Considering that both were associated with the Democrats, that’s not as close as it looks. Might have been 70% if we had 2008-level turnout.
I wonder if this will effect Corrine Brown’s district.
http://nationalatlas.gov/printable/images/preview/congdist/FL03_109.gif
It is practically a textbook gerrymander. With the last election it looks like a blue island sitting in a sea of red.
I don’t see it that way–in fact, it strikes me that you could set up an algorithm that drew perfectly compact districts without any human judgment whatsoever. Simply give the computer all the information about where everyone lives, and how many equally populated districts to divide the state into. Then the computer looks for the configuration in which the total length of all boundary lines between districts is the smallest number.
Optionally, you could preserve a bit of the gerrymandering prerogative of the state government by allowing them to take this map as a baseline and then giving them leeway to draw a map with total length of boundary lines up to X% longer than the baseline (hopefully no more than 10 as a value for X).
I’m not optimistic, but I’d really love to see a constitutional amendment require something like this because I’m so sick of gerrymandering.
I read quite a bit here before realizing the thread was a zomby. I decided to post; the point I wanted to make is the same as that of the zomby-reviver:
There may be no clear best solution or best algorithm, and a global optimization like SlackerInc’s might be computationally prohibitive. Different algorithm designers would produce different districtings.
But that doesn’t matter! With no info on ethnics, politics, etc., the districting produced would have no systemic bias.
On the last level, you can have the courts approve it instead of the legislature. Don’t worry, I missed it the first time round, too. 
Looking for problems–what’s to prevent the algorithm’s designer from designing multiple algorithms, and choosing the one that draws the districts most favorable to his party?
Perhaps the Governor’s Association (or whatever they’re called) could contract with an expert commission to develop guidelines for such an algorithm, and then contract with someone else to develop the algorithm using open-source code. They could do something like the secretary problem: run a billion sample district lines, choose the best of those, and then run more samples until either they’ve found one better than the best-in-a-billion, or they’ve reached an arbitrary number that means running more samples has decreasing chance of finding a better sample.
If I understand the proposal, equal districts with shortest lines guarantees that districts are as close as possible to circular. And that’s a good goal. However, I wonder if a better goal would be to have the lowest range in district perimeters–that is, the district with the smallest perimeter is as close as possible to the district with the largest perimeter? This prevents a solution with a lot of normal-shaped districts at the cost of one ridiculously-shaped district. Or maybe you figure out what a rectangular grid looks like, and you try to get the lowest score compared to this ideal district?
Neither of those would work, since population is concentrated unevenly. Look at a state like Nebraska, for instance: There’s one district that’s basically Omaha, one that’s the rest of the eastern quarter of the state including Lincoln, and one that’s the entire rest of the state. Omaha by itself is about a third of the state’s population, and any district including it is going to have to be tiny, while any district that covers the rural western parts has to be huge to encompass enough population. Trying to make all three districts the same size would require gerrymandering of the worst sort.
I do have an idea for a modification of SlackerInc’s idea that would be computationally feasible, though. Start by having some commission come up with a map, using whatever methods they like. If everyone is OK with that map, fine. But then, any state legislator can come up with a new map, and whichever map has the lowest total boundary length replaces the one the commission comes up with.
I imagine that in practice, this would lead to each of the two major parties making a map, and the final map would end up being one drawn by one of the parties. But even though it’d be a party-drawn map, they still couldn’t get away with any blatant shenanigans, because if they tried, the other party would be able to steal the map from them.
It’s worth noting that the anti-gerrymandering amendments have yet to go into effect because Governor Babyeater has effectively blocked them by failing to submit them for Justice Department preclearance.
I like it. It’s a variant on the classic “one gets to cut the pie, the other gets first choice of the pieces”. Worked great when my kids were fighting over the last of the ice cream or something. Enlightened self interest (read: simple greed) forced the person doing the dividing to split the booty evenly right down to the microgram.
If indeed my proposal is computationally non-feasible, then this looks like a great modification. As CannyDan says, it’s a bit like the ol’ “one divides, the other chooses” (though with my kids, it is a struggle to get them to decide which has to divide, that being seen as the short end of the stick with some justification).
I’m a bit disappointed though to learn that despite having supercomputers that can do massive simulation of weather systems and so on, they can’t do what I wanted to task them to do? Really? Huh.
Even if the computer model is feasible, I kind of like the semi-bidding structure of Chronos’s model better, since it neatly sidesteps any accusations about the partisanship of the modelers.
Actually, I think how I’d set it up would be like this:
The majority party submits their map.
The minority party either accepts their map, or else they submit an alternate map with lower total boundary lines by some significant margin (5%, perhaps)
If the minority party submits a qualifying alternate map, that’s the one that’s used. Otherwise, the majority party’s map is used.
I’m not sure what the need would be for a commission, if it’s going to be assumed that everyone ignores the commission’s map.
That’s brilliant! Absolutely wonderful.
My only caveat would be that political parties are not technically required in legislative bodies; also, it is in theory possible that such a body could have more than one minority party (and for that matter, could have no single majority party). What kind of mechanism could we use that would get around this, while accomplishing more or less the same thing?
I think the mechanism we could use would be to enact this in the United States, where every state legislature (AFAIK) is composed overwhelmingly of Democrats and Republicans. If at some point this is no longer true, a future House could change the law to accommodate their changed circumstances.
Although that also points to a fatal flaw in the process: how often does the majority in a legislative body support a proposal that gives more power to the minority? I could barely see it happening during a session midway between Censuses, but not otherwise.
That’s why in my idea I started with a commission and let each individual legislator challenge it. That way you don’t need to worry about what party those individuals are, and if the Stark Raving Loony Party manages to get a seat in the legislature, they get a voice. And the occasional seat held by a third party is hardly unheard-of, at the state level: One state once managed to even elect someone under the banner of the NASCAR party.
Right, good point. I’d suggest though at least preserving the LHoD’s concept of requiring that any “improved” map submitted be at least 5% less in boundary line distance. And we need to have some kind of deadline for the last possible submission (then, I guess, if a bunch of legislators hand in maps right at the end, the lowest wins–or if none were more than 5% lower than the others, the whole legislature could vote among the lowest qualifiers for the one they prefer).
Oh, and:
“Find the best possible solution to problem X” tends to be very hard-- Such problems are often provably NP-hard, which is computer jargon for “if there exist any really hard computer problems at all, this is one of them”. Finding such an absolute-best solution quickly turns into the sort of thing that would take a computer longer than the lifetime of the Universe. However, for such problems there’s often a fairly simple algorithm that’s good enough: It won’t guarantee that it’s actually the best, but it’ll almost certainly be within some small percentage of being best, such that finding a better solution would be very difficult. So you wouldn’t want a law to say that the state has to use the absolute best districting, since you could never be sure if you have it or not.
None of these will work, realistically. These ideas are concerned only with boundary lines and perimeter distance – they ignore several other criteria that are supposed to be considered in redistricting:[ul]
[li]paying attention to major natural or human delineations, like rivers, freeways, etc. A river is a clear dividing line, and often makes a good divider. But they follow the contour of the land, and not the ‘shortest perimeter’.[/li][li]paying attention to existing boundaries, like city limits or county lines. It’s better to keep a district all within a single city or county whenever possible. [/li][li]paying attention to existing neighborhoods within towns & cities. [/li][li]paying attention to housing patterns, to create ‘minority opportunity districts’. For[/li] example, if there are several neighborhoods with a large number of black voters, you should try to keep those within a single district, to increase the chance of this minority electing one of their own. As opposed to splitting those communities among several districts, thus diluting their voting power.
[li]finally, paying attention to historical lines – where the old district lines were. People are used to those, and having the least changes needed is better.[/li][/ul]
All of these factors are included in redistricting guidelines, and even in state law in some places. A computer program that goes simply on compactness or shortest perimeter would really mess up some of these, and produce an unacceptable map.
I don’t especially understand why any of these, other than the last two, are better. What makes district A, that follows the contours of the river, better than district B, that doesn’t?
As for minority districts, I don’t know enough to have a firm opinion on the subject. I suspect they exist in response to previous anti-minority gerrymandering; with the elimination of gerrymandering, they may become irrelevant. There may also be the beneficial effect of forcing political parties to respect minority concerns in multiple districts instead of in just one.
And the advantages of historical districts would be very transitory and trivial, compared to the advantage of eliminating partisan gerrymandering.
I agree with all of this. Also, it’s unlikely that compact districts would dilute minority representation to an unacceptable level. Districts that do that tend to be those that take a metropolitan area, including the inner city and the inner-ring suburbs that tend to have a higher population, plus the further suburbs and exurbs, and divide them up like a pizza, with triangular “slices” that use the outlying white areas (near the “crust”, if you will) to form a majority. Not much chance of this happening with a system of compact boundaries.
And I know the other considerations are not universally applied. I lived in Lawrence, KS for a year, and because I lived in the western side of the city I was in one congressional district (and yes, this was for the U.S. Congress, not the state legislature); but other people who lived a few blocks to the east were in a different district. I’m not sure any of the advantages claimed for keeping cities within one district aren’t outweighed by advantages of having them represented by more than one U.S. Representative.
In any event, nothing about the plan we’ve been advocating is negated by any of this (except maybe the “historical boundaries” issue, which is the biggest canard of them all as I’ve seen districts get radically redrawn many times), if it is deemed to be so important. We can simply state that any map submitted by a legislator must follow whatever state law already applies in terms of municipal boundaries, rivers, etc. Then as long as it is found to be kosher in that respect, it can be judged secondarily based on the length of the boundary lines between districts.