Just for completeness, lets talk about the gerrymandering issue.
Even the best, most fair districts are going to evolve over the 10 years between censuses. The single seat districts contribute strongly to our two party system, and anyone who has seen a map of these can see how often they just don’t look right. So, lets take as a given that the current district layout system sucks (“sucks” is a legal term, BTW).
The only other options I know of are all variations on proportional representation, or instant-runoff in single seat elections. In my opinion instant runoff is a good way to do single seats (president, mayor) in a system where proportional representation fills most legislative seats, but I am not sure how it might work with all single winner districts.
So, is there a solution to gerrymandering that does not involve some form of proportional representation?
Every so often, people introduce proposals that districts must be “convex.” No bays or peninsulas, so to speak. This would remove the temptation to join two neighborhoods by a narrow isthmus – dumbbell-shaped districts (reflective of the dumbbells elected thereunto?)
It’s a brute-force solution, and would result in lots of little oddities, like having only one house in an entire housing development in District 1, while all of the other houses would be in District 2. But it might be a small price to pay.
(There’s also the problem of court-mandated districting, to help assure minority representation. This is worthy of a whole separate debate!)
Well, if by “gerrymandering” we mean “deliberately drawing the boundaries of electoral districts with a view to conferring an advantage on one party”, the obvious solution is to delegate the drawing of electoral boundaries to a non-partisan agency with no interest in securing a beneficial outcome for any party, as is done in many countries.
The popular vote percentage and the house representation percentage have favored republicans, a lot, since 2012. That is largely what brought up this concern for me.
There are about 50 different systems of doing it in the US. Many of them attempt to make it as non-partisan as possible.
Am I having a confirmation error problem here? Are there only a handful of badly gerrymandered states, and they just need to get with the program? How do the better nations/states keep it (more) fair?
This is a political problem, not a technical one. Dividing voters into districts fairly is not inherently difficult. There are all sorts of clustering algorithms out there that will impartially solve the problem without making overly large districts. More simply, you can take a north-south line, start at the western edge of the state, and sweep eastward until you reach enough voters. Repeat the process and you get vertical slices of the state, all with the same number of voters, determined perfectly impartially. If you insist on having a commission redistrict the state by hand, you can at least insist that they be reasonably compact with respect to some objective metric, e.g., diameter (i.e., the largest distance between two points in the district). As Trinopus said above, you can require that districts be convex: the straight line between any two points in the district also lies in the district.
And so on. The real issue is that we let partisan bodies determine districts, and without any clear, objective conditions on the “compactness” of those districts. It’s downright criminal for something as egregious as Maryland’s 3rd district to exist in a democracy. Honestly, though, if the criterion for fair districts is that the resulting Congressional delegate should represent the overall voting preferences in the state, then it would make more sense just to go ahead and adopt statewide proportional representation. Obviously that’s a nonstarter, but we could at least have nonpartisan commissions to do the redistricting.
Ultimately, the problem is that state legislatures mostly control the redistricting, and they use it to their advantage. That sucks, but so do state legislatures and Congress.
Indeed. We are just about to go into a major bout of this in the UK, since the government has set the Electoral Commission the task of reducing the overall size of the House of Commons and (as in all previous redistributions) at least trying to equalise the number of voters per constituency. But underlying this is the statistically obvious fact that the existing distribution has worked in favour of the Labour Party holding a larger number of seats with smaller electorates.
The Commission uses concepts like matching constituency boundaries to those of local authorities, and some sense of “natural communities” where possible - where people go to work and to shop, for example. But inevitably the Labour Party is already alleging gerrymandering, and there will be appeal after appeal, based on all sorts of technical arguments (e.g., it’s been done on an out-of-date register when more people registered this year, nobody in this ward thinks they belong to that lot in the proposed new constituency, and so on), when of course the political parties are frantically trying to work it all to their advantage.
What won’t be possible is for Parliament to sit down and re-do all the Electoral Commission’s work and re-draw boundaries as they wish.
If the state being subdivided into districts is itself non-convex, it is possible that there is no way to divide it into convex districts (for reasonable definitions: i.e. you can’t form a district from parts of disjoint peninsulas connected through sea sounds).
But this is beside the point. Crude tranching could evade the legitimate purposes of districting — a next, worse, step is using the last three digits of the SSN.
There is a simple solution that requires no constitutional amendment or such, just the will of Congress: Base seniority on how much of the vote a Representative gets. The committee member with 68% vote will get the Chairmanship, not the one who got 52%. This creates a specific disincentive to typical Gerrymandering.
Maine might actually do instant runoff for congressional elections. Question five on the Maine ballot is to use instant runoff for the Governor, State legislator, Senate, and US Congress races, starting in 2018.
Based on a quick glance at the Wikipedia articles on state congressional districts, it seems to me that most of the states that are significantly gerrymandered are in the rust belt. Ohio is 12-4 in favor of Republicans, Pennsylvania 13-4 in favor of Republicans, Michigan 9-5 in favor of Republucans, Indiana 7-2 in favor of Republicans, and Wisconsin 5-3 in favor of Republicans. Other than Wisconsin all those states look like they have very oddly shaped districts. To me it’s obvious that these states, other than maybe Wisconsin, have significantly gerrymandered districts. I don’t, however, have any idea what the solution to this might be.
If we increased the number of representatives to one for every 30,000 Americans (the legal minimum number of people a representative can represent), we would increase the number of house members by 2000% but gerry mandering would almost become a non-issue. WTF would be the point in gerrymandering a district of 30,000 people?
You can decompose any connected state into convex regions; it follows from the fact that any polygon in the plane can be triangulated. For an arbitrary state, just divide the connected components into compact districts. (This doesn’t quite work for states like Hawaii with many connected components and few districts, but that’s an easily-handled edge case.)
As UDS and Patrick London point out, in most countries the electoral boundaries are drawn by non-partisan commissions. That’s how it works in Canada, at both the province and federal levels. Gerrymandering is rarely an issue here.
We should just have proportional representation. X% of a state voted D in the last presidential election, Y% voted R, therefore the seats are split that way, with the national party appointing members to serve in those seats.
This would allow smaller groups to have representation, because even the Greens or Libertarians can muster up a couple of percentages even if their candidates don’t win the presidential. 5% of Congress is libertarian? 5% Green? Sure, I’m ok with that.
There is no such thing as non-partisanship in the US. Such a committee might believe itself to be non-partisan, but at least one party will insist it is not, and it probably isn’t. We don’t believe in draws, only winners and losers. And when we lose we insist it’s because the other side had an unfair advantage.
As I said, it becomes a matter of definitions. At one extreme … how do you divide a non-convex state into* one convex district*?
A more interesting example, again as stated in my post, would involve interleaved peninsulas (consider the west coast of Scotland) where your “convexity” is achieved by connecting across sea sounds, contrary to a natural sense of connectivity or compactness.
In my whimsical quibbling about “convexity” in the face of natural barriers, my more substantial point got overlooked:
Gerrymandering gives the opposition a few seats they win with high margins. If a system is in place — as would make sense; it’s logical to give more power to high-vote Representatives — to reward these high-vote Representatives, gerrymandering would be much less of a problem.
(But the quibbling about convexity was not wholly whimsical. Computer-drawn districts might combine parts of Marin and San Francisco counties, or connect towns on opposite sides of the Mojave district. At some point you lose the very reason “compactness” makes sense in the first place.)
You certainly can’t divide an arbitrary connected state into a bounded number of convex districts; take n line segments meeting at a single point, for example.
I’m not quite sure what the grouping criterion is supposed to be for districts. If it’s having similar political opinions, then we might as well group a hypothetical 60% Republican and 40% Democratic state into three 100% Republican districts and two 100% Democratic districts. If it’s having similar political interests, then geography isn’t a good proxy for that; it’s hard to claim that a single 20-year-old working a minimum wage job and a married 50-year-old business executive with kids in college have similar interests just because they live in the same town.
Come to think of it, I actually like the idea above about grouping people by their social security numbers (or some similar arbitrary criterion). It’s as defensible a grouping as any other, and it would make gerrymandering impossible.
If the group has nothing in common, though, where is the value in grouping them at all? Once you get to that point any objection to multi-member districts or even one huge multi-member district basically evaporates.
