html#other It would require a thorough study to determine if these methods biased political power towards or away from population centers. Many political issues have an urban rural split, so that's not a trivial issue. They would be at least an incremental improvement over intentionally gerrymandered districts though, where the probability of bias is approximately 1.

samatman 10y ago
That study has been done, by Chris Fedor. Salient quote: “In both the 2008 and 2010 elections, the actual districts used by the 111th Congress […] were in fact about as fair as you could reasonably expect from a voting system based on contiguous geographies.” https://priceonomics.com/algorithm-the-unfairness-of-gerrymandering/ https://priceonomics.com/algorithm-the-unfairness-of-gerryma...
- brownbat 10y ago
  Interesting. Mattingly and Vaughan found slightly different results, even looking at the same state: https://politicalwire.com/2014/11/29/study-shows-how-gerrymandering-distorts-election-results/ https://politicalwire.com/2014/11/29/study-shows-how-gerryma... Seems like a contentious issue in political science.
- brownbat 10y ago
  After a closer read of Fedor's research, what it reveals is that methods that look purely at geography and population density aren't sufficient, because of urban / rural clustering patterns. Gerrymandering relies on a guideline to conserve resources in game theory: win by a little and lose by a lot. That is, pick the districts you lose and pack opponents into those districts. Edge out competition in others. If you incorporated exit polling data and revised your constraints to draw districts to create as many tossup districts as possible based on recent elections, then you should find a district map that is the most weakly gerrymandered. It will still have some aesthetically unappealing shapes, some salamander districts. But while the shape of districts can illustrate gerrymandering, I don't think it's the real concern. Fedor judged systems on how likely it was for the representation to match the statewide split in the voters. If that's the real standard of fairness, the easiest solution would probably just be proportional representation. Thanks for flagging the article, interesting stuff.

samatman 10y ago

That study has been done, by Chris Fedor. Salient quote: “In both the 2008 and 2010 elections, the actual districts used by the 111th Congress […] were in fact about as fair as you could reasonably expect from a voting system based on contiguous geographies.” https://priceonomics.com/algorithm-the-unfairness-of-gerrymandering/ https://priceonomics.com/algorithm-the-unfairness-of-gerryma...

brownbat 10y ago
Interesting. Mattingly and Vaughan found slightly different results, even looking at the same state: https://politicalwire.com/2014/11/29/study-shows-how-gerrymandering-distorts-election-results/ https://politicalwire.com/2014/11/29/study-shows-how-gerryma... Seems like a contentious issue in political science.
brownbat 10y ago
After a closer read of Fedor's research, what it reveals is that methods that look purely at geography and population density aren't sufficient, because of urban / rural clustering patterns. Gerrymandering relies on a guideline to conserve resources in game theory: win by a little and lose by a lot. That is, pick the districts you lose and pack opponents into those districts. Edge out competition in others. If you incorporated exit polling data and revised your constraints to draw districts to create as many tossup districts as possible based on recent elections, then you should find a district map that is the most weakly gerrymandered. It will still have some aesthetically unappealing shapes, some salamander districts. But while the shape of districts can illustrate gerrymandering, I don't think it's the real concern. Fedor judged systems on how likely it was for the representation to match the statewide split in the voters. If that's the real standard of fairness, the easiest solution would probably just be proportional representation. Thanks for flagging the article, interesting stuff.