How could gerrymandered districts be un-gerrymandered?

What’s the simplest and most achievable way to get rid of those snakelike monstrosity districts for good?

California’s new system of having a citizen commission draw the boundaries instead of the politicians dong it seems to have worked reasonably well.

Make people move. A lot of those districts were court ordered to achieve minority participation in the electoral process.

I believe there are actually some computer algorithms that have been developed that will analyze districts and determine whether they have been gerrymandered. I think the same algorithms can propose fairer-solutions.

Also, I believe that some states leave redistricting up to non-partisan boards. That’s another solution.

I agree that a computer method is probably the best solution. I would suggest agreeing on an algorithm then locking it to kick in 10 years so that fighting over the algorithm itself is moot.

Impose a basic mechanical constraint, such as “A district must contain at least X% of the inhabited area within some circle that surrounds the entire district.” That reins in abuse without a futile attempt to take the politics out of politics. The only difficulties are in setting an appropriate value of “X” and a definition of the minimum population density that defines an “inhabited area” (a necessary caveat to account for regions containing large stretches of water or desert).

If that’s the main goal, we could switch to a proportional representation system.

And that’s the problem right there. If it doesn’t look like a salamander, it looks like a politician’s dong.

One (small) part of the problem for Democrats is that under the Voting Rights Act (or some other civil rights law or principle) there needs to be a certain level of “minority” districts, so as to allow enough minorities to be elected. This generally has the impact of helping Republicans, because these minority districts generally concentrate a very high percentage of Democratic voters into fewer districts.

I imagine this would also complicate any computer algorithms.

As a practical matter, there’s not really any such thing as “non-partisan”.

I don’t think that’s exactly right, but it’s perhaps close enough for this discussion.

I believe the explicit language of Section 2 of the VRA is that in a majority-minority district redistricting can not be done in such a way as to dilute the voting power of the minority in question.

Basically, that you can’t have a white-majority state splitting the minority vote amongst several districts in order to make sure no minority is ever elected.

There has been at least one case that struck down a gerrymandered district that kept too many minorities together in order to insure a minority representative (Georgia, maybe?).

I do agree with the general point that concepts of racial “fairness” and the history of suppression of minority voters throws a wrench in the idea of simple, compact districting.

I remember fights over this someplace, although I don’t remember how it came out. As I recall it, it was an oddball coalition of Republicans and African Americans against white Democrats.

In the United States, perhaps, but other western countries manage to have non-partisan drawing of electorate boundaries – for example.

They say Abraham Lincoln once asked a guy “how many legs would a donkey have if you called its tail a leg?”, and the guy responded “five”. And AL said “no, it still has 4. Just because you call its tail a leg doesn’t mean that it is one.”

As a practical matter, there’s not really any such thing as “non-partisan”.

My favorite solution is this:

The majority party draws districts however they want, as long as population is distributed relatively equally (say, no more than a 5% variation in population between the most and least populous districts). Add up the length of all the borders in all the districts.

The minority party then has two choices. Either they accept the district map as it is, or they submit an alternative map in which the total length of borders is at least 10% (or other significant number) less than the majority’s proposal. If they can submit such an alternative map, then the minority party’s map is the one that applies; otherwise, the majority map applies.

Such a procedure not only allows for partisanship, it leverages partisanship to try to get the least gerrymandered maps possible. If the majority party does too much gerrymandering, they’ll not get to draw the final map; and if they go ridiculously overboard with gerrymandering, then they let the minority party submit a gerrymandered map of their own.

Two modifications that might be worth discussing:

  1. The majority party could choose whether to go first or second in the process. They really should have an advantage, after all, having received the most votes, and if it turns out that going second is advantageous, let them have it.
  2. The process doesn’t end at the second iteration. Instead, after the second map has been proposed, the other party gets another chance to beat it with a map of their own that improves by 10% on the last one, and so on, until both sides agree on a map.

Interesting idea overall, but I would quibble with this a bit. Sometimes the majority party actually got fewer votes, but is the majority party due to prior gerrymandering. And even if they got the most votes, this is sometimes also due to gerrymandering. Because the party controlling the redistricting process generally tends to push as many of their opponent’s voters into as few districts as possible, thus creating a few districts with 90% or so for the others and many districts with 60% for them. As a result, a high percentage of the opposing party districts are not remotely competitive, which greatly depresses voter turnout.

[It’s not strictly true, and it’s been said that the overriding goal in redistricting tends to be preserving as many seats for incumbents of both parties as possible. But it’s also a big part of it.]

My proposal is like Dorkness’s, but even simpler: Every member of the legislature can submit a map, and whichever one has the shortest total boundary length, wins.

If you want, you can encourage districts that correspond to the current geography, by (say) making boundary lines along navigable bodies of water free, and boundary lines that follow county lines only half cost.

I think Nate Silver should develop a model that splits states mathematically, objectively and fairly.

I’m sure I’m not the first Doper to notice this. I’m just surprised I’m the first Doper classless enough to point it out. Are we all still hung over?

These proposals address tadpole shaped districts, and similar. You can also split up an urban area with regular shaped districts, and to control against that requires rules based on more than geometry.

My favorite algorithm works like this: you minimize the sum of the squares of the distances between each person in a state, and the center of population of their proposed Congressional district, subject to the usual stipulation that the same number of people reside in each district.

The advantage of this over the minimizing-boundaries proposals is that this gets you the most compact districts in terms of people, rather than land.

But I’m for practically any system that doesn’t have control of the House of Representatives for a given decade being dictated by which party was in control of a bunch of statehouses in years ending in 1.