Per SCOTUS, initiative created redistricting commissions may be l'histoire

[muon2]

There is a simple fix available to states with independent commissions. Follow IA's lead and send plans to the legislature for an up or down vote. In IA the bureau that drafts the bills is given the task of drawing the legislative and congressional maps following statutory criteria. The bureau sends the plans to the legislature for an up or down vote without amendment. If the plan fails it goes back to bureau for another try, perhaps guided by comments the legislature is permitted to send with their rejection. The process is repeated up to three times. If the third plan is rejected, then the matter goes to the courts.

In commission states, just replace the bill drafting bureau with the commission, but otherwise follow that procedure. The legislature stays in the process with the final say on the map.

[muon2]

There is a simple fix available to states with independent commissions. Follow IA's lead and send plans to the legislature for an up or down vote. In IA the bureau that drafts the bills is given the task of drawing the legislative and congressional maps following statutory criteria. The bureau sends the plans to the legislature for an up or down vote without amendment. If the plan fails it goes back to bureau for another try, perhaps guided by comments the legislature is permitted to send with their rejection. The process is repeated up to three times. If the third plan is rejected, then the matter goes to the courts.

In commission states, just replace the bill drafting bureau with the commission, but otherwise follow that procedure. The legislature stays in the process with the final say on the map.

But in Iowa, didn't the legislature create the structure in the first instance? I am not sure an initiative based law not passed by the legislature, that just has the legislature involved at the end with a mere veto power to send it to the courts, will pass muster, assuming SCOTUS strikes down the AZ structure.

I think that will be a question many will be looking at the opinion to determine. There are lots of initiative-based election laws in states that could be impacted based on the questions from the liberal justices. If the ruling stays narrow to the act of redistricting and overturns AZ, will it be because it was by initiative or because it was a plan without legislative approval. The legislature need not be the sole approver of the plan since most states require the Gov's signature on the redistricting bill. It seems to me that initiatives may still be able to dictate the manner in which the legislature performs its redistricting function, even if they can't strip that function entirely.

[muon2]

When you are drawing those DRA maps for a Dem gerry, I would ignore the 2008 pres numbers. The Dems would certainly want to be no worse than Brown in 2010 to protect against midterm defeats.

[muon2]

When you are drawing those DRA maps for a Dem gerry, I would ignore the 2008 pres numbers. The Dems would certainly want to be no worse than Brown in 2010 to protect against midterm defeats.

2010 was not a mere midterm, but a BIG Republican victory year. Even then many House incumbents performed reasonably well: they were incumbents. Depends on what the objective is: if it is to build a firewall, you may be right. But if the objective is to win most seats most of the time, one can take certain risks: and hope that incumbents would know their districts.

IL in 2011 was a good example of two maps drawn with different data sets. The state house map was based on mid-term election data in 2010 and the supermajority gains in 2012 were held in 2014 with no losses, despite a big win for the Pubs for Gov. OTOH the congressional map was designed for the big win assuming 2012 would repeat 2008 in the state. The goal was 13 of 18 seats, but that gamble did not pay off and after 2014 the Dems only have 10 seats. That's less than they could have had if they played it like the legislative map and had drawn 11 safe seats.

[muon2]

When you are drawing those DRA maps for a Dem gerry, I would ignore the 2008 pres numbers. The Dems would certainly want to be no worse than Brown in 2010 to protect against midterm defeats.

All of the Democratic districts are Brown 2010 districts. The two swing districts are Obama/Whitman districts although Obama won both districts in 2012 and Brown won them in 2014.

It was worth stating since the one district in your maps that had numbers (CD 48), was an Obama/Whitman district.

[muon2]

However, I am not sure kraxen's map is legal under the VRA. Creating an ersatz Hispanic CD like that, losing Hispanics on the perimeters in various locations to take in the core in disparate locations, might be viewed as racial gerrymandering ala that NC map, where SCOTUS axed that Watt CD back when. It may have taken erosity a bit too far. The point being that if an Hispanic (black) CD can be created that is compact, is it legal to make one hideously erose, going all over the state? Interesting question. Maybe Muon2 has a thought on it. It's one thing to create a majority minority CD that can only be done by going all over the place (not mandated of course, but legal); quite another when it is unnecessary to do so, to create the requisite majority minority CD.

I think the current VA case touches on that very issue. The lower court, citing other cases, said that VA-3 was drawn with race as the primary factor. The unusual shape including the hopscotch down the river led to that conclusion, and it was recognized that there were other districts that could provide an opportunity for the minority to elect a candidate of choice without the gerrymander.

At the other extreme is a district like IL-4. It was originally created in 1991 and the bizarre shape was upheld, since to directly connect the two Hispanic neighborhoods would have bisected a black majority CD. The operating conclusion was that an unusual shape for a VRA district had to be strongly justified by the state - basically that the burden is on the state to show that a strange-shaped district was the only means to comport with section 2. That concept applied in part in LULAC (TX), where the remade Hispanic CD could not be justified and the map was overturned.

I think Krazen's map could be easily attacked and would be hard to defend.

I found this map that I put together in Nov 2011 during the AZ commission debate.


I wasn't using political data when this was designed, only putting in two Hispanic CDs. It ends up with 6 firm R, 1 lean R and 2 firm D districts.

Also, my districts have an average deviation of only 66, an inequality score of 2. I'm not sure how to calculate erosity in AZ yet.

[muon2]

However, I am not sure kraxen's map is legal under the VRA. Creating an ersatz Hispanic CD like that, losing Hispanics on the perimeters in various locations to take in the core in disparate locations, might be viewed as racial gerrymandering ala that NC map, where SCOTUS axed that Watt CD back when. It may have taken erosity a bit too far. The point being that if an Hispanic (black) CD can be created that is compact, is it legal to make one hideously erose, going all over the state? Interesting question. Maybe Muon2 has a thought on it. It's one thing to create a majority minority CD that can only be done by going all over the place (not mandated of course, but legal); quite another when it is unnecessary to do so, to create the requisite majority minority CD.

I think the current VA case touches on that very issue. The lower court, citing other cases, said that VA-3 was drawn with race as the primary factor. The unusual shape including the hopscotch down the river led to that conclusion, and it was recognized that there were other districts that could provide an opportunity for the minority to elect a candidate of choice without the gerrymander.

At the other extreme is a district like IL-4. It was originally created in 1991 and the bizarre shape was upheld, since to directly connect the two Hispanic neighborhoods would have bisected a black majority CD. The operating conclusion was that an unusual shape for a VRA district had to be strongly justified by the state - basically that the burden is on the state to show that a strange-shaped district was the only means to comport with section 2. That concept applied in part in LULAC (TX), where the remade Hispanic CD could not be justified and the map was overturned.

I think Krazen's map could be easily attacked and would be hard to defend.

I found this map that I put together in Nov 2011 during the AZ commission debate.


I wasn't using political data when this was designed, only putting in two Hispanic CDs. It ends up with 6 firm R, 1 lean R and 2 firm D districts.

Also, my districts have an average deviation of only 66, an inequality score of 2. I'm not sure how to calculate erosity in AZ yet.

Since the second Hispanic CD is not legally mandated, your map and mine violate a ton of the rules that we have evolved and which we agree upon.   What should be done is to draw a compact Tucson CD that does not chop Tucson. Drawing the second Hispanic CD creates a ton of chops, and massive erosity. It was only done because it had been done before. I doubt either of us would have gone there absent that.

No doubt we have evolved in thoughts about mapmaking. To apply our current metrics to AZ would require a bit of work to identify county subdivisions and determine how to treat reservations and connections between counties and subdivisions.

[muon2]

The real problem of course, is this non legally mandated ersatz second Hispanic CD, that violates just about every rule in the book per Muon2’s little “rules.” And yes, Tucson is a Dem town, and needs and deserves a Dem CD. It just should not be gerrymandered to create a Dem CD, and a tilt Dem CD. In fact, if one abandons that unfortunate frolic and detour, suddenly one is able to create a map which I suspect would score very high (might be hard indeed to score higher) per Muon2’s metrics. This assumes that chops to accommodate and keep together Indian Reservations don’t count as chops or anything else, and one pretends that they are just not there (which I think is appropriate). Given that exception, the map below I think can hardly be described as  a Pub gerrymander, and would score very high in a map contest, using the rules that we have evolved. The chop by AZ-02 into Apache County is a microchop by the way, so it does not count.
 
 

Oh, for desert, I offer up a legal 8-1 Pub gerrymander (in Pima, AZ-03 takes the Hispanics, and AZ-02 and AZ-05 split up the more troublesome psephologically speaking for Pubs, white liberal zone), that I think masks the mischief reasonably well. It cuts down AZ-05 to 53.7% McCain, which is a tad tight, but it should hold and be just beyond dummymander territory. AZ-02 is 53.5% McCain, so it moves about 3 points in the Pub direction, and the former Hispanic CD, AZ-03 is now 52.9% McCain. The current incumbent will not be able to win there, and it would be in play, but should have a Pub tilt. One really cannot do much better I think, without a ridiculously looking map ala sbane’s most creative efforts at Dem gerrymandering in CA. The problem is that it is next to impossible to do a quad-chop of Pima County, which is needed for a safer 8-1 map (absent krazen’s probably illegal effort to create an erose Hispanic CD, that packs all the Dems in), given the huge precincts in area over the empty desert, and the Indian Reservation blockage issue. One does not want to mess with Native Americans in AZ. That is a bridge too far, as both parties acknowledge.

Anyway, the AZ Pubs can send this sort of map to the CA Dems as a bargaining chip perhaps, if SCOTUS axes the AZ commission’s map. It would be far preferable to just use the higher scoring map per Muon2’s metrics instead however, which I think would be hard to characterize as any sort of gerrymander at all, much less a slash and burn one. Thus if the CA Dems go there, it would hardly be a symmetrical retaliation.

Here are a couple of tidbits about the application of muon2 rules to AZ.

Yuma and Pima aren't connected, so the first plan above would be disallowed.

There are lots of UCCs in AZ. The big one is Phoenix with Maricopa and Pinal. Together they make up 5.9 CDs. The cover is 6 and the pack is 5, making the first plan above a great candidate since it incurs no penalty there, except for the connection problem.

The Tucson UCC is a single county one, but it is larger than a CD so there must be a whole CD nested within the county to avoid a UCC pack penalty. When we discussed dropping the penalty for single county UCCs, I don't think that we intended to avoid a large single county UCC like Pima.

Yavapai, Movave, Yuma, Coconino, and Cochise are all small single county UCCs. Torie hasn't commented on the first round of scoring in the MI thread, but I'm leaning towards not penalizing single county UCCs with less than a macrochop. If so macrochops should be specifically avoided except in Maricopa and Pima where they must exist. Of course they probably should be avoided for erosity reasons, too.

I like the idea of Torie's reservation exemption, which I might describe as reservations acting as extra counties independent of the statutory counties. They are connected if there are local connections to the counties that the reservations overlap.

With all that the best way to avoid macrochops and UCC penalties is to link Maricopa, Pinal, and Gila extended to the reservations that overlap Gila and fill that with 6 CDs. That leaves a donut around the state with 3 CDs and Yuma and Pima can't connect.



I'm think that some UCC penalties might be preferable to this CD 3.

[muon2]

I don't agree with a rule that you cannot connect counties connected by land mass with no state highway connection I don't think. Yes, if you sever them, it helps with the erosity score, since nothing is severed, but I think that is about it. It can be fixed by doing another county chop, but why? At worst, you could count it as another chop, but to me, it just seems wrong. Keeping counties together is a good thing, not something to be thwarted by some hard and fast rule, that just says no.

You were a proponent of that rule when it came to mountains. It made sense in CA and WA. We use that rule to avoid connecting counties in the Midwest that adjoin by only a small segment with no highway, and that seemed fine in MI, your maps included. It also prevents leaping bodies of water with no bridge or ferry within a state to connect counties. In VA it allows one to determine where one can cross the James and Chesapeake, and distance across the water doesn't always help given the long bridge that links the Delmarva.

You can't ditch the rule just because it forces a UCC penalty in AZ to avoid an erose mess. It's just another example of local geography constraining the map.

edit: One of the tools in the gerrymanderers kit is to connect areas with no road connections. Some states even have language requiring that it be convenient to travel between areas in a district to block that type of district. One map expert I spoke to thought it would be one of the most useful anti-gerrymandering rules one could have. That's why I use it here.

[muon2]

Well, “hideous” is too strong a word for Train’s modest Dem gerrymander. My objections to his map (he requested elaboration, and I apologize in being tardy in providing it) are mostly aesthetic.  I note them on his map below

That little hook between 6 and 9 is unfortunate; I'd like to fix it.  I don't think there's anything wrong with my District 1, though: it's more of an east-west line (modulo drawing around reservations and such) and Cochise doesn't really fit any better with Tucson IMO.  I'd argue it's nicer-looking and better-connected than your District 2, as well.

Your way of dealing with reservations is of course correct, I think we can be unanimous on that point.

And, yes, everything depends on whether Grijalva's district is protected/encouraged/mandated or not. I suspect that we are better off keeping it.

Is keeping it a state choice, like in Bartlett where NC would be permitted to draw a non-mandated district? Do you think there's some mandate given the 24% Latino CVAP in AZ? It would be hard to make the case that Yuma and Tucson are close enough for Gingles. If the state eliminated it I assume there would be a suit to preserve it, but could the suit prevail?

[muon2]

We will disagree on the highway business. I think it is about far more than CoI as I noted in my comment. Connectivity is a vital notion that goes beyond contiguity and the erosity measure was a bonus for me. I will will stick to the principle that one should be able to conveniently travel between significant populations within a district without going through other districts. I have heard it espoused enough in other venues that I remain confident that it can be sold to policy makers.

Anyway here's a version of AZ that only keeps one HVAP-majority CD (56.8%). There is only one extra cover in the Phoenix UCC and no pack penalties. I had a choice between an extra chop into Pinal or a chop into one of the large suburbs, so I went with the Pinal chop. It turns out that Phoenix has 15 recognized "villages" within the city limits. My CD 7 chops none of those to the extent that precincts allow, and only one of the 15 is chopped between CD 8 and 9. I could have avoided that chop but the erosity would suffer.




Would you really describe this as a "crappy" map?

[muon2]

I concede that my map chops up Pinal, and as I said in my comment on that map I could trade the Pinal chop for chops of the large burbs around Phoenix and I opted for the former, but I'm not wedded to it. For example your offering without the bridge has a pretty nasty chop of Glendale. I'm not sure my map's really that much more erose than yours, and if I chopped some burbs it might well match or betters yours in erosity. We'd have to agree on county subdivisions within the large counties to make a definitive determination.

I gave you the end of free microchops and start of penalties for UCC underpacks, so I think have been eminently reasonable when I comes to good ideas that have demonstrable value. However on this one there is public value to disallow any districts where you can't drive from one part of the district to another without passing through other districts. The only exception I consider is when a county is disconnected internally due to water, deserts or mountains and the plan keeps the county whole. As to the public acceptance I will simply cite the Washington state statute on redistricting (RCW 44.05.090), "Areas separated by geographical boundaries or artificial barriers that prevent transportation within a district should not be deemed contiguous." That is very good public policy IMO and has been recommended at many panels I have attended as a strong tool to fight gerrymandering.

[muon2]

I gave you the end of free microchops and start of penalties for UCC underpacks, so I think have been eminently reasonable when I comes to good ideas that have demonstrable value. However on this one there is public value to disallow any districts where you can't drive from one part of the district to another without passing through other districts. The only exception I consider is when a county is disconnected internally due to water, deserts or mountains and the plan keeps the county whole. As to the public acceptance I will simply cite the Washington state statute on redistricting (RCW 44.05.090), "Areas separated by geographical boundaries or artificial barriers that prevent transportation within a district should not be deemed contiguous." That is very good public policy IMO and has been recommended at many panels I have attended as a strong tool to fight gerrymandering.

What is an example of either a geographical boundary or artificial barrier that prevents transportation within a district?

In WA it is primarily the Cascades where major highways are needed to justify a link, and Puget Sound where only ferries and bridges count as connections. In principle it can apply to any other part of geography that interrupts transportation.

[muon2]

I gave you the end of free microchops and start of penalties for UCC underpacks, so I think have been eminently reasonable when I comes to good ideas that have demonstrable value. However on this one there is public value to disallow any districts where you can't drive from one part of the district to another without passing through other districts. The only exception I consider is when a county is disconnected internally due to water, deserts or mountains and the plan keeps the county whole. As to the public acceptance I will simply cite the Washington state statute on redistricting (RCW 44.05.090), "Areas separated by geographical boundaries or artificial barriers that prevent transportation within a district should not be deemed contiguous." That is very good public policy IMO and has been recommended at many panels I have attended as a strong tool to fight gerrymandering.

What is an example of either a geographical boundary or artificial barrier that prevents transportation within a district?

In WA it is primarily the Cascades where major highways are needed to justify a link, and Puget Sound where only ferries and bridges count as connections. In principle it can apply to any other part of geography that interrupts transportation.

The Cascades and Puget Sound are not artificial barriers.

The two I listed are geographical. I believe that an example of an artificial barrier would be a road along a boundary between two areas without a road that separately connects into each of those areas.

[muon2]

Indeed the connections are not friendly in NV. There are two counties where a majority of the population is disconnected from the seat of government (Esmeralda and Nye). Here is the connection map I put together last year. The yellow lines indicate state highway connections to the isolated population centers, and could be used to establish internal connectivity.



If one is willing to give up a county chop to reduce erosity, then Henderson can be linked to Nye. Within Clark this only chops Paradise. With higher inequality, Mineral could be moved to CD 2.

[muon2]

You  have a double chop of Paradise. One precinct from your NV-4 is entirely within Paradise, and two or three more have most of their population in Paradise. I could have done a double chop of Paradise too, but decided it better to do the second chop in another jurisdiction (not sure of the name of the place directly east of Las Vegas).

Thanks for the catch. I thought I had all overlapping precincts. If I place that one in CD 1 the population is still OK. If I place the overlapping precincts (or some fraction of them) in CD 1 then CD 4 needs to pick up the shore of Lake Mead and Moapa Valley to rebalance the population with some loss of erosity (but not that much since it's largely open desert). That gets the Clark muni chop back to 1.

[muon2]

You  have a double chop of Paradise. One precinct from your NV-4 is entirely within Paradise, and two or three more have most of their population in Paradise. I could have done a double chop of Paradise too, but decided it better to do the second chop in another jurisdiction (not sure of the name of the place directly east of Las Vegas).

Thanks for the catch. I thought I had all overlapping precincts. If I place that one in CD 1 the population is still OK. If I place the overlapping precincts (or some fraction of them) in CD 1 then CD 4 needs to pick up the shore of Lake Mead and Moapa Valley to rebalance the population with some loss of erosity (but not that much since it's largely open desert). That gets the Clark muni chop back to 1.

Well, another difficulty in NV (and, really, in much of the South and West) is that precincts don't line up with town lines.  Like, okay, you have to split Paradise because of all the precincts it shares with Enterprise and Winchester. 

I would think that there could, instead, be some sort of effort to fudge a standardized boundary that counts as non-chopped, and which is as close to actual as you can get.  (And, perhaps, that effort might want to make sure to distinguish between what are actual incorporated towns, and what are just CDPs.)

Cutting the voting districts to conform to town boundaries would be a good thing to do in reality... but we can't do so here in DRA.  You need to dig into the weeds of GIS to get it done.

...

As for those NV counties where the center of population and county seat are disconnected... possibly we could cut them and make fictitious counties which are internally contiguous, and draw based on that?

For the VA maps I was making a best approximation to the actual lines using precincts, and I used the same method for the Detroit hoods. The mapping exercise shouldn't be dependent on the chance that certain jurisdictions add up exactly, so as long as there is agreement about the precinct-level boundaries in advance then the task is a fair representation of what would happen with actual boundaries at a commission.

To calculate erosity it is necessary to fully assign every precinct to a county subdivision. Clark county only has 5 recognized cities: Boulder City, Henderson, Las Vegas, Mesquite, and North Las Vegas. All the other communities on DRA are unincorporated Census-designated places (CDPs). That's part of why the precincts overlap those boundaries. Presumably areas not in a city or CDP have to be assigned to a CDP.

My view of the internally disconnected counties is to treat them as whole counties, but allow state highway connections to the county through the disconnected parts as shown by the yellow lines. Cutting yellow lines doesn't add to erosity, nor does a chop that separates the disconnected parts. A chop is still a chop however.

[muon2]

You  have a double chop of Paradise. One precinct from your NV-4 is entirely within Paradise, and two or three more have most of their population in Paradise. I could have done a double chop of Paradise too, but decided it better to do the second chop in another jurisdiction (not sure of the name of the place directly east of Las Vegas).

Thanks for the catch. I thought I had all overlapping precincts. If I place that one in CD 1 the population is still OK. If I place the overlapping precincts (or some fraction of them) in CD 1 then CD 4 needs to pick up the shore of Lake Mead and Moapa Valley to rebalance the population with some loss of erosity (but not that much since it's largely open desert). That gets the Clark muni chop back to 1.

Well, another difficulty in NV (and, really, in much of the South and West) is that precincts don't line up with town lines.  Like, okay, you have to split Paradise because of all the precincts it shares with Enterprise and Winchester. 

I would think that there could, instead, be some sort of effort to fudge a standardized boundary that counts as non-chopped, and which is as close to actual as you can get.  (And, perhaps, that effort might want to make sure to distinguish between what are actual incorporated towns, and what are just CDPs.)

Cutting the voting districts to conform to town boundaries would be a good thing to do in reality... but we can't do so here in DRA.  You need to dig into the weeds of GIS to get it done.

...

As for those NV counties where the center of population and county seat are disconnected... possibly we could cut them and make fictitious counties which are internally contiguous, and draw based on that?

You could split the precincts on the city boundaries.

We are constrained to DRA for mapping software, and the 2010 data is given by VTD. If you know of another free web product that has finer granularity, I'm sure we'll be interested.

[muon2]

Wasserman redraws CA. He mostly just shores up the more vulnerable Democrats and gets rid of Valadao.

At the very least, Denham and Knight could be put in swing districts in addition to getting rid of Valadao. There is no reason to strengthen the other districts to the extent that he did.

You are thinking like the chair of the DNC as opposed to a member of the CA Assembly. State lawmakers have working relationships with their federal counterparts, and that relationship is important to get federal funds and projects to the state. It is much better from a local political perspective to have secure members of Congress than ones who are always worried about an upcoming tough election. CA was well known for creating bipartisan incumbent protection maps prior to the Commission this cycle.