Well, as long as you’re satisfied to leave your assertions completely unsupported and end on a wrong note, then I’m good too.
Of course, if it were me, I wouldn’t assert things I couldn’t back up to begin with, I’d try not to miss the point and then condescend when challenged, and I’d strive to clearly label my gut feelings and assumptions as gut feelings and assumptions. But like I said, this is a pattern with you.
To be nitpickingly correct, Fotheringay-Phipps has not provided a mathematical proof that winner-takes-all voting produces skewed results. He has, however, provided multiple examples of it under plausible conditions, while nobody has offered counterexamples. Further, he does not appear to be arguing that gerrymandering is not a problem, simply that it’s not the only problem. Why is this contentious?
I have a counterexample: the 2012 election.
But one example proves very little, and he’s admitted that that SPECIFIC example could be the result of gerrymandering.
His central point is that in a FPTP system, the pattern will usually be that the party winning the most votes will usually - not always, but usually - win a number of seats greater than their percentage of the vote without gerrymandering. That is in fact absolutely correct, and strikes me as being common sense. But I’m a baseball fan, and this phenomenon, which we call the “Pythagorean projection,” is just kind of par for the course for us. If you score 55% of the runs, you won’t win 55% of the games. You’ll win about 59% of them. Score two thirds of the runs, you’ll win better than three quarters of your games. Binary electoral politics should logically have a similar relationship - again, without gerrymandering - though the precise exponents you’d have to raise your variables to I do not know.
If you assume no gerrymandering and then run 1000 simulations of a 435-seat election, I assure you the vast majority of the time the party that wins most of the vote will win a greater percentage of seats than their percentage of vote. You will also get, by random chance, situations like Ohio where a party with a small percentage advantage in popular vote lucks out and scores big in seats. (I am not saying Ohio in particular was luck.) The more parties you introduce, the greater the disparities will be. This is a ridiculously obvious phenomenon if you watch enough elections. Heck, it’s equally true of PRESIDENTIAL elections. While you get the odd outlier like 2000, usually the EV split is extreme compared to the popular vote split.
I’m not disputing that there can be variations from the expected results. The basic pythagorean expectation (and I’m not claiming I’ve made any vigorous calculations, just the quick and dirty) suggests the Democrats should have ended up closer to 220 seats, rather than the 197 they ended up with. But this is the first election since the redistricting of 2010, which Republicans controlled for the most part, and this is only the fourth time this has happened in the last 100 years, and in states like Ohio there were an awful lot of 60-40 results.
I guess what I’m saying is that it seems pretty disingenuous to wave the whole thing away just because it’s not entirely due to gerrymandering. F-P is essentially telling the people in this thread who are trying to understand how we ended up with a Republican majority in the House again when they got less votes, and what can be done about it, that it’s not actually a problem, as if it’s an insane conspiracy theory we dreamed up just to give us a reason to bitch about conservatives. It’s a pattern he established in the Paterno thread: concede some minor point on the one hand, while with the other hand dismissing your entire point by resorting to his own sloppy logic or dressing his intuitions up as fact.
One of the complicating factors in the overall popular vote for the House is that districts that are a lock for one party or the other don’t attract quality candidates or big money. The faceless Democrat in my district only got 34% of my county’s vote, but Barack Obama and Sherrod Brown both got 41%. Gerrymandering in favor of one party can drive down the other party’s vote total at the state level simply by making it impossible for them to get good candidates (or any candidates) on the ballot.
But would you want every district in this state to be 55% Democrats and 45% Republicans? Or would you want 55% of the districts to be all Democrats and 45% of the districts to be all Republicans?
If I’m understanding you correctly, you’re saying you don’t care as long as the districting plan meets some theoretical standard of geography. Okay, let’s assume you’re presented with two plans like I described above, each of which are equal under the geographic standard. Which one do you pick?
Sure. But you’re not going to take politics out of the situation. Districting is an inherently political process. It’s like trying to take politics out of elections.
Nemo, we don’t really accept your dichotomy.
The question is, what is a district? Is it a group of people who have some kind of common heritage, or have strong geographic or economic ties? Then they can elect someone of them to represent them.
If it’s only something created by a ruling cartel for teleological reasons, it lacks legitimacy. It’s just perpetuating their control by means of a fictional “district.”
If you’re dividing people up into voting districts, then I think that dividing them up on the basis of how they vote is hardly the worst idea. Seems to make a lot more sense to me than grouping them by something like what town they live in, their religious affiliation, their height, their favorite Beatle, or something like that.
Illegal, it was determined in another thread, but sensible (it makes sense in terms of, well, geographic and economic ties to treat Brooklyn as a single district, but nearly one in seven New Yorkers live here; it kinda makes sense to treat the whole city as one district, and more than 40% of the state lives here).
They wouldn’t be equal by a geometric standard. If we’re assuming our world, where political orientation is strongly correlated with population density, the optimum plan would almost certainly be one with districts for each party.
A problem is that the proper goals of districting are far from clear. In the age of high-speed transportation, geographic compaction is just not too important. Instead a salamander-shaped district that combines the Hispanic neighborhoods of a city actually makes good sense for the desirable goal of flocking birds of a feather together.
The problem is when the feather-flocking is abused to give some groups more net political power than their numbers warrant.
Although not likely to happen in the U.S.A., there might be some simple procedures to reduce the adverse effect of gerrymandering. For example, instead of seniority based on years, legislators could be given seniority (chair committees, etc.) based on the percentage of their district’s votes they received. Or even give double weight to the votes of legislators elected with 60%, triple weight to those with 70%. (This sounds absurd, but perhaps there’s some simpler variation.)
ETA:
Oops. Well just disregard my post then.
This is worth considering.
Here’s an idea: self-selected virtual districts. Divide up the state into however many congressional districts it’s entitled to and number them one to whatever. And then registered voters can pick whichever one they want to join.
I’d speculate there would be political motivation behind the district they choose. Why would liberals join a district that they know has a strong conservative majority and will elect conservative representatives? But if people want to join a district for non-political reasons because there are people from the same town or same job as them, they’ve free to do so.
You’re a Libertarian or a Green or a Tea Partier? You can join a district with other like-minded people. Maybe you’ll have a majority.
And you can’t complain about not be represented because you’re in your district by choice. If you don’t like your Congressman, you can always switch to another district. More people will feel like they have the representation they want instead of one they may have voted against.
If I wanted to say that I would say it. In fact I’ve said nothing of the sort, nor have I “essentially” said it, nor implied it etc. etc.
Fabricating arguments out of the whole cloth and claiming that I’ve essentially said them does not speak well for your integrity.
Is there a reason not to just use a method like Shortest Split Line or some other unbiased algorithm? I mean an ethical or logical reason, not a legal reason – obviously laws would have to be abolished to allow for it (like the part of the Voting Rights Act mentioned above), however unlikely that is to happen.
The problem with using an unbiased algorithm is that most of them, by design or accident, are horribly biased. It’s easy to come up with an algorithm that looks reasonable on the face of it, but different people can come up with superficially-reasonable algorithms that give vastly different results, and then which one do you pick?
That is quite curious, inasmuch as here every effort is made to make urban areas their own riding without stretching them into rural areas. Almost any city in Ontario I can think of that is roughly the size of a federal riding is coterminous, or close to it, with its riding boundaries, incluidng:
<snip>
[/QUOTE]
The 2002 re-distribution was a bit unusual. The Commission was composed of well-respected individuals who had strong public service records: a QB judge, a political scientist from the U of S, and a lawyer who would go on to be president of the Canadian Bar Association.
They originally proposed 14 electoral districts, six urban and eight rural. They made that proposal based on their understanding that at the previous distribution, submissions had been strongly in favour of that sort of split.
However, there were some problems with that proposal, namely Saskatchewan’s unusual population distribution. The south-west corner and the south-east corner are very sparsely populated, but that sparse population is evenly distributed. That meant that the proposed ridings for those areas were huge, if they were not to include portions of Moose Jaw, Regina or Saskatoon; my recollection is that the proposed south-west riding was a huge vertical rectangle, stretching from the 49th way up to the Yellowhead Highway. The south-east proposed riding was similar, but not as extreme.
Then people started coming to the commission’s public hearings, to comment on proposals. And the commission said they were surprised by the uniform opposition to the proposal. No-one spoke in favour of it, and there were many suggestions to keep the existing system of “rurban” seats. The commission had assumed that people would favour clear rural-urban splits, but that was not the case. One of the concerns that was repeatedly expressed was that the rural ridings were simply too big, and would not encompass a community of voters.
As part of their statutory mandate, they were required to take public reaction into account. Given the uniform opposition to their proposals, they re-drafted the map to go back to the “rurban” approach. They explained the realities of Saskatchewan’s unusual population distribution as follows:
That is a different type of distribution than in the areas of Ontario that you mention, RickJay, where you have large urban centres regularly spaced in the province, with the population tapering off as you move away from them. It’s a much starker contrast in Saskatchewan, and be-devils electoral map-drawers.
Full report of the 2002 Commission is available here.
Okay, here’s the deal:*
You have 53 districts in California to divvy up. Assume that every voter has a SSN. Assume that the distribution of final three digits of SSNs is fairly uniform.
The representative from District 1 will represent those voters whose SSNs end in 000-018.
The representative from District 2 will represent those voters whose SSNs end in 019-037.
The representative from District 3 will represent those voters whose SSNs end in 038-056.
And so on.
Seven of the districts will represent blocs of voters with eighteen unique final threes.
Each state will adjust this formula to accommodate the total number of congressional districts. There you have it. No gerrymandering.
*Read that to yourself in a Ross Perot voice, so it sounds as crazy as I intend for it to sound
Run all of them and pick which result you use (uniformly) randomly.