Redistricting Commissions and Referenda

[jimrtex]

There needs to be a referendum. I can't think of an uglier gerrymander in recent memory. Even the 2001 districts were much better than this.

I could get behind referenda in both Ohio and Maryland on their horrible maps. I can't agree with unilateral disarmament but we need some kind of counter-weight to the legislative impulse to do the worst they can get away with. It doesn't matter if it's a single-party gerrymander or a bipartisan protection scheme.

Maybe a constitutional amendment establishing a national redistricting commission like other countries?

Congress has the authority to redistrict or provide for the manner in which it is conducted.  No need for a constitutional amendment.

[jimrtex]

Divide the State into atomic districts, 16384 x Number of congressional districts, so each atomic district would have around 40 persons.  Appoint 3 commissioners for each atomic district, chosen by lot from the voter rolls.

Districts would be formed by consolidation of districts on a pairwise basis in rounds.  So in the first round pairs of atomic districts would be merged based on agreement of the commissioners of the districts.  Some of the commissioners would be chosen by lot to represent the merged district.

Repeat for 14 rounds and you have your congressional districts.

[jimrtex]

Divide the State into atomic districts, 16384 x Number of congressional districts, so each atomic district would have around 40 persons.  Appoint 3 commissioners for each atomic district, chosen by lot from the voter rolls.

Districts would be formed by consolidation of districts on a pairwise basis in rounds.  So in the first round pairs of atomic districts would be merged based on agreement of the commissioners of the districts.  Some of the commissioners would be chosen by lot to represent the merged district.

Repeat for 14 rounds and you have your congressional districts.

1) Best solution I've read yet.

2) The initial districting of people into groups of forty is still gameable.

How is it gameable?

[jimrtex]

Divide the State into atomic districts, 16384 x Number of congressional districts, so each atomic district would have around 40 persons.  Appoint 3 commissioners for each atomic district, chosen by lot from the voter rolls.

Districts would be formed by consolidation of districts on a pairwise basis in rounds.  So in the first round pairs of atomic districts would be merged based on agreement of the commissioners of the districts.  Some of the commissioners would be chosen by lot to represent the merged district.

Repeat for 14 rounds and you have your congressional districts.

1) Best solution I've read yet.

2) The initial districting of people into groups of forty is still gameable.

How is it gameable?

There is a street. On one side everyone is Black, and on the other side, everyone is White. In one distribution the street is the dividing line. In another, blocks are the units, taking folks from both sides of the street. If Whites prefer to aggregated with Whites, and Blacks with Blacks, in the first case the street will be a dividing line, while in the second, the aggregation will occur along the street as neither the Black or White groupings will want to aggregate with the street people. Either initial distribution creates its own logic of aggregation.

So if there is a city with 1.4 million persons, and all the Blacks live east of Grey Street, and all the Whites live to the west of Grey street, but equal numbers of Blacks and Whites live on opposite sides of every block of Grey.

So in one case, the Blacks who live between Grey and Charcoal will coalesce with Blacks further to the east, and the Whites who live between White and Beige will coalesce with Whites to the west.

But if the atomic districts cross Gray, they will aggregate north south along Grey.  What happens when that aggregation is finished and the district with 2560 persons along Gray is completed.  Who will they aggregate with then?

[jimrtex]

That's one of the problems with your proposed system. Early in the process, the mergers won't be particularly problematic. As the districts congel, coercion will become increasing the norm.

It is not a problem.  You exclaimed that it was gameable and had to contrive an example, and there was no problem dealing even with your contrived example.

There is no such word as "congel".  You mean coalesce or aggregate (verb).

Suppose at the tip of the Panhandle the districts had coalesced to something like [New Mexico to the West, and Oklahoma to the North and East],

AAAA
BBCC
BBCC
DDDD

Since A is separated from the rest of Texas by B and C, it must merge with one or the other. If B and C want to merge with each other, or D, coercion must occur. If A and D both want to merge with B, but not C, coercion must occur.

As your districts coalesce coercion will be increasingly necessary to maintain contiguous subdistricts, while pairing every district.

There would be constraints on choices of partners.

For example the commissioners for a district could have to choose among neighboring districts:

1) Adjacent district
1a) Adjacent district in the same city.
1b) Adjacent district in the same county.
1e) Adjacent district in another county.

Within larger cities, there might be constraints based on neighborhoods.

2) Nearby districts:
2a) Districts in the same city.
2b) Districts in the same county.
2c) Nearby districts based on relative districts (eg 10% nearest districts).

Each set of commissioner's would rank their preferences among adjacent districts.  Where there was agreement, the districts would be merged.   Those who did not find a partner would try again.  In the early rounds it would not matter that the districts were not contiguous.

And at the end, districts could be fixed up by tunneling.

[jimrtex]

Divide the State into atomic districts, 16384 x Number of congressional districts, so each atomic district would have around 40 persons.  Appoint 3 commissioners for each atomic district, chosen by lot from the voter rolls.

Districts would be formed by consolidation of districts on a pairwise basis in rounds.  So in the first round pairs of atomic districts would be merged based on agreement of the commissioners of the districts.  Some of the commissioners would be chosen by lot to represent the merged district.

Repeat for 14 rounds and you have your congressional districts.

This seems pretty fun, even if possibly too expensive and still potentially prone to abuse.

Rather than actually creating all the atomic districts, I would apportion them to somewhat larger areas, perhaps election precincts.  So instead of dividing a precinct with 1260 persons into 30 areas of 42 persons tiling the area, I would give it 30 districts covering the whole area, that would form a stack of districts.  Each atomic district would represent a 1/30th interest in the precinct, rather than a specific area of the precinct.

Then, much of the initial mergers would simply be automated combinations of districts in each precinct's stack.  If there were an odd number of districts, one of them would have to merge with districts from other precincts.  And overlapping districts would be constrained toward merging with each other to eliminate overlaps.

This would reduce much of the cost in the early rounds.

[jimrtex]

There would be constraints on choices of partners.

That's my point. Your claim was, "So in the first round pairs of atomic districts would be merged based on agreement of the commissioners of the districts."

Agreements often are made within constraints.  The simple fact that pairs of atomic districts are merged, is a constraint.

You must be logged in to read this quote.

Or, my point, they may be boxed in a way that they will forced to pair with one group regardless of their wishes. The system wouldn't be as voluntary as you describe it.

Agreement doesn't necessarily imply voluntary.  You might agree to purchase a used car because you don't want to walk 5 miles to work and that is all you can afford, even though your wish would be to have the Maserati you saw in a magazine.

[jimrtex]

Agreement doesn't necessarily imply voluntary.  You might agree to purchase a used car because you don't want to walk 5 miles to work and that is all you can afford, even though your wish would be to have the Maserati you saw in a magazine.

And, my analogy would be that you do want the Maserati that you can't afford, but, have the money to buy a Ford or a Chevy. You would chose the Chevy, but, you are told to buy the Ford because someone else can afford the Chevy, but, not the Ford. Now, your buying of the Ford wouldn't be all that voluntary. In the same way, some masters over the process are going to have to force, or forbid, certain pairings to assure that every group is successfully paired at every iteration. That will give the overseers certain powers. What are your controls to stop the overseers from abusing their power?

Your plan does have the avantage that small towns will in all probability coalese earlier than larger cities could possibily do so.

There was one Ford and one Chevy.  The seller of the Chevy liked the other buyer more.  He didn't seem to like your negotiating style.  Or maybe it was caprice.  You bought the Ford.

The number of atomic districts was set to a power of 2 for a reason: so that every district can always be paired at every iteration.

The constraints would be rule-based.  No reason for the administrators to abuse them.

[jimrtex]

Suppose Presidio, Jeff Davis and Culbertson counties coalese in the early rounds, and aggregate togeather soon after. What then? Unless the number of groups to the West C-JD-P group boundary is an exact power of two, the process will fail [assuming contigious pairings].

El Paso would have 18,781 atomic districts; Hudspeth 82; Culberson (no t) 56; Jeff Davis 55; and Presidio 183.

Run through a few rounds and explain where it will fail.

[jimrtex]

The constraints would be rule-based.  No reason for the administrators to abuse them.

Suppose Presidio, Jeff Davis and Culbertson counties coalese in the early rounds, and aggregate togeather soon after. What then? Unless the number of groups to the West C-JD-P group boundary is an exact power of two, the process will fail [assuming contigious pairings].

This is one reason why it is difficult for computer programs to automatically draw districts. Mapping algorithms start with seed areas, but generally get stuck when an area is cut off that contains a population not equal to a whole number of districts. Programs then have to iterate by unbuilding one or more districts and try to rebuild such that equal population is restored. There are usually several such iterations, and in the process rules that governed the original construction are weakened or bypassed.

In the system that I have proposed, the intermediate districts form a network, in which it is possible to determine distance (in nodes) between two separated parts of a district.  The isolated parts would choose an adjacent district along a shortest path, which would produce a new set of isolated districts which are closer together.  Repeat until they are next to each other.

[jimrtex]

Suppose Presidio, Jeff Davis and Culbertson counties coalesce in the early rounds, and aggregate together soon after. What then? Unless the number of groups to the West C-JD-P group boundary is an exact power of two, the process will fail [assuming contigious pairings].

El Paso would have 18,781 atomic districts; Hudspeth 82; Culberson (no t) 56; Jeff Davis 55; and Presidio 183.

Run through a few rounds and explain where it will fail.

After six rounds, Jeff Davis and Culberson coalesce because people will want their county not to be split in redistricting. After round seven, Hudspeth coalesces, and Jeff Davis and Culberson merge. In the eight round, Presido has coalesced, and Hudspeth merges with Jeff Davis/Culberson. In round nine, Presido merges with Hudspeth/Jeff Davis/Culberson. In the tenth round, and after, the Presido/Hudspeth/Jeff Davis/Culberson want to merge with blocks to the East. After eleven rounds, there would about 17 blocks West of the P/H/JD/C block, 16 of which will eventually coalesce to form a district, and a seventeenth that would have to merge with districts to the East. That would be the P-H0JD-C block whether it wanted to merge or not.

It is a little trickier than that. After the four counties cut off El Paso from the rest of the state, the detached area would be one, and some fraction, of a Congressional district expressed in binary format.  When that digit is one, the P-H-JD-C block would have to merge to the West, while if it is zero, it would have to merge to the East.  That isn't a voluntary system.

Why would the Trans Pecos prefer to merge in a way that would place them in a district dominated by San Antonio, rather than have more El Pasoans?

And why does it matter if it was voluntary or not?