Gerrymandering and a cure for it – the shortest splitline algorithm • Hacker News Board

Re: Gerrymandering and a cure for it – the shortest splitline algorithm

#11

Post by specialist » Wed, Sep 30, 2015, 1:36 AM UTC

tunesmith wrote: post #2

Isn't gerrymandering a much more complex problem than that makes it out to be? I've always thought the problem was that it's actually difficult to even come up with a definition of what fair is. For instance, do you want each district to be as representative of the population as possible? Or do you want the house of representatives to be as representative of the population as possible? I suspect that you can't have b…

Gerrymandering is simply incumbency protection.

Fairness is simply maximal competitiveness.

Re: Gerrymandering and a cure for it – the shortest splitline algorithm

#12

Post by powera » Wed, Sep 30, 2015, 1:36 AM UTC

specialist wrote: post #10

dragonwriter wrote: post #3

Earlier quoted context omitted.

> Isn't gerrymandering a much more complex problem than that makes it out to be? No, its vastly simpler: the problem with gerrymandering is that there is no good way to draw single-member winner-take-all districts, and the solution is just don't do that . Multimember districts with a system that provides proportional representation within each district (e.g., five-member districts with STV) solve the problem pretty m…

Agreed. But even when the USA sees the light on proportional representation, there will continue to be winner takes all elections. Executive positions (president, governor, mayor), senators, judges, ballot issues, etc. In those cases, approval voting is the correct answer. Easy to explain, easy to tabulate, most mathematically fair, hardest to game. Per Duvenger's Law, approval voting would probably also help replace…

On the other hand, you can't gerrymander Senate seats. (it's not representative now, but it's not changing either)

Re: Gerrymandering and a cure for it – the shortest splitline algorithm

#13

Post by spoiledtechie » Wed, Sep 30, 2015, 1:41 AM UTC

a8da6b0c91d wrote: post #4

Doesn't gerrymandering mostly carve out a lot of districts for black congressmen who otherwise wouldn't have districts? I think that's kinda what it boils down to in America. There are conservative areas adjacent to and mixed with majority black areas and they want different representation.

No one ever thinks of it that way.

Re: Gerrymandering and a cure for it – the shortest splitline algorithm

#14

Post by aaronbrethorst » Wed, Sep 30, 2015, 1:41 AM UTC

a8da6b0c91d wrote: post #4

Doesn't gerrymandering mostly carve out a lot of districts for black congressmen who otherwise wouldn't have districts? I think that's kinda what it boils down to in America. There are conservative areas adjacent to and mixed with majority black areas and they want different representation.

Um, no. Here's one example: http://talkingpointsmemo.com/muckraker/janet-adkins-corinne-...

Re: Gerrymandering and a cure for it – the shortest splitline algorithm

#15

Post by specialist » Wed, Sep 30, 2015, 1:42 AM UTC

powera wrote: post #12

specialist wrote: post #10

Earlier quoted context omitted.

Agreed. But even when the USA sees the light on proportional representation, there will continue to be winner takes all elections. Executive positions (president, governor, mayor), senators, judges, ballot issues, etc. In those cases, approval voting is the correct answer. Easy to explain, easy to tabulate, most mathematically fair, hardest to game. Per Duvenger's Law, approval voting would probably also help replace…

On the other hand, you can't gerrymander Senate seats. (it's not representative now, but it's not changing either)

I'm totally on board with starting over with new state boundaries.

Re: Gerrymandering and a cure for it – the shortest splitline algorithm

#16

Post by kevinoid » Wed, Sep 30, 2015, 1:50 AM UTC

This is an elegant solution with a lot of good properties: Equal population balancing, politically unbiased, reasonably understandable by the layman, typically geographically compact districts. But where's the discussion of the desirable and undesirable properties in a districting algorithm and how this algorithm achieves or doesn't achieve these properties?

specialist already mentioned maximal competetiveness, ROFISH mentioned clustering due to natural features. Is regional or ethnic representation important, or is it an undesirable bias? How about travel distance (either for voters or campaigners)? I'm sure you guys can think of others you'd like to see or to avoid. Feel free to suggest some, I'm curious.

There is a little more information at http://rangevoting.org/SplitLR.html but did I miss the real weighing of the tradeoffs and advocacy for this particular method above all others?

Edit: Found a discussion of the theoretical issues at http://rangevoting.org/TheorDistrict.html which touches on some, but not all of the points raised.

Re: Gerrymandering and a cure for it – the shortest splitline algorithm

#17

Post by afarrell » Wed, Sep 30, 2015, 1:59 AM UTC

specialist wrote: post #15

powera wrote: post #12

Earlier quoted context omitted.

On the other hand, you can't gerrymander Senate seats. (it's not representative now, but it's not changing either)

I'm totally on board with starting over with new state boundaries.

I'm actually really intensely curious how one would design a voting system for people to choose what subentities to belong to. How would you populate the ballot? Would you run a clustering algorithm on the votes?

What if New Jersey and Michigan just decide they want to be as they are, but excluding Camden and Detroit respectively?

Or would you do something whimsical like draw lines based on football/baseball teams and cleave states like CT in half?

Re: Gerrymandering and a cure for it – the shortest splitline algorithm

#18

Post by ashearer » Wed, Sep 30, 2015, 2:03 AM UTC

tunesmith wrote: post #2

Isn't gerrymandering a much more complex problem than that makes it out to be? I've always thought the problem was that it's actually difficult to even come up with a definition of what fair is. For instance, do you want each district to be as representative of the population as possible? Or do you want the house of representatives to be as representative of the population as possible? I suspect that you can't have b…

This is very true--trying to create a single outcomes-based ideal of what's fair is both difficult and subject to the same strategizing that caused gerrymandering in the first place. Instead of choosing what "fair" is and maximizing it, the proposal takes a different approach: by using an extremely simple geometric rule, the algorithm seeks to avoid choices that would be susceptible to influence for political gain. Those are the choices that tend to result in gerrymandering, which by locking down power is expressly designed to be the opposite of fair to future voters. So instead of the difficult task of directly maximizing fairness, the proposal aims to minimize a great source of unfairness. I'd be curious to know how well it would work in practice.

Re: Gerrymandering and a cure for it – the shortest splitline algorithm

#19

Post by tunesmith » Wed, Sep 30, 2015, 2:10 AM UTC

specialist wrote: post #11

tunesmith wrote: post #2

Isn't gerrymandering a much more complex problem than that makes it out to be? I've always thought the problem was that it's actually difficult to even come up with a definition of what fair is. For instance, do you want each district to be as representative of the population as possible? Or do you want the house of representatives to be as representative of the population as possible? I suspect that you can't have b…

Gerrymandering is simply incumbency protection. Fairness is simply maximal competitiveness.

Define "maximal competitiveness" in a 60-40 state with ten districts. Ten 60-40 districts? Eight 51-49 districts, and two 96-4 districts? Either could be argued as incumbency protection.

Re: Gerrymandering and a cure for it – the shortest splitline algorithm

#20

Post by cjensen » Wed, Sep 30, 2015, 2:18 AM UTC

ROFISH wrote: post #7

One problem with this: it makes sense that district lines could be along natural (rivers, lakes, mountains) or political (towns, county) lines. Trusting a simple line algorithm may bisect a town down the middle.

Yep. Consider Eastern California: the Sierra Nevada runs north to south creating a natural barrier. There's a small sequence of cities east of the barrier. During the Spring and Fall campaign season, it can be literally impossible to cross that barrier: you must go a long way around or fly.

The suggested algorithm would unite cities on both sides of the divide. It would be very difficult for an underfunded candidate to campaign on both sides.