I Cut, You Choose
November 10, 2017 8:31 PM Subscribe
CMU researchers suggest a new method of redistricting which might create voting districts which prevent gerrymandering. It's all about who cuts the cake, and who gets to choose their piece.
It calls for one political party to divide a map of a state into the allotted number of districts, each with equal numbers of voters. Then the second party would choose one district to "freeze," so no further changes could be made to it, and re-map the remaining districts as it likes.
The first party then would choose a second district to freeze from this map and proceed to redraw the remaining districts as it sees fit. This back-and-forth process would continue until all of the districts are frozen. For a state such as Pennsylvania with 18 congressional districts, this would require 17 cycles.
It seems like if we had a mix of districts and at large candidates, with the proportion of district votes for each party within the state that didn't result in a district candidate going to determine the at large candidate contest it would mostly remove the incentive for gerrymandering. Especially if there were a few at large candidates on the national scale even to iron out the last bit of discrepancy. Of course, that last option would mean a state that had a tampered election would affect the national count of reps.
posted by BrotherCaine at 9:08 PM on November 10, 2017 [2 favorites]
posted by BrotherCaine at 9:08 PM on November 10, 2017 [2 favorites]
Of course, what I just said would mean we'd get some third party at large candidates, so I think it'd never happen.
posted by BrotherCaine at 9:11 PM on November 10, 2017
posted by BrotherCaine at 9:11 PM on November 10, 2017
It's a little too late in the day for my brain to properly prove this, but my intuition is saying that if a state has a party with a fairly strong majority for one party - 60-40, say - then the best strategy for the majority party is to always draw 60-40 districts, while the best strategy for the minority party is to draw districts as close to the extremes of 0-100 and 100-0 as possible.
60-40 for the majority party: The more of these districts they get, the closer they get to winning all of the state's seats. They can make all districts 60-40 on their turn to carve up the cake, thus giving the minority party no real choice.
0-100 and 100-0 for the minority party: If the majority party chooses the guaranteed win, it brings all the remaining districts closer to a 50-50 tossup. If the majority party chooses the guaranteed loss, they've given up a seat.
Am I close? Or is there some game theoretic stuff that I'm missing?
posted by clawsoon at 9:26 PM on November 10, 2017 [8 favorites]
60-40 for the majority party: The more of these districts they get, the closer they get to winning all of the state's seats. They can make all districts 60-40 on their turn to carve up the cake, thus giving the minority party no real choice.
0-100 and 100-0 for the minority party: If the majority party chooses the guaranteed win, it brings all the remaining districts closer to a 50-50 tossup. If the majority party chooses the guaranteed loss, they've given up a seat.
Am I close? Or is there some game theoretic stuff that I'm missing?
posted by clawsoon at 9:26 PM on November 10, 2017 [8 favorites]
This is cool. Thanks for posting.
posted by latkes at 9:35 PM on November 10, 2017 [1 favorite]
posted by latkes at 9:35 PM on November 10, 2017 [1 favorite]
This is not a problem without precedent. Do what we do in Australia: establish an independent electoral commission to draw boundaries, bound by specific rules, requirements and methods for drawing boundaries.
Getting parties involved with creating electorates is madness.
posted by smoke at 9:57 PM on November 10, 2017 [64 favorites]
Getting parties involved with creating electorates is madness.
posted by smoke at 9:57 PM on November 10, 2017 [64 favorites]
Garbage in terms of real world utility. There aren't two parties and multi player game allows collusion.
FTA: Partially funded by the Office of Naval Research
Choose your battles, man...There aren't two parties, but a third party hasn't ever been a probability without a disruption along the lines of D] None of the above, and revotes. So, in this respect, Voltaire?
posted by lazycomputerkids at 10:02 PM on November 10, 2017 [1 favorite]
FTA: Partially funded by the Office of Naval Research
Choose your battles, man...There aren't two parties, but a third party hasn't ever been a probability without a disruption along the lines of D] None of the above, and revotes. So, in this respect, Voltaire?
posted by lazycomputerkids at 10:02 PM on November 10, 2017 [1 favorite]
> Garbage in terms of real world utility. There aren't two parties
In 2016 95% of the votes for president went to one of two candidates. Among those two, the votes were divided 49-51. I know not every state is evenly divided, but a system that gives these two parties equal power over redistricting sounds very useful in a practical sense. It would be useless almost everywhere outside of the US.
The only problem with real-world utility is that the ruling party would need to agree among themselves to do it.
posted by Pruitt-Igoe at 10:18 PM on November 10, 2017 [5 favorites]
In 2016 95% of the votes for president went to one of two candidates. Among those two, the votes were divided 49-51. I know not every state is evenly divided, but a system that gives these two parties equal power over redistricting sounds very useful in a practical sense. It would be useless almost everywhere outside of the US.
The only problem with real-world utility is that the ruling party would need to agree among themselves to do it.
posted by Pruitt-Igoe at 10:18 PM on November 10, 2017 [5 favorites]
There's probably some obvious reason I'm missing, as I do, but do we even need districts (or precincts, or wards, or whatever) at all? Can't all state and federal offices be statewide at-large? Can't all county ones be county-wide and local ones be city-wide?
posted by FelliniBlank at 10:25 PM on November 10, 2017 [1 favorite]
posted by FelliniBlank at 10:25 PM on November 10, 2017 [1 favorite]
A Democratic Socialist just won a district in Virginia. I don't want them shut out of their chance to grow by further privileging the two major parties. (Already, they get advantages - they're not required to qualify for the ballots the way smaller parties are.)
I do like "one group does the dividing; another group chooses which divisions to take." But also want a nonpartisan commission - a group of people who are not elected and therefore have no direct personal stake in how the districts are arranged. Get a commission, have them research demographics and community identities, DON'T give them any data about party membership - and then split them into two groups: One group divides the state into districts; the other chooses one of those districts to lock into place - and then they switch.
It'd take some time for them to finalize a plan, but when they were done, it'd be as devoid of party-specific bias as we could make it.
posted by ErisLordFreedom at 10:25 PM on November 10, 2017 [7 favorites]
I do like "one group does the dividing; another group chooses which divisions to take." But also want a nonpartisan commission - a group of people who are not elected and therefore have no direct personal stake in how the districts are arranged. Get a commission, have them research demographics and community identities, DON'T give them any data about party membership - and then split them into two groups: One group divides the state into districts; the other chooses one of those districts to lock into place - and then they switch.
It'd take some time for them to finalize a plan, but when they were done, it'd be as devoid of party-specific bias as we could make it.
posted by ErisLordFreedom at 10:25 PM on November 10, 2017 [7 favorites]
I'm curious what this means in terms of incumbents trying to keep their seats. That's not necessarily a red/blue divide, and if the people picking the seats are mostly elected officials, I'd think that'd be a key priority, possibly more than "how can my party get more influence?"
posted by ikea_femme at 10:38 PM on November 10, 2017 [2 favorites]
posted by ikea_femme at 10:38 PM on November 10, 2017 [2 favorites]
The fundamental flaw with this is that it assumes the fundamental purpose of dividing an area into districts is to maximize political advantage, rather than ensure local representation.
There are a number of other flaws. It assumes two parties for perpetuity. It assumes the two parties are evenly distributed -- should 50% of Massachusetts districts really be drawn by the 20% who are Republicans?
It ignores that perhaps the only thing all politicians can agree on is that they would like to be reelected. California spent a long time under a bipartisan gerrymander, where the Democrats had a perpetual majority (but never a supermajority) and everybody got reelected - the 2000 congressional districts had exactly one seat flip in five elections - 1 in 265.
It would be incredibly time-consuming, with parties trading back and forth for months on end. Let's say it takes two weeks to make a "move" in this "game". That's actually not a lot of time to get a proposed map, analyze it, pick a district to freeze, and then draw a new proposed map - especially since the district you pick to freeze depends on what your next proposed map is, so you would actually draw multiple proposed maps each conditional on freezing a single district. FL and NY would take a year to allocate congressional districts, Texas 16 months and California 2 years. That's still reasonable, I think. Except that states have legislatures, and those have boundaries, too. Only 10 upper houses would finish in the first year, and four - MN, GA, IL, NY - would be drawing boundaries two years later. The lower houses are much larger - 20 states would still today be using districts based on the 2000 Census, and Pennsylvania and New Hampshire would still be using 2000 Census boundaries in the 2018 elections.
There is also no guarantee -- in fact, this probably guarantees the opposite -- that districts make any sort of geographic sense. Almost certainly, the final two districts will be completely scattered around the entire state, potentially block by block.
Getting parties involved with creating electorates is madness.
Exactly; the Canadian system also involves independent commissions with public hearings and official open submissions from politicians (often rejected), and produces very reasonable boundaries.
posted by Homeboy Trouble at 11:13 PM on November 10, 2017 [17 favorites]
There are a number of other flaws. It assumes two parties for perpetuity. It assumes the two parties are evenly distributed -- should 50% of Massachusetts districts really be drawn by the 20% who are Republicans?
It ignores that perhaps the only thing all politicians can agree on is that they would like to be reelected. California spent a long time under a bipartisan gerrymander, where the Democrats had a perpetual majority (but never a supermajority) and everybody got reelected - the 2000 congressional districts had exactly one seat flip in five elections - 1 in 265.
It would be incredibly time-consuming, with parties trading back and forth for months on end. Let's say it takes two weeks to make a "move" in this "game". That's actually not a lot of time to get a proposed map, analyze it, pick a district to freeze, and then draw a new proposed map - especially since the district you pick to freeze depends on what your next proposed map is, so you would actually draw multiple proposed maps each conditional on freezing a single district. FL and NY would take a year to allocate congressional districts, Texas 16 months and California 2 years. That's still reasonable, I think. Except that states have legislatures, and those have boundaries, too. Only 10 upper houses would finish in the first year, and four - MN, GA, IL, NY - would be drawing boundaries two years later. The lower houses are much larger - 20 states would still today be using districts based on the 2000 Census, and Pennsylvania and New Hampshire would still be using 2000 Census boundaries in the 2018 elections.
There is also no guarantee -- in fact, this probably guarantees the opposite -- that districts make any sort of geographic sense. Almost certainly, the final two districts will be completely scattered around the entire state, potentially block by block.
Getting parties involved with creating electorates is madness.
Exactly; the Canadian system also involves independent commissions with public hearings and official open submissions from politicians (often rejected), and produces very reasonable boundaries.
posted by Homeboy Trouble at 11:13 PM on November 10, 2017 [17 favorites]
then the best strategy for the majority party is to always draw 60-40 districts, while the best strategy for the minority party is to draw districts as close to the extremes of 0-100 and 100-0 as possible.
My thinking is that no matter whether you are the majority of the minority, whenever you are dividing a map into n segments, you should always draw one district containing 100% of the other party's remaining voters and (n-1) districts containing (100/(n-1))% of your party's remaining voters. The other party is then guaranteed to freeze one of the many districts containing your voters; therefore, every turn, whoever's turn it was to draw the map is guaranteed to pick up a new district. This guarantees that each party will get 50% of the districts, and because getting 50% is possible, neither party has incentive to settle for a strategy that might get them fewer than 50%.
posted by one for the books at 11:15 PM on November 10, 2017 [1 favorite]
My thinking is that no matter whether you are the majority of the minority, whenever you are dividing a map into n segments, you should always draw one district containing 100% of the other party's remaining voters and (n-1) districts containing (100/(n-1))% of your party's remaining voters. The other party is then guaranteed to freeze one of the many districts containing your voters; therefore, every turn, whoever's turn it was to draw the map is guaranteed to pick up a new district. This guarantees that each party will get 50% of the districts, and because getting 50% is possible, neither party has incentive to settle for a strategy that might get them fewer than 50%.
posted by one for the books at 11:15 PM on November 10, 2017 [1 favorite]
Never mind, missed the instruction that districts have to contain "equal numbers of voters."
posted by one for the books at 11:27 PM on November 10, 2017 [1 favorite]
posted by one for the books at 11:27 PM on November 10, 2017 [1 favorite]
For people who favor a nonpartisan commission: how do people decide what counts as "nonpartisan?" In a country as polarized as ours, I think it's not realistic to say that an independent committee would be unable to consider partisanship, or to at least predict it from demographic information, and there's also the danger of capture of the process by covert partisans (particularly if one party's ideology is particularly dominant). If the answer is actually "form a bipartisan committee," then this sounds like a formalized way of accomplishing that.
posted by en forme de poire at 11:53 PM on November 10, 2017 [4 favorites]
posted by en forme de poire at 11:53 PM on November 10, 2017 [4 favorites]
For a completely untrained type, I freaking love game theory, and watching folks like the ones in this thread is why.
posted by Samizdata at 12:26 AM on November 11, 2017 [1 favorite]
posted by Samizdata at 12:26 AM on November 11, 2017 [1 favorite]
I played around with this problem a bit. My solution was to first afix the geophysical center of the state. From that center eight radians are drawn, one in each of the cardinal and ordinal directions. Where those radians intersect the state boundary plus the center point provides nine starting points for a piece of open source software. This software knows only of state borders, existing roadways for dividing lines, population densities, the number of districts required and the +/- allowance for population difference between districts. It is also constrained by trying to produce the most compact (ie perfect circle) districts possible. It knows nothing of party affiliations, income, ethnicity, race, etc. Once run, it should provide nine different maps for the legislature, however comprised, to haggle over and choose from...
posted by jim in austin at 1:19 AM on November 11, 2017 [1 favorite]
posted by jim in austin at 1:19 AM on November 11, 2017 [1 favorite]
how do people decide what counts as "nonpartisan?"
here is a good breakdown of our process.
posted by smoke at 3:02 AM on November 11, 2017 [2 favorites]
0-100 and 100-0 for the minority party: If the majority party chooses the guaranteed win, it brings all the remaining districts closer to a 50-50 tossup. If the majority party chooses the guaranteed loss, they've given up a seat.
Yes, but they don't have to do that very often until party B is out of voters. You end up with however many 55:45 that A drew, the same number of 0:100 that B drew, and then an equal or larger number of 100:0 A:B that are leftover when nobody has choices left. Party B is always better off forcing A to freeze either a district that B is winning 45:55 or take a 100:0 win, which pushes the remaining vote closer to 50:50. I think that you do naively end up with a bunch of close districts and a couple of solid "leftover" 100:0's. Of course A can see the amount of leftover likely to happen and draw some of their maps with more 60:40 or so (or whatever their two-party share is) to have a margin of safety.
I'm curious what this means in terms of incumbents trying to keep their seats.
Unless the opposite party can draw a map that screws all your incumbents at a given stage, you wouldn't freeze that district. It'll certainly be a problem, but no more of one than now.
It assumes the two parties are evenly distributed -- should 50% of Massachusetts districts really be drawn by the 20% who are Republicans?
The minority party draws the same number of districts, but gets the right number of seats. I suppose that out of spite they could draw logistically annoying boundaries.
California spent a long time under a bipartisan gerrymander
Collusion between the parties and non-faithful representation are a real problem with implementation.
It would be incredibly time-consuming, with parties trading back and forth for months on end.
Right now they already use computer programs to draw the optimal districts. I don't see why each step ought take that long (other than one party might want to gum up the process if allowed).
There would certainly need to be some additions to try to preserve some geographic continuity and prevent incumbent-stealing.
posted by a robot made out of meat at 5:15 AM on November 11, 2017 [1 favorite]
Yes, but they don't have to do that very often until party B is out of voters. You end up with however many 55:45 that A drew, the same number of 0:100 that B drew, and then an equal or larger number of 100:0 A:B that are leftover when nobody has choices left. Party B is always better off forcing A to freeze either a district that B is winning 45:55 or take a 100:0 win, which pushes the remaining vote closer to 50:50. I think that you do naively end up with a bunch of close districts and a couple of solid "leftover" 100:0's. Of course A can see the amount of leftover likely to happen and draw some of their maps with more 60:40 or so (or whatever their two-party share is) to have a margin of safety.
I'm curious what this means in terms of incumbents trying to keep their seats.
Unless the opposite party can draw a map that screws all your incumbents at a given stage, you wouldn't freeze that district. It'll certainly be a problem, but no more of one than now.
It assumes the two parties are evenly distributed -- should 50% of Massachusetts districts really be drawn by the 20% who are Republicans?
The minority party draws the same number of districts, but gets the right number of seats. I suppose that out of spite they could draw logistically annoying boundaries.
California spent a long time under a bipartisan gerrymander
Collusion between the parties and non-faithful representation are a real problem with implementation.
It would be incredibly time-consuming, with parties trading back and forth for months on end.
Right now they already use computer programs to draw the optimal districts. I don't see why each step ought take that long (other than one party might want to gum up the process if allowed).
There would certainly need to be some additions to try to preserve some geographic continuity and prevent incumbent-stealing.
posted by a robot made out of meat at 5:15 AM on November 11, 2017 [1 favorite]
I think the idea of treating districting as a two-player game is actually a pretty good one. Here's a twitter thread where I talk about it a bit; the protocol I had in mind was different from Wes's but the fundamental idea is the same. My belief (but this is an empirical question) is that the highly gerrymandered maps we have now are pretty brittle, and would be very hard to maintain against another player with an equal number of moves.
Re some of the issues raised in this thread:
1. Yes, this keeps two-party dominance structurally in place. So do tons of other features of our legislative electoral system. You could argue that ideas like this represent a way to keep one-party dominance from being locked structurally in place.
2. Yes, this would probably lead to more districts drawn to protect incumbents, since games often equilibriate to seek goals that are agreeable to both players, and "I want to stay in office" is something GOP and Dem officeholders can agree on. This is a fundamental tradeoff. For reasons mentioned above, the goals of "fair map not drawn to advantage one party" and "map with as many truly competitive districts as possible" are often in tension.
posted by escabeche at 6:21 AM on November 11, 2017 [7 favorites]
Re some of the issues raised in this thread:
1. Yes, this keeps two-party dominance structurally in place. So do tons of other features of our legislative electoral system. You could argue that ideas like this represent a way to keep one-party dominance from being locked structurally in place.
2. Yes, this would probably lead to more districts drawn to protect incumbents, since games often equilibriate to seek goals that are agreeable to both players, and "I want to stay in office" is something GOP and Dem officeholders can agree on. This is a fundamental tradeoff. For reasons mentioned above, the goals of "fair map not drawn to advantage one party" and "map with as many truly competitive districts as possible" are often in tension.
posted by escabeche at 6:21 AM on November 11, 2017 [7 favorites]
Winner Take All is always going to end up with a two party system. We don't have proportional representation here, I don't know why the Founding Fathers [who according to some had the Constitution handed to them directly by Jesus] didn't go that route. But in their unfailing wisdom, we abide by their sacred, unchangeable choices.
Oh wait, Senators used to be chosen by the State legislatures. Now by the popular vote. Hrm.
posted by hippybear at 6:31 AM on November 11, 2017 [1 favorite]
Oh wait, Senators used to be chosen by the State legislatures. Now by the popular vote. Hrm.
posted by hippybear at 6:31 AM on November 11, 2017 [1 favorite]
Getting parties involved with creating electorates is madness.
As the primary deciders certainly but perhaps given a certain number of "veto's" in a process that attempts to be independent but with some kind of sequential dividing up process. Then significant independent parties could be included with a proportionate "veto" voice.
As for taking a long time, I'd expect it'd be 20% that were complex disputes and the rest fall out somewhat naturally.
posted by sammyo at 6:55 AM on November 11, 2017
As the primary deciders certainly but perhaps given a certain number of "veto's" in a process that attempts to be independent but with some kind of sequential dividing up process. Then significant independent parties could be included with a proportionate "veto" voice.
As for taking a long time, I'd expect it'd be 20% that were complex disputes and the rest fall out somewhat naturally.
posted by sammyo at 6:55 AM on November 11, 2017
Am I close? Or is there some game theoretic stuff that I'm missing?
Nope. Those were most of the same criticisms that I would make.
I'd add that the result that the promoters seem to think would happen -- trading off choosing districts just barely reliably on your side -- is exactly the same thing as "Every state will be gerrymandered in favor of its minority party." Same as plans that try to maximize competitive districts.
You'd probably need constitutional tinkering to do it, but the Germans (and allied occupation leaders) got this one right: people vote for a representative of their district, but they also vote for a party in the Bundestag. After the election, the size of the Bundestag is increased with members from party lists until the party ratios in the chamber closely approximate the votes for parties.
posted by GCU Sweet and Full of Grace at 7:15 AM on November 11, 2017 [3 favorites]
Nope. Those were most of the same criticisms that I would make.
I'd add that the result that the promoters seem to think would happen -- trading off choosing districts just barely reliably on your side -- is exactly the same thing as "Every state will be gerrymandered in favor of its minority party." Same as plans that try to maximize competitive districts.
You'd probably need constitutional tinkering to do it, but the Germans (and allied occupation leaders) got this one right: people vote for a representative of their district, but they also vote for a party in the Bundestag. After the election, the size of the Bundestag is increased with members from party lists until the party ratios in the chamber closely approximate the votes for parties.
posted by GCU Sweet and Full of Grace at 7:15 AM on November 11, 2017 [3 favorites]
There's probably some obvious reason I'm missing, as I do, but do we even need districts (or precincts, or wards, or whatever) at all? Can't all state and federal offices be statewide at-large? Can't all county ones be county-wide and local ones be city-wide?
They don't have to be by any means, but at-large elections were a tool in the Jim Crow toolkit. Say you have a town that's 70\% white, 30\% black with the black voters concentrated in one part of town. If you elect from 5 or 10 districts, it would be hard to avoid drawing at least one or two majority-black districts, so black people will have a direct input into city policy (and probably still lose).
These towns would sometimes switch to at-large elections so that every election was in that full-sized 70/30 districts. If you combine this with a sufficiently racist white electorate, which they had, then racist white voters can refuse to vote for black candidates or candidates who'd made overtures to local black voters, and keep an entirely white-supremacist city government.
posted by GCU Sweet and Full of Grace at 7:24 AM on November 11, 2017 [10 favorites]
They don't have to be by any means, but at-large elections were a tool in the Jim Crow toolkit. Say you have a town that's 70\% white, 30\% black with the black voters concentrated in one part of town. If you elect from 5 or 10 districts, it would be hard to avoid drawing at least one or two majority-black districts, so black people will have a direct input into city policy (and probably still lose).
These towns would sometimes switch to at-large elections so that every election was in that full-sized 70/30 districts. If you combine this with a sufficiently racist white electorate, which they had, then racist white voters can refuse to vote for black candidates or candidates who'd made overtures to local black voters, and keep an entirely white-supremacist city government.
posted by GCU Sweet and Full of Grace at 7:24 AM on November 11, 2017 [10 favorites]
From 2008 there's a great report on how Canada got to non-partisan redistricting. Long story short, people got fed up with partisan gridlock in the 1960s and forced a change. And it worked.
California recently switched to a non-partisan process. State Republicans agreed to it because they were so far out of power they felt it was their better choice, then various referendums got put on the state ballot 2008-2010 and now we have an independent districting committee. Jury is still out on whether it really is non-partisan. It doesn't make room for third parties.
Arizona also has nominally non-partisan redistricting. But according to the excellent book Ratf**ked in practice it is anything but non-partisan.
posted by Nelson at 7:28 AM on November 11, 2017 [6 favorites]
California recently switched to a non-partisan process. State Republicans agreed to it because they were so far out of power they felt it was their better choice, then various referendums got put on the state ballot 2008-2010 and now we have an independent districting committee. Jury is still out on whether it really is non-partisan. It doesn't make room for third parties.
Arizona also has nominally non-partisan redistricting. But according to the excellent book Ratf**ked in practice it is anything but non-partisan.
posted by Nelson at 7:28 AM on November 11, 2017 [6 favorites]
hippybear: Winner Take All is always going to end up with a two party system.
That has turned out to only be true in the U.S., which means that it's not true. Having winner-take-all lead to a two party system turns out to be the anomaly which needs explaining. Most other nations with winner-take-all regularly produce third and fourth parties which upset the electoral calculus. (Having a strong-president system without ending up with a collapse or a coup by either president or legislature is also anomalous. In short: US democratic politics is weird and exceptional.)
posted by clawsoon at 7:43 AM on November 11, 2017 [6 favorites]
That has turned out to only be true in the U.S., which means that it's not true. Having winner-take-all lead to a two party system turns out to be the anomaly which needs explaining. Most other nations with winner-take-all regularly produce third and fourth parties which upset the electoral calculus. (Having a strong-president system without ending up with a collapse or a coup by either president or legislature is also anomalous. In short: US democratic politics is weird and exceptional.)
posted by clawsoon at 7:43 AM on November 11, 2017 [6 favorites]
Previously, and I like the idea of gaming the boundary, but it needs local input, because the problem in legal history was always about race and representation.
posted by Brian B. at 8:13 AM on November 11, 2017
posted by Brian B. at 8:13 AM on November 11, 2017
I don't trust any sort of scheming and strategizing as a redistricting method. If two or more sides are competing to get an advantage, someone will have better strategists and make better choices. Someone will figure out how to game the system, or deals will be struck in back rooms, or money will exchange hands.
Redistricting should be as automatic, transparent, and data-driven as possible. Given an open algorithm and the latest census data (including race and income and so on), let redistricting automatically happen after each census. If people moved in or out, redistricting would reflect it.
It would require accurate census data, but all proper governance requires accurate census data. This would just add pressure to keep it accurate.
posted by pracowity at 8:29 AM on November 11, 2017 [1 favorite]
Redistricting should be as automatic, transparent, and data-driven as possible. Given an open algorithm and the latest census data (including race and income and so on), let redistricting automatically happen after each census. If people moved in or out, redistricting would reflect it.
It would require accurate census data, but all proper governance requires accurate census data. This would just add pressure to keep it accurate.
posted by pracowity at 8:29 AM on November 11, 2017 [1 favorite]
In terms of real-world utility, this is exactly like fantasizing about FTL drives.
posted by aramaic at 8:29 AM on November 11, 2017 [1 favorite]
posted by aramaic at 8:29 AM on November 11, 2017 [1 favorite]
Start at state capitol building/city hall, draw 360 pie slices out to the borders of the state/county. Starting point is chosen at random. Add slices until wedge population=number of districts÷population of state/county (+/- as close as possible). Keep the 360 slices (so you don't have to keep figuring which building is where), but redraw the districts (starting with a new random start point) at every new census.
posted by sexyrobot at 8:37 AM on November 11, 2017 [1 favorite]
posted by sexyrobot at 8:37 AM on November 11, 2017 [1 favorite]
Start at state capitol building/city hall, draw 360 pie slices out to the borders of the state/county.
This would significantly and systematically advantage rural voters by dividing up voters who live in the capital / town. It would be more of an issue in states where the capital is one of the larger cities.
I think a geometric center point would be more fair.
posted by straight at 9:02 AM on November 11, 2017 [1 favorite]
This would significantly and systematically advantage rural voters by dividing up voters who live in the capital / town. It would be more of an issue in states where the capital is one of the larger cities.
I think a geometric center point would be more fair.
posted by straight at 9:02 AM on November 11, 2017 [1 favorite]
independent electoral commission
No such thing
posted by 922257033c4a0f3cecdbd819a46d626999d1af4a at 9:30 AM on November 11, 2017 [1 favorite]
No such thing
posted by 922257033c4a0f3cecdbd819a46d626999d1af4a at 9:30 AM on November 11, 2017 [1 favorite]
I never heard of this method of cake-cutting before, but we used to use something similar when we split up bags of weed.
posted by Cookiebastard at 9:31 AM on November 11, 2017 [2 favorites]
posted by Cookiebastard at 9:31 AM on November 11, 2017 [2 favorites]
That has turned out to only be true in the U.S., which means that it's not true.
Are there any examples of other countries with a winner-take-all system that doesn't have two dominant parties? As far as I know, most countries that hold real elections have proportional representation. I'd love to hear a tale of a country with our system that has more than two parties vying for power. Please share!
posted by hippybear at 10:19 AM on November 11, 2017
Are there any examples of other countries with a winner-take-all system that doesn't have two dominant parties? As far as I know, most countries that hold real elections have proportional representation. I'd love to hear a tale of a country with our system that has more than two parties vying for power. Please share!
posted by hippybear at 10:19 AM on November 11, 2017
hippybear: Are there any examples of other countries with a winner-take-all system that doesn't have two dominant parties?
Canada comes to mind first, since I live here. :-) In the past 20 years, we've had 5 different 2nd-place parties (Bloc Quebecois, Reform, Conservatives, Liberals, NDP). The UK has had, I think, 4 parties winning respectable numbers of seats in recent history (Conservatives, Labour, Liberal Democrats, Scottish National Party). Recent reviews have found that the same is true of every other first-past-the-post country other than the US.
posted by clawsoon at 10:38 AM on November 11, 2017 [3 favorites]
Canada comes to mind first, since I live here. :-) In the past 20 years, we've had 5 different 2nd-place parties (Bloc Quebecois, Reform, Conservatives, Liberals, NDP). The UK has had, I think, 4 parties winning respectable numbers of seats in recent history (Conservatives, Labour, Liberal Democrats, Scottish National Party). Recent reviews have found that the same is true of every other first-past-the-post country other than the US.
posted by clawsoon at 10:38 AM on November 11, 2017 [3 favorites]
> we've had 5 different 2nd-place parties
... and that still leaves you with two dominant parties?
posted by farlukar at 11:02 AM on November 11, 2017
... and that still leaves you with two dominant parties?
posted by farlukar at 11:02 AM on November 11, 2017
As long as we're not talking about the practicality, I will throw this back out:
This. isn't. hard.
posted by Seeba at 11:07 AM on November 11, 2017
This. isn't. hard.
posted by Seeba at 11:07 AM on November 11, 2017
One of the advantages of this method is that it is very simple and therefore actually has a chance of being codified into law. Mathematical tests for gerrymandering and tools for drawing non-gerrymandered "equilibrium" districts have gotten very sophisticated, and Pegden is actually an author of one of these. But for something so fundamental to our political system, I don't think the electorate is going to be enthusiastic about a solution that requires advanced math to implement or understand. Especially when mistrust of expertise is such a mainstay feature of American politics.
Reading about the Australian system, I'm still skeptical about a "non-partisan" committee in the USA, where literally every aspect of our system has been captured by partisan politics. I'm still worried that all such a committee would do would be to make the process of redistricting more opaque, and to hide partisan decisions beneath a veneer of plausible deniability. The page smoke linked says the following:
posted by en forme de poire at 11:29 AM on November 11, 2017 [1 favorite]
Reading about the Australian system, I'm still skeptical about a "non-partisan" committee in the USA, where literally every aspect of our system has been captured by partisan politics. I'm still worried that all such a committee would do would be to make the process of redistricting more opaque, and to hide partisan decisions beneath a veneer of plausible deniability. The page smoke linked says the following:
The committee draws up a proposed redistribution that considers the current boundaries, geography, communication and travel within a proposed electorate; and economic, social and regional interests.... which to me seems so vague as to give cover for literally any decision ("regional interests?" Hmm). One of the things the last 15 or so years of American politics have proven to me is that norms are not enough for setting standards by which politicians behave; instead you need explicit rules. And again, if such a committee amounted to in practice representatives of both major parties coming to an agreement, this process seems like a better way to formalize and standardize that process.
posted by en forme de poire at 11:29 AM on November 11, 2017 [1 favorite]
farlukar: ... and that still leaves you with two dominant parties?
Not really. Not nearly to the degree that it's the case in the US, anyway, where a 3rd party hasn't got more than a couple of seats on the national level for, what, a century? (I suspect I'm wrong about that. However, any third parties which did get national seats didn't last very long. In the US, the logic has applied: The two big parties expand just enough to absorb the closest third party so that they retain their chance of winning.)
The first time that we had a 3rd party get more than 10% of seats in a Canadian federal election was 1921. In the 28 national elections starting with that one, the top two parties have only gotten more than 90% of seats between them twice (1930 and 1958). (How many times has that happened in the US since 1921?) One of our major parties broke into three pieces and the rump was snuffed out (the PCs, in 1993) and then swallowed (by 2004). The NDP has gotten seats as a third party for decades. Even the Socreds got seats in every election for a 40 year stretch.
An interesting thing about the US is that it can't seem to generate persistent regional parties. That seems to be a regular source of lasting third parties in the other first-past-the-post systems.
On the other hand... really, we've had one major party. The Liberals ran Canada for almost as much of the 20th century as the Communists ran Russia.
posted by clawsoon at 12:19 PM on November 11, 2017 [2 favorites]
Not really. Not nearly to the degree that it's the case in the US, anyway, where a 3rd party hasn't got more than a couple of seats on the national level for, what, a century? (I suspect I'm wrong about that. However, any third parties which did get national seats didn't last very long. In the US, the logic has applied: The two big parties expand just enough to absorb the closest third party so that they retain their chance of winning.)
The first time that we had a 3rd party get more than 10% of seats in a Canadian federal election was 1921. In the 28 national elections starting with that one, the top two parties have only gotten more than 90% of seats between them twice (1930 and 1958). (How many times has that happened in the US since 1921?) One of our major parties broke into three pieces and the rump was snuffed out (the PCs, in 1993) and then swallowed (by 2004). The NDP has gotten seats as a third party for decades. Even the Socreds got seats in every election for a 40 year stretch.
An interesting thing about the US is that it can't seem to generate persistent regional parties. That seems to be a regular source of lasting third parties in the other first-past-the-post systems.
On the other hand... really, we've had one major party. The Liberals ran Canada for almost as much of the 20th century as the Communists ran Russia.
posted by clawsoon at 12:19 PM on November 11, 2017 [2 favorites]
I have a question about compactness-based algorithms. It seems like in the US, urban constituencies with common interests naturally form geographically compact units. However, rural constituencies with common interests are likely to be widely dispersed and irregularly shaped. It seems like it would actually be appropriate to create dense, compact districts for city dwellers, but big, weirdly shaped districts for rural people, so that they can have a representative. Can compactness-based algorithms accomplish this, or do they usually end up creating districts that split up the rural population and append a few rural people onto each urban district?
posted by agentofselection at 12:30 PM on November 11, 2017 [4 favorites]
posted by agentofselection at 12:30 PM on November 11, 2017 [4 favorites]
I think last year's election shows that people don't vote on the basis of much rationality. I'd rather work towards things like keeping money out of politics, and empowering people to know how to vote in their own best interest. We should be in a situation where the candidates who will do the best work for their constituents win, no matter how the districts are cut up. And hopefully that will roll up to states and the country also acting in the best interests of their residents.
posted by mantecol at 1:15 PM on November 11, 2017
posted by mantecol at 1:15 PM on November 11, 2017
Recent reviews have found that the same is true of every other first-past-the-post country other than the US.
Thanks for that link, though I would note that Duverger's law as outlined in Wikipedia excludes "double ballot systems" or run-offs, which Canada allows.
Also, as gerrymandering goes, the practice has been ordered by courts to achieve racially representative districts, and was never something to completely abandon, as many assume.
posted by Brian B. at 1:29 PM on November 11, 2017
Thanks for that link, though I would note that Duverger's law as outlined in Wikipedia excludes "double ballot systems" or run-offs, which Canada allows.
Also, as gerrymandering goes, the practice has been ordered by courts to achieve racially representative districts, and was never something to completely abandon, as many assume.
posted by Brian B. at 1:29 PM on November 11, 2017
Brian B.: Thanks for that link, though I would note that Duverger's law as outlined in Wikipedia excludes "double ballot systems" or run-offs, which Canada allows.
Interesting, though I'm a little confused: When has Canada had a national run-off election or double ballots?
posted by clawsoon at 1:43 PM on November 11, 2017
Interesting, though I'm a little confused: When has Canada had a national run-off election or double ballots?
posted by clawsoon at 1:43 PM on November 11, 2017
Thanks for that link, though I would note that Duverger's law as outlined in Wikipedia excludes "double ballot systems" or run-offs, which Canada allows.
The thing that reliably fucks up Duverger is just regionalism, like the BQ or SNP. Even apart from regional parties it doesn't do great given the long term survival of the Liberal Democrats in the UK. You'd be better off thinking that SMD plurality elections create a pressure towards fewer parties.
posted by GCU Sweet and Full of Grace at 1:54 PM on November 11, 2017 [1 favorite]
The thing that reliably fucks up Duverger is just regionalism, like the BQ or SNP. Even apart from regional parties it doesn't do great given the long term survival of the Liberal Democrats in the UK. You'd be better off thinking that SMD plurality elections create a pressure towards fewer parties.
posted by GCU Sweet and Full of Grace at 1:54 PM on November 11, 2017 [1 favorite]
The cake division problem can be extended to three or more players (for example, as explained by mathematician Hannah Fry), so there should be a three-or-more-party version of this solution.
Fundamentally, though, I do think there's a problem with the notion of having the elected parties pick their electorates, rather than the other way around, even if it can be done fairly. The electoral commission should get to select an algorithm that draws boundaries in a nonpartisan, transparent, and independently-assessable manner, not draw the boundaries themselves.
posted by biogeo at 2:04 PM on November 11, 2017 [2 favorites]
Fundamentally, though, I do think there's a problem with the notion of having the elected parties pick their electorates, rather than the other way around, even if it can be done fairly. The electoral commission should get to select an algorithm that draws boundaries in a nonpartisan, transparent, and independently-assessable manner, not draw the boundaries themselves.
posted by biogeo at 2:04 PM on November 11, 2017 [2 favorites]
When has Canada had a national run-off election or double ballots?
Nevermind, according to the link they only do it for parties, not federal elections.
posted by Brian B. at 2:20 PM on November 11, 2017 [1 favorite]
Nevermind, according to the link they only do it for parties, not federal elections.
posted by Brian B. at 2:20 PM on November 11, 2017 [1 favorite]
The idea that First Past The Post == Two Parties is so ingrained in the American mindset that when I first moved to the UK, I assumed they must have proportional representation. There are currently eight parties that have seats in the House of Commons (as well as five independents). And turns out, it's First Past The Post here, too.
posted by kyrademon at 4:34 PM on November 11, 2017 [2 favorites]
posted by kyrademon at 4:34 PM on November 11, 2017 [2 favorites]
No mention of the case before the Supreme Court this year regarding gerrymandering?
The idea that the most fair district is one that has a particular relationship between area and diameter (IIRC)? The greater the divergence, the more gerrymandered it is (due to the lengths gerrymanderers go to distort the natural bounds and thus an inefficient area). I can't remember the details, but something like that.
posted by symbioid at 4:49 PM on November 11, 2017 [1 favorite]
The idea that the most fair district is one that has a particular relationship between area and diameter (IIRC)? The greater the divergence, the more gerrymandered it is (due to the lengths gerrymanderers go to distort the natural bounds and thus an inefficient area). I can't remember the details, but something like that.
posted by symbioid at 4:49 PM on November 11, 2017 [1 favorite]
Have...have you considered an arms length agency? As I understand in Canada electoral ridings are designated by Elections Canada, and while people who want a popular vote system claim that there is gerrymandering, most of them are pretty much squares, whereas US election maps I've seen look like bugsplats.
posted by Canageek at 4:56 PM on November 11, 2017 [1 favorite]
posted by Canageek at 4:56 PM on November 11, 2017 [1 favorite]
Canada's not perfect by any means. I'm in a riding of 108,000. Meanwhile, PEI, population 146,000, has 4 ridings.
posted by Pruitt-Igoe at 8:34 PM on November 11, 2017 [2 favorites]
posted by Pruitt-Igoe at 8:34 PM on November 11, 2017 [2 favorites]
The page smoke linked says the following: The committee draws up a proposed redistribution that considers the current boundaries, geography, communication and travel within a proposed electorate; and economic, social and regional interests.
... which to me seems so vague as to give cover for literally any decision ("regional interests?" Hmm).
Regional interests mean interests in the regional areas. For example, we have just had a boundary change in my regional state electorate that now takes in a whole swathe of land to the west, and gives up areas to the east. The east's areas are urbanising and work better with electorates further east, while the predominance of my electorate has an agricultural aspect which better suits the bit added on the west.
The people determining these boundary changes live a thousand miles away, are public servants, and thus likely have no vested interest in creating a political environment more suitable for one party than the next, and with the checks and balances and scruitiny they have to go through to get a change passed, a deliberately biased redistribution is unlikely.
I don't think people living in America understand just how odd it is to people from Australia that Americans vote for people like head prosecutor, or school board, or fire chief or sheriff or whatever. You guys have so many built in opportunities for corruption and vote buying and spending time collecting support rather than just doing the job, that you can't envision a vastly less corrupt and more representative system.
posted by Thella at 9:32 PM on November 11, 2017 [5 favorites]
... which to me seems so vague as to give cover for literally any decision ("regional interests?" Hmm).
Regional interests mean interests in the regional areas. For example, we have just had a boundary change in my regional state electorate that now takes in a whole swathe of land to the west, and gives up areas to the east. The east's areas are urbanising and work better with electorates further east, while the predominance of my electorate has an agricultural aspect which better suits the bit added on the west.
The people determining these boundary changes live a thousand miles away, are public servants, and thus likely have no vested interest in creating a political environment more suitable for one party than the next, and with the checks and balances and scruitiny they have to go through to get a change passed, a deliberately biased redistribution is unlikely.
I don't think people living in America understand just how odd it is to people from Australia that Americans vote for people like head prosecutor, or school board, or fire chief or sheriff or whatever. You guys have so many built in opportunities for corruption and vote buying and spending time collecting support rather than just doing the job, that you can't envision a vastly less corrupt and more representative system.
posted by Thella at 9:32 PM on November 11, 2017 [5 favorites]
Commissions are one of our best ways to implement redistricting. There are a few flavors, including partisan and non-partisan ones. For states worried about removing elected officials from the process, Iowa's system is interesting: “an advisory commission drafts congressional and state legislative district boundaries. The state legislature retains final authority to implement district maps.”
Compact geometry is only ever valuable as a sign that something hinky is going on. A high Reock or Polsby-Popper score may be a hint to dig deeper, but is not on its own evidence of wrongdoing. As with the famous Illinois "earmuffs" district, you have to know the political geography to understand what's going on.
Right now, everyone thinking about this problem is on pins and needles waiting for a SCOTUS decision in Gill v. Whitford.
posted by migurski at 10:38 PM on November 11, 2017 [1 favorite]
Compact geometry is only ever valuable as a sign that something hinky is going on. A high Reock or Polsby-Popper score may be a hint to dig deeper, but is not on its own evidence of wrongdoing. As with the famous Illinois "earmuffs" district, you have to know the political geography to understand what's going on.
Right now, everyone thinking about this problem is on pins and needles waiting for a SCOTUS decision in Gill v. Whitford.
posted by migurski at 10:38 PM on November 11, 2017 [1 favorite]
The idea that the most fair district is one that has a particular relationship between area and diameter (IIRC)? The greater the divergence, the more gerrymandered it is
Not necessarily. In flat-terrain areas with the same population density (all urban, all rural, etc.), this might be true; in a city with a river running along one edge, or mountain towns, that's not the case. You don't get circles or squares that contain the same number of people, and it's ridiculous to group people by distance rather than by local communities - two towns a few miles apart may have similar politics, similar interests and needs; a town that's much closer to one of them - but on the other side of a line of hills - may have an entirely different economy.
Algorithms can give you a good starting point; they can't recognize which adjacent physical areas have people with the same interests. It needs a lot of hand-holding to figure out that a string of towns along the coast have similar needs; move a few miles inland and the politics are very different - so a long thin district might make sense.
What I would like to see is, "here's what the algorithm came up with, and here's our final version, and here's the report that explains what we changed and why those changes were made."
Party and racial gerrymandering would be a lot harder to hide - and if the explanations seemed fishy, they could be challenged in court, and the people who worked on it (and their aides) could be called to testify to find out if there was an agenda that the report didn't admit.
posted by ErisLordFreedom at 11:20 PM on November 11, 2017 [2 favorites]
Not necessarily. In flat-terrain areas with the same population density (all urban, all rural, etc.), this might be true; in a city with a river running along one edge, or mountain towns, that's not the case. You don't get circles or squares that contain the same number of people, and it's ridiculous to group people by distance rather than by local communities - two towns a few miles apart may have similar politics, similar interests and needs; a town that's much closer to one of them - but on the other side of a line of hills - may have an entirely different economy.
Algorithms can give you a good starting point; they can't recognize which adjacent physical areas have people with the same interests. It needs a lot of hand-holding to figure out that a string of towns along the coast have similar needs; move a few miles inland and the politics are very different - so a long thin district might make sense.
What I would like to see is, "here's what the algorithm came up with, and here's our final version, and here's the report that explains what we changed and why those changes were made."
Party and racial gerrymandering would be a lot harder to hide - and if the explanations seemed fishy, they could be challenged in court, and the people who worked on it (and their aides) could be called to testify to find out if there was an agenda that the report didn't admit.
posted by ErisLordFreedom at 11:20 PM on November 11, 2017 [2 favorites]
The US Congress has been responsible for enacting rules defining "electoral districts," apportioning of the number of congressional representatives to each of the states according to decennial census, and ultimately limiting the total number of congressional representatives in contravention of the US Constitution, which explictly excluded specific ethnic groups from calculation of a census.
Here is a summary of historical apportionment acts by the US Congress primarily provoked by population growth and admission of states. Each state has exercised constitutional authority to construct electoral districts. Until passage of the Civil Rights Acts of the 1960s persons disenfranchised by states' districting and other voter suppression laws that had little recourse to challenge violations the 14th Amendment of the US Constitution. Accordingly, the two dominant parties' rhetoric explaining districting design has shifted to appeasement or division (between them) of previously excluded ethnic groups.
Here are renderings of "optimized (and) compact" electoral districts that conform with the statutory limitation to 435 congressional representatives AND the "spirit" of Article II. Sec 2 which does not in fact recognize monopoly of two-party governance .
posted by marycatherine at 7:39 AM on November 12, 2017
Article I. Sec 2: The number of Representatives shall not exceed one for every thirty thousand, but each state shall have at least one Representative; The term "gerrymandering" dates to 1812, commemorating a patent districting bill in MA designed to give Gov. Gerry's party, Democratic-Republican Party (opposed to Federalist Party), numerical advantage in apportioning congressional representation. This event epitomizes the objective of districting designs within a state and historical collusion between parties to monopolize electoral control of popular preferences. At that time the franchise didn't even extend to all "white" males.
Here is a summary of historical apportionment acts by the US Congress primarily provoked by population growth and admission of states. Each state has exercised constitutional authority to construct electoral districts. Until passage of the Civil Rights Acts of the 1960s persons disenfranchised by states' districting and other voter suppression laws that had little recourse to challenge violations the 14th Amendment of the US Constitution. Accordingly, the two dominant parties' rhetoric explaining districting design has shifted to appeasement or division (between them) of previously excluded ethnic groups.
Here are renderings of "optimized (and) compact" electoral districts that conform with the statutory limitation to 435 congressional representatives AND the "spirit" of Article II. Sec 2 which does not in fact recognize monopoly of two-party governance .
posted by marycatherine at 7:39 AM on November 12, 2017
State Republicans agreed to it because they were so far out of power they felt it was their better choice,
No they didn't. The CA Republican party wanted redistricting, but the elected R legislators did not. This is why the ballot measure was required (one of the few ballots I have campaigned for in my life).
posted by benzenedream at 9:05 AM on November 12, 2017
No they didn't. The CA Republican party wanted redistricting, but the elected R legislators did not. This is why the ballot measure was required (one of the few ballots I have campaigned for in my life).
posted by benzenedream at 9:05 AM on November 12, 2017
Those optimized and compact districts are not a great approach to the issue of redistricting. One of the ways a political scientist I know defines gerrymandering is that it’s any privileging of a single concern above all others. A racial gerrymander focuses only on race while a partisan gerrymander accounts only for party. A purely-geometric district process is no different, it considers only abstract shape. Some political scientists jokingly refer to the geometry experts in this field as “the Cubists.”
It’s not enough to measure distances and areas, you have to know where people live and stay attuned to the various ways in which they divide themselves. Race was once the overarching concern in the South, but now party is in the driver’s seat. Per-district remedies for racial gerrymanders are different from whole-plan remedies for partisan ones.
posted by migurski at 10:34 AM on November 12, 2017 [3 favorites]
It’s not enough to measure distances and areas, you have to know where people live and stay attuned to the various ways in which they divide themselves. Race was once the overarching concern in the South, but now party is in the driver’s seat. Per-district remedies for racial gerrymanders are different from whole-plan remedies for partisan ones.
posted by migurski at 10:34 AM on November 12, 2017 [3 favorites]
Pruitt-Igoe: Yeah, blame that on old laws from when PEI joined confederation and was guaranteed a set number of seats. All of the Atlantic provinces get far more seats then they should.
posted by Canageek at 11:20 AM on November 12, 2017 [1 favorite]
posted by Canageek at 11:20 AM on November 12, 2017 [1 favorite]
« Older It is to laugh | An Extremely Close Reading of Pop Song Duets Newer »
This thread has been archived and is closed to new comments
posted by forgettable at 8:53 PM on November 10, 2017 [13 favorites]