CA CD Wine Country Map Quest poll

[muon2]

I've started looking at the whole county analysis I did for the Iowa-style states and apply it to my regional approach to CA. One interesting region I found is Colusa, Contra Costa, Lake, Marin, Mendocino, Napa, San Francisco, Solano, Sonoma, and Yolo comprise the population of 5 CDs with only 625 extra people. Within that region the core wine counties of Mendocino, Napa, and Sonoma are only 5298 over pop for a district and Colusa, Lake, Solano, and Yolo are only 2628 under population for a district. I'll try to post a wine county region map based later today.

[muon2]

That really is the main goal here. What rules need to be set up, which make sense for CA, so we don't get another cf map next time?  I mean they are not needed really for a Commission made up of the Sbane, Lewis, Muon2, dpmapper, and Johnny's of this world, since we know all the rules, and will make each other's lives a living hell if we play games, since we know all the games too, but well we aren't on the Commission, and our types never well be. They will put folks on who can be gamed to death, and then some.

This is the question for me. Some states do fine even without a commission, and in part that seems to be a set of criteria with enough definition to constrain the worst of games. Even in CA, counties get some cred as units to be preserved. From commission testimony, they have at least the weight of municipalities, if not more. As Torie has often noted the problem lies in shifting definitions for socioeconomic groupings.

With that in mind, I push for minimal county fragments (BTW I don't need a 3-chop in San Joaquin), and if there are chops the preference is to avoid municipal chops and create minimal population incursions into a county when needed. Other C of I factors can then guide mapping beyond the predefined C of I represented by counties and municipalities.

As I mentioned, I have looked again at multi-district regions as if they were large single districts. I can apply the formula from the Iowa type districts to estimate the following ranges of deviation between the smallest and largest deviation for CA with 58 counties:

2 regions, range 45
3 regions, range 229
4 regions, range 723
5 regions, range 1767
6 regions, range 3665

I can draw 5 regions with a range of 1033, but it has one region that is not very compact. There is a 4-region plan based on the suggestion I had above with a range of 896. It's consistent with the estimates, and is the basis for a fresh look at the plan.

Region A: 4 CDs, deviation -188, Butte, Del Norte, Glenn, Humboldt, Lassen, Modoc, Nevada, Placer, Plumas, Sacramento, Shasta, Sierra, Siskiyou, Sutter, Tehama, Trinity, and Yuba.

Region B: 5 CDs, deviation +625, Colusa, Contra Costa, Lake, Marin, Mendocino, Napa, San Francisco, Solano, Sonoma, and Yolo.

Region C: 14 CDs, deviation -271, Alameda, Alpine, Amador, Calaveras, El Dorado, Fresno, Inyo, Kings, Madera, Mariposa, Merced, Mono, Monterey, San Benito, San Bernardino, San Joaquin, San Mateo, Santa Clara, Santa Cruz, Stanislaus, and Tuolumne.

Region D: 30 CDs, deviation -166, Imperial, Kern, Los Angeles, Orange, Riverside, San Diego, San Luis Obispo, Santa Barbara, Tulare, and Ventura.

Though this might not always group counties the way one likes, it imposes a rational constraint to keep counties whole and reduce the exposure to arbitrary influence.

[muon2]

Yeah, the commission map is all over the place. Still, I don't know if they did anything as bad as putting Richmond with Marin County. I mean, Marin County has a median income approaching 100k and Richmond is right around 50k. And it's not as if Marin county is only 20 or 30,000 people or something, and the same with Richmond and surroundings.

Yes, yes, but it mitigates the CA-03 chop in CCC. It is all a balancing test, which is why I wanted input. Once I get the dpmapper map done, the new poll will have nine maps in it.

Right, but what I'm saying is that CA-3 chopping into CCC makes infinitely more sense than a district from Marin County chopping into CCC. It is a balancing test, and the 2nd map is the right mix. It's not perfect obviously, no map is going to be, but it's the best of the lot. Maybe I'm missing it, but does crossing the Richmond-San Rafael bridge lead to one less chop than CA-3 chopping CCC?

If one assumes that as in IA (or AR in 2000) de minimus population deviations are permitted to maintain county integrity, then my plan makes for fewer county chops. Even with an exact population requirement, the needed chops would be preferable since the counties would remain largely whole.

[muon2]

Here's my offering for wine country showing how it fits into the northern half of the state. All CDs are within 100 of the ideal population. All counties with enough population have a district entirely within, and Sacto and Alameda have two CDs entirely within. No county has part of more than two CDs not counting those entirely within. County splits are organized to split off only a small part where possible leaving most of the county intact. For example these are the populations of the fragments split:

Butte: 7.8K
Colusa: 0.6K
Napa: 3.3K
Placer: 0.5K
Sonoma: 2.0K

[muon2]

Ah, well I already knew that you were smarter than me, Mike.  Given the walls (see below), I thought it was impossible, but then I thought it was a crime to chop Section 5 protected Merced (which to do your twist, you chopped). I guess it is OK to remove white people from it, is that were its at? And didn't you while losing a San Joaquin chop, pick up two others (Madera and Merced)? Did you avoid a chop somewhere else?  I assume your map is based on the Ridgecrest based chop. And didn't you reduce the Hispanic percentage in the Monterey based CD (not that that is dispositive of anything of course)?

Oh, I think you did that extra chop for VRA reasons, even though the Commission didn't care much about the Hispanic percentage in CA-16.  So you took CA-08 farther north?  You didn't do the Ridgecrest chop, but chopped Placer County instead. Taking empty Alpine is OK I guess, but not San Andreas. And that picked up about half of the population that you needed, and you got the other half from the extra CA-16 area county chop, excusing it with the VRA?  I am just wondering where you picked up the population is all.

The Placer affair of course illustrates the conundrum here. The issue with your algorithm, is when to allow it to be violated.  Sometimes the geographic barriers or communities of interest, or issues of compactness, or protecting cross county metro areas from being chopped (e.g. West Sacramento from Sacramento, although the Sacto area needed to be chopped anyway), are just too compelling. Or you need an extra county chop to avoid an ugly chop of some town with some population. For example, I don't think it acceptable to chop Woodlands in half, and considered that a constraint. So how do we allow for human judgement while still being sufficiently leashed to avoid going where this Commission went?

The Placer problem was actually quite easy. My 2 northern CDs plus Sacto (region A) has a pop deficit of 188 but it becomes a surplus of 437 when I give it 625 from Colusa to balance the north bay region. There's a little precinct in Placer on the shore of Lake Tahoe that has the right pop and keeps the Tahoma community together. The split actually reduces town splitting so that's a plus. That split actually goes with the South Lake Tahoe pop (about 39K) to join the Owens Valley in the Victorville CD. I haven't finished the south yet, but you can sense my walls from the regions I posted.

I also haven't gotten to a VRA analysis yet, but I wanted to start with the algorithm and it said that San Mateo+Santa Cruz+Monterey = 2 CDs less 10K pop which I picked up from San Benito. So my wall is where the commission put it between Monterey and SLO. If I can't pass Section 5 with Monterey I can violate my subregion and draw much the same CD the commission did with a few small shifts in SJ. The VRA would be one of the times to violate the algorithm. The VRA is what prevents me from putting a district entirely within Fresno, for example.

So I have two areas of human judgement in map making so far. Making necessary county splits in ways that minimize community splits, and accommodating the VRA.

[muon2]

There doesn't appear to be a road between Alpine and Placer.

There is through South Lake Tahoe. My Stockton CD would have 39K too many people including all of the usual Sierra Nevada counties (40K with Alpine). Something needed to go with the Owens Valley so I chopped off South Lake Tahoe from El Dorado for that population. It then provides the link to my one Placer precinct along the lake.

[muon2]

There doesn't appear to be a road between Alpine and Placer.

There is through South Lake Tahoe. My Stockton CD would have 39K too many people including all of the usual Sierra Nevada counties (40K with Alpine). Something needed to go with the Owens Valley so I chopped off South Lake Tahoe from El Dorado for that population. It then provides the link to my one Placer precinct along the lake.

I will look for the road again. That Monterey, SLO wall makes a cf of Ventura County. That is the problem with a strict algorithm.  You get a nasty mess in a county, or you do something that will never sell, like appending Placer to a Victorville CD.  But we have been over this before. I wonder if there is any way to get some discipline, but allow exceptions.

Other than some minor pop balancing what I did is place Lake Tahoe with the east side district that includes Owens Valley. If you come up US 395 from OV into Alpine and go north on CA-89 it connects the miniscule piece of Placer that I use on Lake Tahoe. It's exactly where I was vacationing last summer so I had a good chance to see the area. It was easier to go from Alpine to either Mono or Lake Tahoe than over the passes to the west. Often it seemed like West Nevada, minus the casinos.  From my visit, it is hard to convince me that Tahoe can't be appended equally well to OV and the deserts south as it is to suburban Sacto.

I think the exceptions too often come from preconceived notions. An outsider assessing the situation neutrally, might not have those same notions. That's part of the beauty of the Iowa system. The politicians and public aren't always happy, but there's a recognition that game-playing is

I'll get back to you on the Ventura problem. Given the crazy current CD 24 there, and the issues noted it's clear that both commission and political gerrymanders don't satisfy.

[muon2]

My initial thoughts on Ventura starts with SLO+SB which is 9K short of a district and can be completed like the commission did. In Ventura there is a ridge the separates Santa Paula from the south, and south of that ridge extended across Ventura to the ocean is a population equal to a district. Next Ventura+Kern+Tulare is only 4K short of 3 CDs. A VRA district can be constructed from Tulare going into Bakersfield. That leaves the rest of Kern linking to Santa Paula. There is almost the right amount of population west of I-5 in LAC to make a link that doesn't rely only on the windy road in north Ventura.



The districts are accurate to within 100 persons and the I've added some color to show the inclusion of Hungry Valley SVRA split from its block group.

[muon2]

I have little to add, but I'll mention that I agree with sbane about class being important. Richmond being appended to Marin County is as clear a no-no as can be; I can't imagine any other two places that vote the same way being more different.

Of course, I would have placed Cupertino with Saratoga and Los Altos and Palo Alto and then perhaps up to Menlo Park and thereabouts, so my opinion is to be taken with a grain of salt.

This where I have my greatest disagreement with some of the commission results. Drawing districts to group by class is as bad as drawing districts to group by political party.

[muon2]

My initial thoughts on Ventura starts with SLO+SB which is 9K short of a district and can be completed like the commission did. In Ventura there is a ridge the separates Santa Paula from the south, and south of that ridge extended across Ventura to the ocean is a population equal to a district. Next Ventura+Kern+Tulare is only 4K short of 3 CDs. A VRA district can be constructed from Tulare going into Bakersfield. That leaves the rest of Kern linking to Santa Paula. There is almost the right amount of population west of I-5 in LAC to make a link that doesn't rely only on the windy road in north Ventura.



The districts are accurate to within 100 persons and the I've added some color to show the inclusion of Hungry Valley SVRA split from its block group.

Well we do have a fundamental disagreement here. I don't think it acceptable to append Victorville (oh it is  Bakersfield, which is even worse) to Santa Paula and bits of Ventura city and Ojai - unless demanded by the VRA (which of course it isn't). If you are going to go this route, you need to append to Ventura County the suburbs west of Woodland Hills in LA City. You have CA-25 in three counties to boot, so it chops twice, just so you have your little road connection.

I want some workable balancing test between respecting obvious communities of interest,  jurisdictional boundaries, metro areas, and compactness and geographical barriers. For example, I would rather chop a county, than chop a sizable town in half - unless there is a compelling reason like the Vallejo thing (one reason the dpmapper map has some attraction is that it avoids chopping Vallejo, which is less acceptable really than chopping Solano). And I might chop another county chop going my version of your route, to unite Tracy. Nabbing Tracy to me is a lot more attractive than Victorville nabbing South Lake Tahoe, which has no ties to the rest of the district (yes, I found the road that I did not know exists over the Sierras via "Luther Pass"  (which I did not know existed either:)). Sure it is easier to have to hew to just one loadstar, but among other things, I don't think the public will accept it if presented to them, and they thought about it. The trick is to find a methodology that will effectively leash abuse and partisan/incumbent/aspiring incumbent games.

I might add that in general, counties have less psychological importance in CA than most places - perhaps in part due to geography and ethnic diversity, and perhaps  in part due anomie and rootlessness, which characterizes CA in a way not as present elsewhere perhaps. But counties are useful, because of course they have some importance, and it is a good leashing mechanism. Splitting cities is less acceptable.

And isn't class a communities of interest factor? Sure it should not trump more important considerations, but at the margins, it is certainly an appropriate metric no? I guess where I am going is some kind of hierarchy of "needs" as it were. In the Silicon Valley, after respecting town boundaries, either going Asian, or going by class seems OK with me. I see no reason why one should trump the other.

And for you uber class warriors, you know the "slums" need to go somewhere. Since Oakland has its own CD, Richmond needs to be appended to middle class areas, or tied to more rural Solano. The SF metro area is rather light on slums - it is just too expensive.

In most plans there will be at least two districts in LAC that are not wholly contained in the county. By placing my cut where I did I can keep the SLO-SB district as two whole counties, so it does reduce chops compared to other plans. Counties may be less relevant than in some other states, but the commission's narrative does give weight to them, often providing justification for a split.

Identifying geographical barriers is important in any set of constraints, and it's hard not to slip into an arbitrary definition. For constructing regions and subregions I have preferred to require that all counties in the grouping are connected such that one can go from one county seat to another on designated US or state highways without leaving the group. My subregion of Tulare, Kern, Ventura, SLO, and SB meets that test. Making districts within the subregion only requires a regular road to provide connection since at that point county fragments come into play, though highways are preferred.

[muon2]

If I stick to my Ventura subregion, here's the chance for human intervention. This is the other rotation that keeps splits of Ventura to one but splits SLO instead. Like the other plan this needs 13 K from LAC for pop balance.

[muon2]

If I stick to my Ventura subregion, here's the chance for human intervention. This is the other rotation that keeps splits of Ventura to one but splits SLO instead. Like the other plan this needs 13 K from LAC for pop balance.

That won't sell/work either Mike. Sorry. But by all means finish your algorithm map. I have decided to get more rigorous about this myself, and I will try to delete Chino Hills from the LA County Asian CD, unless the VRA precludes it, because it dilutes too much the Hispanic CD in SB County. SD County should have put one chop out of it, by the way.  Extra county chops beyond the VRA require another very good reason, like keeping a metro area together, or due to compelling geographic barriers, like between Kern and SLO or Ventura County for example.

So we need to define geographic barriers with some rigor so it is clear when it permits the violation of other rules. I made my suggestion based on types of highway connection, do you have one to offer?

My notion of county splits would include three levels. One is for de minimis splits of under 0.5% of a district required to satisfy OMOV. The next is for minor splits that are less than 5% of a district and less than 20% of a county. Finally there are major splits that exceed 5% on both pieces. My preference is for lesser splits. Districts entirely within a county do not count as a split.

So applying this to the southern wine county of Santa Barbara, my first offering has a minor split of the remainder of Ventura and a minor split in LAC. The second offering removes the spit from Ventura and replaces it with a major split of SLO. Using neither of these but keeping the wall around the north of this subregion requires at least a return to the minor split of Ventura and two major splits in LAC (not counting the inevitable split on the east).

BTW I find that even with Chino Hills I get a 63.1% HVAP CD entirely in SB.

[muon2]

That won't sell/work either Mike. Sorry. But by all means finish your algorithm map. I have decided to get more rigorous about this myself, and I will try to delete Chino Hills from the LA County Asian CD, unless the VRA precludes it, because it dilutes too much the Hispanic CD in SB County. SD County should have put one chop out of it, by the way.  Extra county chops beyond the VRA require another very good reason, like keeping a metro area together, or due to compelling geographic barriers, like between Kern and SLO or Ventura County for example.

So we need to define geographic barriers with some rigor so it is clear when it permits the violation of other rules. I made my suggestion based on types of highway connection, do you have one to offer?

My notion of county splits would include three levels. One is for de minimis splits of under 0.5% of a district required to satisfy OMOV. The next is for minor splits that are less than 5% of a district and less than 20% of a county. Finally there are major splits that exceed 5% on both pieces. My preference is for lesser splits. Districts entirely within a county do not count as a split.

So applying this to the southern wine county of Santa Barbara, my first offering has a minor split of the remainder of Ventura and a minor split in LAC. The second offering removes the spit from Ventura and replaces it with a major split of SLO. Using neither of these but keeping the wall around the north of this subregion requires at least a return to the minor split of Ventura and two major splits in LAC (not counting the inevitable split on the east).

BTW I find that even with Chino Hills I get a 63.1% HVAP CD entirely in SB.

This is the remaining chop of Ventura that makes any sense with my regional walls to reduce county splits. As I alluded to, it requires the unpleasant chop into Lancaster (with a chop of 124K, most of the city goes to Kern). If you think this is better than the first two options, I need to know how to recognize this as an appropriate time to bend the algorithm to human input.

[muon2]

Meh, Torie's map in his sig (and iirc I had a similar map) is much preferable to that. And between the two options you posted, I would say they are equally flawed (or good).

Quite, but with the SLO/Monterey wall, your map is the best one can do I think.

By that do you mean my first one with the Kern-Ventura link? That's the one most consistent with the algorithm.

You must be logged in to read this quote.

That's why I avoided it until my first two maps with links to northern Ventura were treated so unkindly.

You must be logged in to read this quote.

I assumed that it was the easiest path to meet section 5, keeping the district largely as is. I'm curious to see what you find out.

[muon2]

I have been meaning to ask my "source" about just how it came about that the SLO/Monterey wall was decided upon, which makes such a hash out of everything.

I assumed that it was the easiest path to meet section 5, keeping the district largely as is. I'm curious to see what you find out.

If the Hispanic percentage were upped in a CD in which Monterey is wholly contained from what it was, how could adding a slice of SLO to it raise a section 5 issue?  My new Monterey County CD clocks in at a 46.3% Hispanic VAP population. The previous CD was at 44.3% per the DRA utility's numbers.

I would agree with your interpretation, and I did much the same thing in my January plan. That was also the basis for the Merced split in this month's offering. I just wonder if the commission had advice the drew them towards keeping the sect 5 counties whole to minimize any DOJ issue.

[muon2]

Here is LAC for the plan with the Kern-Ventura connection. The purple CD is 50.9% AVAP (Brea is almost completely intact in the CD). The medium blue CD is 43.3% BVAP. There are 6 CDs with >50% HCVAP. There are only 2 CDs that span the county line and in both cases a part of the split makes up less than 5% of a CD.

[muon2]

We need wall placement rules. Put that on the list.

I think I have firm rules for walls. Walls exist around regions of whole counties that are nearly equal to a whole number of districts. By nearly equal I would use population deviations that have stood up before SCOTUS. Perhaps we'll get a better idea of their view on deviations from the WV case. In any case we can predict what range one should get given the number of regions and counties in a state. If the result is statistically close to that deviation then the regions can be considered.

For CA with 58 counties the predicted ranges for different numbers of regions are
2 regions, range 45
3 regions, range 229
4 regions, range 723
5 regions, range 1767
6 regions, range 3665

For the purposes of forming regions counties in a region must be connected. Two counties are connected if one can travel from the county seat of one county to the count seat of the other county on US or state highways without passing through another county.

[muon2]

Yes, but if a wall location ends up creating a nasty Lancaster chop, or forces stuff to go where it should not go, that is a problem.  And there may be crossings that should be "disfavored."  Having some statistical limits like you suggested might work, which can be violated under certain circumstances. Or maybe we have defined regions, in which there may be only one chop out unless the VRA demands otherwise, or to unite a city that is already mostly in one CD. And Sacto taking W. Sacto, or uniting Yuba City and Marysville, perhaps should not count as a chop for example. 

But can you describe exceptions a priori so that we don't have map makers coming up with rules a posteriori to suit their needs? Iowa has learned to live with some unusual pairings because they trust their rules to prevent mischief. Could other states do the same?

[muon2]

I am thinking we should go with the Michigan rules, with wall placement up to the Commission unless it ends up chopping a metro area. If it does, that wall has to go. I think that horrible SLO /Monterey wall forces a chop of the Santa Cruz metro area. Now that I got rid of the Chino Hills chop, I think my map now has the minimum chops of counties other than two chops forced by the VRA, and after I massaged the black CD, there should be a minimum of city chops (other than as forced by the VRA).  I shaved CA-35 down to a minimum 50% CVAP to minimize the chop into San Bernardino City (I only had to take about 10 precincts or something).

If you think I have an extra county chop, let me know. The idea is that each CD generates two chops, except one CD with which you start (CA-01 here), unless a county can hold at least three CD's, and then at least one most be wholly contained in that county. Interior walls that make one CD a one chopper, forces another CD to be a three chopper (e.g., in my map, CA-04). At least that is my way to try to break through the fog of chop counts, which itself gives one a headache. Let me know your thoughts.

I am still unclear why your map itself uses the SLO/Monterey wall btw. Surely you don't find favor with it do you? Not if it produces maps that are well, like the Commission's - or yours.

If it's to be MI rules, then there has to be something to replace the constraints of the MI townships, or it still seems a bit loose. What you really want is something that generates a wall between LAC and both Kern and Ventura, and I'm not sure how you get that without a Torie on the Commission. I prefer that wall myself, but I'm looking for a model that might lead in that direction. For instance should one consider the magnitude of a county (or city) chop?

On VRA districts how did the Asians fare? Does section 5 really force 3 chops for the Salinas district? That seems hard to justify in your methodology. To break the M-SLO wall one should contain the northern chop to only SCruz or SClara.

[muon2]

On Section 5, there is enough ambiguity that diluting the Hispanic percentage or chopping a Section 5 county that the chop may be justified. Any VRA chop should require a written legal opinion that without doing it, the risk of a Section 5 (or a Section 2 violation for that matter) is more than remote. I would no problem getting rid  of that chop into Gilroy, although it does force a Monterey CD shave into the Santa Cruz suburbs (or if you go into Santa Clara instead, a shave into the Silicon Valley). However, hewing to a clear chop rule is more bright line, and should trump the metro shave consideration, where the two are in conflict.

As to the type of chop, I have been pondering some sort of percentage limitations, and am thinking that the statute should have some preferred guidelines, which if violated require a written finding by the Commission justifying breaking the guidelines. That at least should concentrate the minds of those there that are just dialing it in. However, for very small cities (say 15,000 or less, maybe 10,000 or less), sometimes they might need to be chopped more severely to avoid a map looking grotesque. For larger ones, a 20% guideline might be reasonable. I think in fact I hewed to such a guideline throughout the state, except for Vallejo. I would sign off on a written finding on that one I think. Obviously, that was the toughest part of the state to map.

The same concept might be stated for the separation of CD's along the Sierra-Techachpi line in the south, keeping CD's from crossing in and out of the Central Valley, and keeping CD's not in a metro area from nibbling in (the Santa Cruz thing).  

It might also be good to require staff to draw up alternative maps hewing to all these guidelines. I think if the Commissioners could see a group of well done alternatives, they might have done a better job. And of course, the partisan numbers should be revealed. I suspect that some knew the numbers, and some of the dial it inners did not.

It is kind of interesting that in the end, my obsession with Michigan gave me some "training" for this exercise. One never knows where things will lead sometimes, does one?  I doubt a year ago I could do what I can do now. I think I maybe have a new skill.

Oh, as to the Asians, since the rule in the 9th Circuit is 50% CVAP, no Asian CD's are required. However, an Asian CD should have been drawn like we did in LA County, because it can be drawn without violating other guidelines. But that is a judgment call. In the Silicon Valley, one can go either the Asian route or the class warfare route (one can't get to 50% AVAP anyway up there without violating other guidelines, much less 50% ACVAP of course). That to me is a fairly close call.

That may work for the chop rules. For walls I would suggest a page out of Iowa. If an initial set of walls creates a problem, the commission can send it back asking for a version with a specific requirement such as no wall between Monterey and SLO and at least no wall between at least one of either SCruz or SClara to insure a section 5 map. The walls can create regions according to the statistical tables above. Swaps between regions of counties can only occur in one county as with CDs for MI.

BTW what is your differential between AVAP and ACVAP?

[muon2]

As to the type of chop, I have been pondering some sort of percentage limitations, and am thinking that the statute should have some preferred guidelines, which if violated require a written finding by the Commission justifying breaking the guidelines. That at least should concentrate the minds of those there that are just dialing it in. However, for very small cities (say 15,000 or less, maybe 10,000 or less), sometimes they might need to be chopped more severely to avoid a map looking grotesque. For larger ones, a 20% guideline might be reasonable. I think in fact I hewed to such a guideline throughout the state, except for Vallejo. I would sign off on a written finding on that one I think. Obviously, that was the toughest part of the state to map.

I have been thinking about the chop rules and I agree that it makes sense to require specific justification for more than a "minor chop". I would stick to 20% of a county or city as a threshold. For the example you gave of a small city chopped to avoid a bizarre shape you can make that one of the possible findings. I would also explicitly allow any chop into a county where all other CDs in the county are wholly contained (eg SF).

I would however put an absolute upper limit on a minor chop size. Consider a CD spanning LAC and OC. A 20% county chop limit allows a completely free hand in crossing the border since a whole CD is less than 20% of LAC and just over 20% of OC. I think that much freedom in mapmaking is worth avoiding, so I will continue to advocate for an upper limit of 5% of a CD in a fragment to be considered a minor chop.

BTW do I see a tiny chop into Pomona that I assume you could not avoid?

[muon2]

As to the type of chop, I have been pondering some sort of percentage limitations, and am thinking that the statute should have some preferred guidelines, which if violated require a written finding by the Commission justifying breaking the guidelines. That at least should concentrate the minds of those there that are just dialing it in. However, for very small cities (say 15,000 or less, maybe 10,000 or less), sometimes they might need to be chopped more severely to avoid a map looking grotesque. For larger ones, a 20% guideline might be reasonable. I think in fact I hewed to such a guideline throughout the state, except for Vallejo. I would sign off on a written finding on that one I think. Obviously, that was the toughest part of the state to map.

I have been thinking about the chop rules and I agree that it makes sense to require specific justification for more than a "minor chop". I would stick to 20% of a county or city as a threshold. For the example you gave of a small city chopped to avoid a bizarre shape you can make that one of the possible findings. I would also explicitly allow any chop into a county where all other CDs in the county are wholly contained (eg SF).

I would however put an absolute upper limit on a minor chop size. Consider a CD spanning LAC and OC. A 20% county chop limit allows a completely free hand in crossing the border since a whole CD is less than 20% of LAC and just over 20% of OC. I think that much freedom in mapmaking is worth avoiding, so I will continue to advocate for an upper limit of 5% of a CD in a fragment to be considered a minor chop.

BTW do I see a tiny chop into Pomona that I assume you could not avoid?

Your text confuses me a bit. You are saying for county chops, if less than 5% no written finding, if between 5%-20% just what again (?), and of course a written finding over 20%.  You could deal with the county size issue by having the 20% limit apply both to the percentage of the county and the percentage of the CD involved in the chop, no?  For a small county, having to justify a 5%-20% chop seems just silly (the answer is because the county is small!).

If I was confusing it's because there are two different criteria at play. A minor chop into a county or city is both a) less than 20% of the political jurisdiction and b) less than 5% of the CD. For small counties only part a) matters, so it matches your statement. Part b) is to address chops in large counties. The transition occurs when a county is larger than 25% of a CD.

You must be logged in to read this quote.

Ah. I'll have to trace your tree of linkages in this latest map.

[muon2]

If I was confusing it's because there are two different criteria at play. A minor chop into a county or city is both a) less than 20% of the political jurisdiction and b) less than 5% of the CD. For small counties only part a) matters, so it matches your statement. Part b) is to address chops in large counties. The transition occurs when a county is larger than 25% of a CD.

You must be logged in to read this quote.

Ah. I'll have to trace your tree of linkages in this latest map.

I will send you my file tonight after I cleanse the black CD of chops. I would appreciate your fly specking  it.

Got it on your formula. The 5% of a CD figure might be low. I mean, what is wrong with a chop that takes in an entire city on the border of a county part of a larger mass of tracts that has a population of say 85,000 (unless it is a county seat, which is one of the reasons that I decided that CA-04 should not take Yuba City (shearing a county of its major town and county seat seems wrong to me if it can be avoided), while W. Sacto is the not the main action in town for Solano County).  Also if the percentage is too low, you might have a map to goes from Victorville to S. Lake Tahoe, or something (absent a written finding to the contrary). No! 

Pending further discussion, go KISS baby, and have the same percentage for both - 20%. 

How about compactness and shape?  Compactness really only obtains for a couple of areas of California, but compactness was on my mind always when drawing CA-01 and CA-02. And that Palos Verdes to Hancock Park "class warfare" CD that the Commission drew is just an obscene sin.

And how about "unnecessary" sword cuts into a county sometimes over empty zone, when less of a sword cut crossing no empty zone was available (switch out the sword cut of the Commission into Ventura County from Simi Valley to the Thousand Oaks area), and get rid of that horrid Westminster sword cut into OC.

Should erosity and wanderlust and sword cuts require written findings justifying them in lieu of the alternative that would avoid that (assuming in both instances the chop rules are otherwise met)?

My initial tracing has me content except for the Salinas CD (keep it out of either S. Cruz or S. Clara.) I'm also not happy about the Sacto region, but I'll put something together later.

The 5% actually stems from my desire to extend this to state house districts as well. At 5% I can leave the fragment out and still meet federal standards. My KISS is simply different than yours. That same motivation also drives my wall tolerance of 0.5%, which could possibly stand up federally for the compelling state interest of counties intact. My macro rules should be independent of the state, then fine tuned for the particulars of the state.

[muon2]

I don't think those are good enough reasons to keep it as low as 5% for CD's, or that matter in CA, assembly districts as well, which are about 450,000 persons a pop. National standards are nice and all, but some other time. I just don't think it is much of a problem to have 20% of a CD in another county from a public policy perspective.  5% is only 35,000 people. As I said, that may push a map in the "wrong" direction, and I really don't want to have to get a written finding to push it back in the right direction.  15% would probably work, maybe even 10% (I will have to see how big my chops are ). The CA-20 double chop is just a section 5 issue as we discussed, and would require a legal opinion to do it, so I think we have resolved that (presumably one would be written).  I don't think Sacto generates another chop having analyzed it to death, but if it does, let me know. That was the trickiest part of the state to chop count. Thanks for looking at it. (I also don't think Sacto taking W. Sacto should count as a chop really (a situation unique to the state really), but carving out an exception for it is probably not worth the candle.)

If all counties were smaller than a district, then for n districts the minimum number of county fragments with exact population equality is 2n-2 (assuming there isn't a perfect combination of counties as in WV). Mathematically, every district completely embedded in a large county removes two fragments from the computation. So when I see no districts entirely in Sacto county when two are possible, I know there are more splits than the minimum.

I'm not convinced that 140 K is a "minor split". If that's your desire I'd rather see 5% across the board as a forgiven without justification, and skip 20% entirely. I was just trying to be accommodating for small counties that a commission needed to split.

[muon2]

I don't think those are good enough reasons to keep it as low as 5% for CD's, or that matter in CA, assembly districts as well, which are about 450,000 persons a pop. National standards are nice and all, but some other time. I just don't think it is much of a problem to have 20% of a CD in another county from a public policy perspective.  5% is only 35,000 people. As I said, that may push a map in the "wrong" direction, and I really don't want to have to get a written finding to push it back in the right direction.  15% would probably work, maybe even 10% (I will have to see how big my chops are ). The CA-20 double chop is just a section 5 issue as we discussed, and would require a legal opinion to do it, so I think we have resolved that (presumably one would be written).  I don't think Sacto generates another chop having analyzed it to death, but if it does, let me know. That was the trickiest part of the state to chop count. Thanks for looking at it. (I also don't think Sacto taking W. Sacto should count as a chop really (a situation unique to the state really), but carving out an exception for it is probably not worth the candle.)

If all counties were smaller than a district, then for n districts the minimum number of county fragments with exact population equality is 2n-2 (assuming there isn't a perfect combination of counties as in WV). Mathematically, every district completely embedded in a large county removes two fragments from the computation. So when I see no districts entirely in Sacto county when two are possible, I know there are more splits than the minimum.

I'm not convinced that 140 K is a "minor split". If that's your desire I'd rather see 5% across the board as a forgiven without justification, and skip 20% entirely. I was just trying to be accommodating for small counties that a commission needed to split.

Well think about 10% at least. I will take a look at my chops on the  map, and try to get a real on the ground feel as to how much of a problem a tight constraint is, and just how necessary a written finding might be and when. (I sent you the drf file by the way.)

I understand your Sacto point, and saw it too (which bothered me of course, so I thought and thought and thought about it), but I think when what would otherwise be two nestings CD within a county, are moving out of that county to service the chop quota of two other CD's in separate appending counties (here CA-03 and CA-04),  then no additional chop is in fact created. If you nest a Sacto CD, then either CA-03 or CA-04 (depending on which CD the wandering nester is servicing), will have to chop some other county.  Play with it, and see if you can find a way to lose a chop. I could not, and I think the above might be why.  Hey, you're the genius on these things. Figure it out!  

You may be counting counties chopped, instead of the chops themselves. For example you have three CDs that go into Sacto county and also include part one one neighboring county. That's a total of six chops, but only three counties because you've tri-chopped Sacto. If you rearranged those three districts to place one CD entirely in Sacto there are still three chop fragments outside of Sacto, but only two inside. One chop disappeared because the CD is entirely within the county. One of the remaining pieces has three fragments, and in principle it can be brought to two, but it may not if you have more than two CDs with only one fragment.

I'll think about 10%, but it seems arbitrary. At least I can point to a federal decision on which to rest 5%.