• Welcome to the new Internet Infidels Discussion Board, formerly Talk Freethought.

Redistricting for the US House and the US state legislatures

West Virginia now has maps for Congress and its state legislature. Its two House districts divide the state roughly north-south.

Georgia Democrats have responded with their own map, which gives Democrats an additional seat, as opposed to the originally-proposed map, which gives Republicans an additional seat.
 
Or create electorates based solely upon numbers, not previous voting patterns?
What do you mean?

Looking at some of the maps (thank you for the work of linking) it seems to me that some of the maps are designed to maintain the status quo i.e. electorate X returned Demo/Rep/Ind at the last election and we will try to maintain that. Based upon population movements we will make the following adjustments to keep that result.

The mechanics of a voting system are to ensure that a number of voters (The number, N +/- %N, to be as consistent as possible in all electorates) have the opportunity to have their votes cast, counted and acknowledged as fairly and correctly as possible. It is not to maintain or attempt to cause a pre-ordinated outcome.
in post 37 you stated

I look at these maps and I think: why doesn't anyone ever think of multimember districts? That would make it easier to ensure that everybody will get some compatible Representative, even if only one of several in a district.
The attempt to maintain a previous outcome is the worse way to do a voting system.
Any system should not necessarily give voters what they desired rather it gives them the opportunity to vote.
 
West Virginia now has maps for Congress and its state legislature. Its two House districts divide the state roughly north-south.

Georgia Democrats have responded with their own map, which gives Democrats an additional seat, as opposed to the originally-proposed map, which gives Republicans an additional seat.

Therein lies a problem.
Political parties should not be drawing up electoral maps. That is the role of a an independent body, answerable to only the parliament, house or authority that raised it. Political parties can certainly critique the results and propose amendements but they do not decide. In Australia, I as a citizen, can also critique new electoral boundaries but it is the electoral commission, not i who finally decides.
Political parties will always try to maintain, at the very least, the status quo. That is one of the worse reasons to have an electoral system.
 
Texas now has a Congressional map. It's very gerrymandered, but Republicans didn't gain much in it. That is because the map was designed for incumbent protection, and to expand the Republicans' reach would have risked the electoral fortunes of several Republican incumbents. It strengthens Republicans in several seats, and it also strengthens Democrats in their seats.

In Illinois, the Democratic-dominated state legislature has released another map, on even more favorable to Democrats than the earlier one. It has a majority-Hispanic district near Chicago, and it shoves Reps. Sean Casten and Marie Newman into the same district.

Iowa now has a second proposed map, with district partisanship R+2, R+4, R+6, R+27.

What Redistricting Looks Like In Every State | FiveThirtyEight
 
Americans Don’t Trust Their Congressional Maps To Be Drawn Fairly. Can Anything Change That? | FiveThirtyEight
But Americans aren’t necessarily confident that the process will be a fair one. Just 16 percent of U.S. adults said they thought their states’ congressional maps would be drawn fairly, while 44 percent said they thought the maps would be drawn unfairly, per an August YouGov/Economist poll. Another 40 percent of adults said they were unsure if the maps will be fair. That might be one reason why independent commissions, which aim to empower ordinary citizens to draw map lines, have grown in use since the last redistricting cycle. In that same YouGov/Economist poll, 50 percent of Americans said they thought independent commissions should be responsible for the redistricting process in their state.

However, it’s unclear whether independent commissions will be enough to help build trust in the redistricting process. For some, the redistricting process is simply “the most political activity in American politics,” according to Michael Bitzer, a professor of politics and history at Catawba College and author of “Redistricting and Gerrymandering in North Carolina: Battlelines in the Tar Heel State.”

And Bitzer doesn’t see that changing anytime soon.

...
Redistricting has been a problem in a number of states, to the extent that the court has had to intervene. According to the nonpartisan redistricting website All About Redistricting, the courts rejected all or part of the maps in five states and drew new maps in 12 other states. And there is already one lawsuit pending this year.

...
For instance, a panel of federal judges threw out a partisan gerrymandering case in Wisconsin involving the state’s 2011 map following the Supreme Court’s 2019 decision, leaving that map in place. Wisconsin is now, of course, redrawing its map, but given that the government is divided — the governor is a Democrat, but Republicans control the state legislature — and there are already two lawsuits over the redistricting process, it’s possible that the courts will need to draw the lines if Democrats and Republicans can’t agree on a map.

It’s why in some states, voters are trying to address redistricting themselves in the form of independent redistricting commissions.
That article contains absolutely zero mention of the best solution IMO: proportional representation. The big problem here is disproportion, especially deliberate disproportion to favor one side or the other -- gerrymandering. PR is a very good cure for it, and it can also encourage alternatives to the D-R duopoly.

Many nations already use PR, so we can see how well it works in them.
 
From that 538 article:
ControlStatesDistricts
Republican20187
Democratic775
Split971
Independent commission896
One district66
(Total)50435
 
Alabama: Jan. 28, 2022 - Deadline for congressional candidates to file (therefore map should be set by this date) - Redistricting committee proposes a map.

Arizona: Jan. 1, 2022 - Date that candidates begin filing for congressional runs (therefore map should be set by this date) - Redistricting commission has preliminary approval of a map that would make 5 R's, 2 C's, and 2 D's.

Georgia: March 7, 2022 - Date that candidates begin filing for congressional runs (therefore map should be set by this date) - two maps, one proposed by Republicans and one by Democrats.

Illinois: March 7, 2022 - Date that candidates begin filing for congressional runs (therefore map should be set by this date) - the IL House and Senate have approved a map. Marie Newman decided to challenge Sean Casten rather than Chuy Garcia. R's Mike Bost and Mary Miller are now in the same S IL district, and the Springfield Strip, now somewhat D, is now an open seat.

Iowa: Feb. 28, 2022 - Date that candidates begin filing for congressional runs (therefore map should be set by this date) - now on its second map, with 2 R seats and 2 C seats.

Michigan: Dec. 31, 2021 - Date by which commission says it will enact a map - the redistricting commission has proposed 22 maps, and it is now down to 4.

New Mexico: Feb. 1, 2022 - Deadline for congressional candidates to file (therefore map should be set by this date) - one map is now a top pick, with 2 weak D and 1 C district.

North Carolina: Dec. 6, 2021 - Date that candidates begin filing for congressional runs (therefore map should be set by this date) - 10 maps so far.

Virginia: Oct. 25, 2021 - Deadline for commission to submit congressional map to legislature - Nov. 9, 2021 - Deadline for legislature to approve or reject congressional map - Nov. 23, 2021 - Deadline for commission to submit new map to legislature if initial map is rejected - Nov. 30, 2021 - Deadline for legislature to approve or reject new map if initial map is rejected - 7 maps so far, with the D and the R members of the redistricting commission at loggerheads.


C = competitive
 
Colorado now has a map. Joe Neguse got CO-02, a district extending NNW from Denver and D+32, and Lauren Boebert got CO-03, in the western and southern part of the state, and R+15. The state's new districts have partisan leanings much like its old districts, except for the district it gained, which is R+3.
  • Old: +47 +23 +15 +12 -12 -21 -22
  • New: +55 +32 +16 +6 -3 -15 -18 -26
 
Arkansas now has a map, though by the state's governor letting it become law instead of him signing the bill for it.

Iowa's legislaure has approved that second map.

Massachusetts: Feb. 15, 2022 - Date that candidates begin filing for congressional runs (therefore map should be set by this date)

The state's redistricting committee proposes a map for the state's 9 districts that is not much different from its present map. All of them are solidly Democratic, the weakest district being in Cape Cod, in the southeast part of the state, ad D+17. That suggests that disproportion can be due to lack of concentration, and the appropriate fix for that is proportional representation.

Montana: Nov. 14, 2021 - Deadline for commission to enact congressional map

The state's redistricting commission is down to two maps, both with a weakly Republican western strip and with a strongly Republican rest of the state.

Ohio: Nov. 30, 2021 - Deadline for legislature to pass temporary map if commission doesn't succeed

Now it's up to state legislators to decide on a map.

Oklahoma: April 13, 2022 - Date that candidates begin filing for congressional runs (therefore map should be set by this date)

Five strongly-Republican districts, with one district made more Republican by moving it northward and out of some of Oklahoma City.

Wisconsin: April 15, 2022 - Date that candidates begin filing for congressional runs (therefore map should be set by this date)

There's another issue. To quote 538:
There are no versions of the map that put Democrats on an equal footing to Republicans in purple Wisconsin, and that is in large part thanks to geography. Democrats are highly concentrated in Dane and Milwaukee counties, while Republicans aren’t as highly concentrated in any one part of the rest of the state. It simply would be very difficult to make Democrats competitive in an equal number of seats to Republicans without drawing funkier lines and breaking up municipalities, which the commission was directed not to do.
Proportional representation would fix this problem without drawing any weirdly-shaped districts.
 
Had a thought on a simple fix for gerrymandering:

Every political party with at least one seat submits a proposed map. The map with the lowest total boundary distance wins. Secret "bids".
 
Brian Olson's site has BDistricting -- "What redistricting is and what it could be." - 2010 Redistricting Results

"Below are district maps for US House and state legislatures that have been optimized for equal population and compactness only. No partisan power plays. No gerrymandering."

With House and state-legislature maps for all 50 states.

BDistricting - About

"What is a district for? ... Districts break down governance into managable pieces."

"What is a good district? ... Across all districts and all people, the best district map is the one where people have the lowest average distance to the center of their district. ... It's also possible for human drawn districts to actually better represent 'communities of interest' and other fuzzy but real sociological features."
 
Continuing with Brian Olson's "about" page, he takes on the issue of "Compactness Measures":

  • Travel time?
  • District perimeter/area
  • District convex-hull area / precise area
  • Just pick simple boundaries along zip-codes/counties/rivers/etc.
Also concerns like
  • Will this break up communities?
  • Won't this disenfranchise minorities?
  • What about competitive districts?
He proposes proportional representation as a solution. I agree with him that it's better than contorted districts.
How does your solver work?

My implementation is a heuristic based gradient descent solver with simulated annealing jitter. It looks at the boundaries between districts and tries to make things better by flipping one block from district A to district B (and possibly over some number of steps, other blocks from B to C and C to A). It doesn't actually directly optimize the measure of population compactness, but looks at related measures like the ratio of the block's distance to the average edge blocks' distances from each district's center, and the ratio of the populations of the two districts the block might go into. Each district grabs up to one block, then centers are recalculated and the cycle begins again checking all the edge blocks.
 
RangeVoting.org - Splitline districtings of all 50 states + DC + PR with the splitline algorithm described in RangeVoting.org - Gerrymandering and a cure - shortest splitline algorithm
The algorithm:
  1. Start with the boundary outline of the state.
  2. Let N=A+B where A and B are as nearly equal whole numbers as possible.
    (For example, 7=4+3. More precisely, A = ⌈N/2⌉, B=⌊N/2⌋.)
  3. Among all possible dividing lines that split the state into two parts with population ratio A:B, choose the shortest. (Notes: since the Earth is round, when we say "line" we more precisely mean "great circle." If there is an exact length-tie for "shortest" then break that tie by using the line closest to North-South orientation, and if it's still a tie, then use the Westernmost of the tied dividing lines. "Length" means distance between the two furthest-apart points on the line, that both lie within the district being split.)
  4. We now have two hemi-states, each to contain a specified number (namely A and B) of districts. Handle them recursively via the same splitting procedure.
 
Auto-Redistrict - the algorithm tries to optimize districts by:
  • Geometry
    • Equal population
    • Contiguous
    • Compact
    • Minimal county / municipality splits
  • Equality
    • Competitive
    • Proportional
    • Minimal partisan gerrymandering (maximal "partisan symmetry")
    • Minimal racial gerrymandering (maximal voting power equality)
It uses a genetic algorithm:
  1. Initialize. First, the population of potential maps is initialized. It could either start off totally random, or you could use the current electoral district shapes to start it off.
  2. Evaluate. Then the fitness of each map is evaluated on each of the criteria (in our case, compactness, equal population, competitiveness, proportional representation, etc.)
  3. Select. The best scoring maps are selected for reproduction,
  4. Recombine. and randomly recombined to form new maps, that are hybrids of the best maps.
  5. Mutate. Finally these maps are "mutated" slightly so that other potential maps that are similar to them are explored.
  6. (Repeat). This is the new population of potential maps. The process repeats from step 2.
The page then goes into more detail about the optimization criteria.

I could not find anything on the algorithm's results or how long it takes to run.
 
Had a thought on a simple fix for gerrymandering:

Every political party with at least one seat submits a proposed map. The map with the lowest total boundary distance wins. Secret "bids".
Or have an independent body produce the map. A body that has no ties to any pollie or party, that stands to gain nothing from an electoral map. the parties/polles are free to critique said body's map but they them selves do not generate a map.
Having pollies generate a map means they will aim to maintain, at the very least, the status quo, thus guaranteeing gerrymandering will continue.
 
uses a genetic algorithm:
  1. Initialize. First, the population of potential maps is initialized. It could either start off totally random, or you could use the current electoral district shapes to start it off.
  2. Evaluate. Then the fitness of each map is evaluated on each of the criteria (in our case, compactness, equal population, competitiveness, proportional representation, etc.)
  3. Select. The best scoring maps are selected for reproduction,
  4. Recombine. and randomly recombined to form new maps, that are hybrids of the best maps.
  5. Mutate. Finally these maps are "mutated" slightly so that other potential maps that are similar to them are explored.
  6. (Repeat). This is the new population of potential maps. The process repeats from step 2.
The page then goes into more detail about the optimization criteria.

I could not find anything on the algorithm's results or how long it takes to run.

Genetic algorithms aren't a good thing here--they are not reproducable so there's no way to figure out if somebody tampered with the results a bit. Redistricting should be done by a reproducible means.
 
Had a thought on a simple fix for gerrymandering:

Every political party with at least one seat submits a proposed map. The map with the lowest total boundary distance wins. Secret "bids".
Or have an independent body produce the map. A body that has no ties to any pollie or party, that stands to gain nothing from an electoral map. the parties/polles are free to critique said body's map but they them selves do not generate a map.
Having pollies generate a map means they will aim to maintain, at the very least, the status quo, thus guaranteeing gerrymandering will continue.

But how do you ensure they are actually independent?
 
Genetic algorithms aren't a good thing here--they are not reproducable so there's no way to figure out if somebody tampered with the results a bit. Redistricting should be done by a reproducible means.
That's a problem with stochastic algorithms in general, like simulated annealing. A deterministic alternative is gradient descent, with avoiding the most recently-used change directions: "tabu search". Another deterministic alternative is K-means clustering, though it cannot optimize features like district perimeters. All these algorithms need initial conditions, and those are usually calculated randomly.

Brian Olson gets around the randomness part by doing repeated runs with different random-number values, and one can also do so by specifying the random-number algorithm and the seeding.
 
One may want to use some programming language's standard-library function for random numbers, but that has a problem. It may not be very well-documented. If one uses an open-source implementation like GCC or NumPy, one may be able to find the algorithm if one burrows through the source code in some archive somewhere. But that's not very feasible with a proprietary implementation.

In fairness, C++11 introduced some well-documented random-number generators in its standard library.

This is much more of a problem with random numbers than it is for (say) square root, log, exp, or trig functions, because different algorithms produce very different sequences of numbers.


Back to redistricting.

Alabama now has a map. Its 7 districts are one strongly Democratic one that includes the state's biggest cities, Birmingham and Montgomery, and 6 strongly Republican ones for the rest of the state.

538 is confused about North Carolina. Does it have a map? Or is its new map still in the final stages of legislative approval?

Michigan: Dec. 31, 2021 - Date by which commission says it will enact a map - now at 3 maps, named "Apple", "Birch", "Chestnut". What next? Let's see ... "Dogwood", "Elm", "Fir", "Ginkgo", "Hickory", "Ivy", "Juniper", "Kapok", "Larch", "Maple", "Nut", "Oak", "Pine", "Quercus", "Redwood", "Spruce", "Tamarack", U? V? W? X? "Yew" Z?

New Hampshire: Nov. 18, 2021 - Deadline for House Special Committee on Redistricting to propose draft of congressional map - Democrats want both districts to be neck-and-neck, while Republicans want one Democratic-leaning district and one Republican-leaning district.

Ohio: Nov. 30, 2021 - Deadline for legislature to pass temporary map if commission doesn't succeed - a third map, from State Senate Republicans, even more gerrymandered than the State House Republicans' map.
 
Had a thought on a simple fix for gerrymandering:

Every political party with at least one seat submits a proposed map. The map with the lowest total boundary distance wins. Secret "bids".
Or have an independent body produce the map. A body that has no ties to any pollie or party, that stands to gain nothing from an electoral map. the parties/polles are free to critique said body's map but they them selves do not generate a map.
Having pollies generate a map means they will aim to maintain, at the very least, the status quo, thus guaranteeing gerrymandering will continue.

But how do you ensure they are actually independent?
Fortunately in Australia our Electoral Commissions, both State and Commonwealth, have been in existence since just after WW1. They have a near perfect record of impartiality and competence. I say near perfect as this century we have had a few examples of carelessness in handling ballots, necessitating new elections or rather part elections.
The way you Yanks divide yourself into opposing factions so quickly would make it difficult but you have to start somewhere.
As a start the body is answerable only to the parliament or legislative body that formed it, not the government of the day. No existing, former or wanna-be politicians can be members. No members can belong to a political party or have worked for a political party or candidate, paid or unpaid.
In the state of Victoria where I live you cannot work for the Electoral Commission if you have represented or worked for a political party or candidate within the last 5 years or 2 electoral cycles, whichever is longer. You cannot even hand out pamphlets or flyers for a party or candidate. You cannot put a political sign on your property including a vehicle as your impartiality will be called into question. Paid or volunteer work does not matter. Similar rules apply for the Commonwealth Electoral Commission.
Members are paid by the parliament, not the parties or others. They are considered to be public servants and follow the public servant rules.
They produce the maps and then put them up for public discussion for a time. These maps are based upon population only. The maps merely show the boundaries of the electorates, not where the polling stations etc. will be. That is a separate process. All interested parties including political parties, candidates, individuals etc. can lodge a request or protest to amend an electorate based solely upon population. Any changes are made based solely upon population and are made by the commission. Naturally the perceived independence of the commission is important.

There are 2 important features
1. The body is independent of the parties, candidates, government of the day etc. Responsible only to the parliament that created it.
2. Boundaries are based solely upon population. If another criteria is allowed esp. past voting patterns then gerrymandering will occur.
 
Back
Top Bottom