Local vs regional road connections

[muon2]

One detailed difference is with areas like the one between Bothell and Kirkland. I have it split between two school districts. The few precincts just south of Bothell are like the SE corner of Plain in Stark. They are adjacent to a larger unincorporated township, but they are kept separate as a fragment of Plain. I do the same in King. There a number of similar places where you have merged small unincorporated areas that are in separate school districts.

That's fine. I didn't intend to ignore school district lines outside city or town entities. Any other differences beyond this concept? I see that you kept separate on your map bits of land surrounded by a city that is not itself a city or town, such is in Bellevue and Kent? Why did you do that, rather than just merge it?

A second potential difference is how those fragments are handled. Yes, a whole fragment can be shifted to a different district with no chop penalty - that's the Stark equivalence. However, it's still a chop and that means the shifted fragment could contribute to erosity, just like a chop anywhere else. I say could, because it could work either way or have no effect, again that's just like what happens with how we've applied erosity to chops in other states. To make it clear I show all fragments in the same school district with the same color.

Typically it would not matter with a CD border fragment right, unless doing it the other way, avoided any pavement cut right? I don't think I have a problem with that, although I did mention a preference regime. Yes, I know, you don't like preferences. More on that below.

Anyway, subject to the above about those surrounded non town or city fragments surrounded by cities (this is an instance, where the guy who does not like lines erased, thinks that they really should be erased), I think I sign off on your map. Nice job. Thanks for your hard work on this. I appreciate it.

As you have observed I've tried to model WA subunits (and hopefully other troublesome states) on work we've done elsewhere. OH seems to be a model that could be adapted to fit. In OH we left the surrounded pockets with the township, so consistent with that model I left the inclusions with the other unincorporated parts of the underlying school district. There's no penalty for grouping them with the surrounding city, but it makes the subunit creation rule much clearer.

Here's my version of Pierce using the same methodology I finalized for King. There are 39 subunits: 18 cities, 5 incorporated towns, and 16 unincorporated school districts.

[muon2]

I think my resistance to the one bite rule stems from how well our maps have performed without such a rule. Your MI offering is an excellent example that worked without the one-bite. I am resistant to disrupting that kind of success. I'd like to see that a particular goal is impossible without one-bite.

Yes, one bite happened to not be necessary, but then there were not subunit chops anywhere, and Detroit did not matter. The one bite rule will rarely be necessary to avoid an unfortunate skew result. It would happen in Phoenix if we had a population accident as I demonstrated. So I know it can happen, when it should not. The one bite issue really potentially matters in subunits for large townships and cities. For artificial subunits outside such large subunits, it probably will never matter. If might make for an uglier map, but it is unlikely to have much of a partisan difference. And with enough subunits, such as we now have in King, it is unlikely to result in an ugly map to boot. So I am focused I think about subunits within large townships and cites. That is where I think leeway is needed provided that the map does not get erose.

I'm also resistant to an endless preference list that isn't part of the score. I think one of the successes of the UCC model is that we made into a scoring modifier, not a mere preference. There may well be other items that should function like the UCCs and that means working out a scoring rubric for them. My example along this line is the MCC to handle the rural minority counties of the South. We should look at each preference and see how to create a balanced score for it.

Yes, the preference list does seem endless, doesn't it. It may grow yet longer!  The thing is, is it is a good tool for getting the balancing test right. I think it is right for bridge chops as defined by me, and certainly for Indian reservations (although reservations have less national importance). If reservations are treated like counties, than one gets no credit for keeping a county whole. In fact one is punished. Not good. And surely you don't want to treat the same a map that chops both Apache County and the Navajo Indian Reservation as opposed to a map that chops but one of them do you? And wouldn't a chop that divides an Indian Reservation, but unites a county, but better than another county chop elsewhere? If you have another way to get the balancing test right, not using the preference mechanic, it certainly should be considered. But it does need to get to the same place in my view. The fragment preference is more minor. To me it seems like good policy to keep a subunit together that is divided all things being equal, but it's not really that important. I think it is just good policy, and I think a judge would agree. Judges would like the preference concept I think. That is something they would understand.

The thing is with all of this, is that the rules should facilitate what a fair minded person would do when drawing maps in good faith, and not frustrate it. Another example is cities divided by county lines. The first thing I would do, all things being equal, when chopping into a county, is to unite a divided municipality. Wouldn't you? That is just common sense to me.

So I hope you will be open minded on this sort of stuff, even at the cost of elegance. I am trying to be helpful, rather than obstructive. I am trying to keep this metric system from getting into trouble, and having a situation where a map drawer, says, well this rule is really silly, and forces one to do that one with common sense would not do, or is just bad policy.

The idea of the UCC began with a nasty map I drew that chopped up Lansing. Our attempt to correct for it started as a mandate, became a preference, but then emerged as a scoring adjustment. Eventually we tried three different scoring adjustments and found two that worked, but the third didn't (one county UCCs) so we dropped it entirely.  One of the ones that did survive is not one I was particularly enamored of (pack rule), but it is still with us.

So, I'm willing to entertain a look at any of these preferences. I hope you're prepared to view them as scoring modifiers as well. I think that is in the spirit of a Pareto choice using numeric variables and recognizing that it actually worked last year with the UCCs.

[muon2]

Here's Snohomish. Along with King and Pierce, these are the counties that must be macrochopped. At this point I can see how my erose Dem-pack district in King compares with my more traditional offering.

[muon2]

I want to follow up on what I think is a way to do this consistent with some of our past practice.

Interesting questions. I was focused more about chops that are created by these "artificial" subunits, which are less salable out there. That is a very substantial concern of mine. So there is the issue of whether to create distinctions between real towns and cities, and these constructs, particularly cities I guess. I am not sure what to do about that. It is more an empirical issue. In all events, a 0.5% limitation is way too small. That is only 3.500 people or so and trivial. That is not going to do much good at all. I would think a limitation of 5%, or 35,000 people, is more appropriate. Tentatively, I would tend to think that the one bite rule will not apply to cities, but would to everything else, with the 5% limitation. That way, we do not get disparate results between those states that have little towns here, there and everywhere (NY has them taking in all real estate, as does Michigan), as opposed to states that do not.

At 5% it becomes a macrochop of the subunit, so you are suggesting (I think) that any simple chop of an unincorporated subunit, or neighborhood of a macrochopped city is subject to the one bite rule. Presumably that means townships in the Midwest, too. If so, there would seem to be no incentive to keep them intact.

As I noted above 5% is a simple chop and does not create a macrochop. We have the practice that with a simple chop we don't look at the subunits. However we don't want wholesale chopping of subunits just because there isn't a macrochop. How about this formulation:

A simple chop may be comprised of a number of subunits, no more than one of which may be chopped. A simple chop counts as one chop in scoring whether or not a subunit is chopped. Each disconnected fragment of a simple chop counts as a single unit for determining erosity.

A macrochop requires that each subunit of the macrochopped unit be treated as a unit both for assessing chops and for determining erosity. That is in a macrochop any chop of a subunit increases the chop score.

A chop of a subunit that only separates disconnected parts of the subunit does not increase the chop score. This applies in units with either a simple chop or a macrochop.

This gives one-bite relief for chops that stay below 5%, but not in units that are macrochopped. It gives multi-bite relief in macrochops for disconnected fragments of a subunit.

[muon2]

Here's an illustration of the suggested rule in my previous post. Kitsap has 4 cities and 6 school districts, one of which is the same as a city (Bainbridge Island) and one is a fragment from a neighboring county North Mason. The city of Bremerton divides the Bremerton SD into a lot of fragments.

Edit: there's a 4 person disconnected part of Bremerton adjacent to Port Orchard not shown on the map below. The magenta outline north of the SW part of Port Orchard indicates the location.



Populations by subunit:
Bainbridge Island (dark blue): 23,025
Bremerton city (red): 37,546
Port Orchard city (green): 11,144
Poulsby city (dark slate): 9,200
Bremerton SD (pink): 10,420
North Kitsap SD (light blue): 37,738
Central Kitsap SD (medium blue): 64,844
South Kitsap SD (light green): 56,861
North Mason SD (orange): 345

Note that Bainbridge Island is only connected to North Kitsap and North Mason is only connected to Central Kitsap. Only one fragment of Bremerton SD (southernmost) is connected to the main part of South Kitsap, and the SD fragments south and west of Kitsap lake are only connected to Bremerton city. Bremerton is not connected to Port Orchard. There are mountains and a lot of water without bridges or ferries here.

Now consider the map for the Puget sound area I posted much earlier in this thread. Assume that it corrects the error that left out the sparsely populated part of SW Bremerton from CD-2 (green). It has a macrochop of Kitsap putting about 160K in CD2 (green) and 91K in CD6 (slate). To do so it chops the Central Kitsap SD for a total of two chops, one for the county and one for the subunit. Other arrangements lead instead to a chop of Bremerton city.



Here's a map that shifts Grays Harbor and Mason to CD2 and reduces the chop into Kitsap to 26K. That's a simple chop, so the fact that it chops North Kitsap SD doesn't count against it by my suggested version of one-bite. That reduces the Kitsap chop to just 1.



However, because of connections, CD9 slid up into Kitsap and cost a UCC pack. Thus the net chop score including UCC is a wash between the two plans. That seems like a reasonable balance, leaving it to erosity to discriminate between the two plans.

I think it is good policy to discourage large chops when small chops will do. Macrochops define the line between large and small chops. If the one bite rule applies to macrochops as well then there is no incentive not to macrochop Kitsap. I don't like that result. Maybe the issue only applies to subunit chops in city subunits (basically a subunit of a subunit) as per the Phoenix example; I haven't seen as a good an illustration elsewhere.

The issue returns to the subject of population accidents. But this is exactly what we addressed with train's concern in MI last year. If a chop is used to overcome a population accident, and it reduces inequality then it rewards the chopper. The one bite in macrochops could do the same thing. There are judges who think lower inequality is important; in the 7th circuit when faced with a choice between VRA compliant maps in the 90's  they explicitly picked the one with the lower inequality. I think this should be given consideration.

[muon2]

If the folks in Washington were to ask whether this methodology could also be used for legislative districts, what would you tell them? What if any adaptations would be needed?

Washington has 49 legislative districts, which elect one senator and two representatives by position. The constitution requires nesting of representative districts within senate districts, so formally, senate and representative districts are coterminous.

It is constitutional to divide a senate district into two distinct representative districts. Such division need not be done on a statewide basis.

So consider several possibilities:

(1) Draw 49 legislative districts as now.
(2) Draw 49 senate districts, then divide each into two representative districts.
(2a) Make an object decision on whether to actually split a senate district into the representative districts.
(3) Draw 98 representative districts, then pair into senate districts.
(3a) Make an objective decision on whether to dissolve the pair of representative districts into a single district.

Illinois also has nested representative districts in senate districts, but must create two separate house districts in each senate district. When I applied my technique to IL legislative districts, it worked better to build the senate districts then make the division into house districts. When creating districts, I like grouping whole counties in a whole number of districts just like we did with FL. If I use the house districts, I'm tempted to use an odd number of house districts in a region, and that leads to more chops for the senate.

So, for WA I would lean towards your option 2a. Create the senate districts, then allow the state to determine if it wishes to divide the senate districts into pairs of representative districts.

[muon2]

3.Did the 7th circuit case involve CD's? Most states require absolute equality of population with CD's. If equality suddenly became important, we would suddenly be getting more chops over a few people. Not good.

The 7th circuit case involved CD's at a time when IL was not strict about absolute equality. It is still not codified to have absolute equality, nor are any other provision codified other than language related to minority districts. However, after that decision both parties have drawn maps with absolute equality to remove the possibility they would be beat in court by an inequality argument, since there are no other codified criteria.

In the WV case there were a number of whole county plans presented with lower deviation. The actual plan because it had the least deviation for the particular state goal which is to minimize the shifted population. The decision noted that if it were solely to avoid splitting counties they would have been required to use a whole county plan with less deviation. That's what they meant by as nearly practicable.

What I worked out for MI and could generalize is a table derived from actual data regressions that limits the ability to chop just to get to absolute equality. A simple example is to consider that absolute equality requires in most cases a number of county chops equal to one less than the number of districts. If the inequality table provides fewer points to gain than the number of districts (ie the number of chops required), then one can't win by chopping to exact equality.

[muon2]

When I applied my technique to IL legislative districts, it worked better to build the senate districts then make the division into house districts. I would have thought the opposite. When creating districts, I like grouping whole counties in a whole number of districts just like we did with FL. If I use the house districts, I'm tempted to use an odd number of house districts in a region, and that leads to more chops for the senate. You want more chops?

Maybe what is driving you, is that you want an even number of seats in the Senate, and an odd number in the House. More likely, I am just in a state of confusion.

Suppose I build house districts first and cluster them by regions of whole counties to minimize chops at that level. Some of those regions will have an odd number of HDs. When there is an odd number of HDs, then at least one of the SDs has to span out of the region to collect its second HD. That becomes an SD that is chopping a region and thus chopping a county. I found that I ended up with more county chops going house first, than I did by going senate first.

I should clarify that there is no problem if the house districts can be made of whole counties - that is single district regions. That isn't true in most of IL where chops are required, but where it is true that single house district can be shifted in the way that minimizes the number of odd house district groups.

[muon2]

Well, you tossed a lot on the table there.

1. Before, you had a flat ban on chops of subunits not involving a macro-chop, and otherwise such ordinary chops had no penalty other than an erosity issue potentially. Now the flat ban is gone (good), and we have no incentive to avoid a chop, other than erosity considerations. Does that make any sense? I think we are back to a preference issue (outside of subunits of subunits, where I do want an unbridled one bite rule).

2. Your problem here is that the macro-chop penalty for the chop of a subunit was offset by a pack penalty (the latter of which you don't like much anyway). So to weaken the pack penalty, you want an additional incentive to not macro-chop. Absent the pack penalty, this issue you have illustrated would not be in play. So you grab onto the one bite rule, in a context about which I am not particularly concerned, to effect your preferences, as opposed to where I am concerned. Very clever! I guess I would need to see how much you have let the genie out of the bottle with your weakening of the pack penalty, and for that matter, the cover penalty. I do agree that outside of a chop of a subunit of a subunit, that a macro-chop is worse than an ordinary chop in theory. It certainly should be a preference item. Whether it should go beyond that, and weaken the pack and cover penalty regime, is another matter.

What I am saying is that the UCC rules which began as a preference were replaced by a score modifier. The initial regime from MI would require that no subunit could be chopped without penalty, and simple chops couldn't chop them at all. I'm proposing the conversion of the preference for smaller chops into a score modifier as well. I think that the price for an unneeded macrochop ought to be worth the price for a missed UCC pack or cover.

I also disagree with your characterization of school districts as arbitrary. They are no more arbitrary than townships in states that allow cities to annex portions away. The only difference I see is that DRA didn't load them and not every county is strict about following their lines for precincts (I found that Kitsap was quite strict in this regard.)

[muon2]

Well, on the inequality thing, by manipulating the point count, you may have a proxy for a preference rule. But generating more chops of counties or subdivisions to minimize inequality is just silly in my view. Thus the preference needs to be a the bottom of the heap.

On this I think you are in the minority. All of the other mappers on this site have given weight to inequality. jimrtex had a model to judge regions based on minimizing population shifts between them. I used a point score based on data regression. train wanted stronger rules to protect equality than we eventually used. Lewis was quite adept at seeking and finding combinations to minimize inequality. I don't see the case to put it at the bottom of the heap.

[muon2]

Well, on the inequality thing, by manipulating the point count, you may have a proxy for a preference rule. But generating more chops of counties or subdivisions to minimize inequality is just silly in my view. Thus the preference needs to be a the bottom of the heap.

On this I think you are in the minority. All of the other mappers on this site have given weight to inequality. jimrtex had a model to judge regions based on minimizing population shifts between them. I used a point score based on data regression. train wanted stronger rules to protect equality than we eventually used. Lewis was quite adept at seeking and finding combinations to minimize inequality. I don't see the case to put it at the bottom of the heap.

That's OK. In the end, the public square will decide these issues, rather than an electorate comprised of three people. I wonder if Train would still want extra chops to reduce inequality. Would he really want to place it above SKEW for example? I wonder if Jimrtex wants extra chops to reduce inequality. Has he voted yet on this one?

How far up the preference tree do you want to place inequality? Just curious.

I put INEQUALITY equal to SKEW and POLARIZATION in preference. They all have fundamental scores in the rubric. The basic metric compares CHOPS to EROSITY. How a state would choose to use other scores is a local decision that can be controlled by modifying the basic metric scores with those other rubric scores. Modifying the basic scores serves to shift the Pareto curve in a way that puts more weight on other parts of the rubric, thus favoring local priorities.

My preference is to use just CHOP and EROSITY to judge a set of WA plans based on the current subunit design. They are sufficient to measure the scores. The scoring would list the other rubric elements as we did in MI. We can then see if the system is producing a good Pareto set, or if modifying the basic metric gives a better set.

[muon2]

I am not clear at all why you are jettisoning what you are jettisoning for Washington, other than it must doesn't matter in practice - maybe. And now you have a free chop regime, with no preference to avoid such, in order to weaken the pack and cover rules, but will also entail collateral damage. One can just do a gratuitous chop because one can. So this negotiation is not going very well, alas. But I bet I can get Train's vote on this inequality thing, given how things have evolved. Whatever. I am not sure what you are getting at with "local priorities." The more one has different rules for different states, the more one worries about unilateral disarmament concerns.

What am I jettisoning? I propose doing exactly what we did in MI. We define the rules for chops and erosity by way of subunits. We score a bunch of plans based on those two values alone, and at the same time show the scores for the other rubric variables. If the plans in the Pareto set are lacking in some way we should see it and perhaps correlate it to one of the other variables. After that we can make appropriate adjustments to the variables. The final step has always been to hand a set of qualified maps to a decision making body along with the metrics that both were used for the selection as well as those that were not used. Where is there a difference today?

[muon2]

I am not clear at all why you are jettisoning what you are jettisoning for Washington, other than it must doesn't matter in practice - maybe. And now you have a free chop regime, with no preference to avoid such, in order to weaken the pack and cover rules, but will also entail collateral damage. One can just do a gratuitous chop because one can. So this negotiation is not going very well, alas. But I bet I can get Train's vote on this inequality thing, given how things have evolved. Whatever. I am not sure what you are getting at with "local priorities." The more one has different rules for different states, the more one worries about unilateral disarmament concerns.

What am I jettisoning? I propose doing exactly what we did in MI. We define the rules for chops and erosity by way of subunits. We score a bunch of plans based on those two values alone, and at the same time show the scores for the other rubric variables. If the plans in the Pareto set are lacking in some way we should see it and perhaps correlate it to one of the other variables. After that we can make appropriate adjustments to the variables. The final step has always been to hand a set of qualified maps to a decision making body along with the metrics that both were used for the selection as well as those that were not used. Where is there a difference today?

Is this a process comment, or a substance comment? By that I mean, is this a protocol, en route to fashioning the same rules for all states, and suggesting the order of the steps to get there, or a suggestion that states might end up with different rules in the end, as to what maps make the cut for consideration?

I think if at all possible the rules should be the same, which is one reason why I decided that creating artificial subunits to take in all real estate is appropriate, because some states have such subunits for all real estate, and some do not. And no states have such subunits by definition within townships and cities, and have never drawn maps based on that, which is why I think different rules are required there, for the reasons I elucidated.

My goal is to have rules for the five metrics that are independent of the states. The Pareto test to find a set of qualifying plans is based solely on the metrics.

The metrics require county subunits and they obviously will not be derived the same way for all states. I would like to have clear principles that will lead to relatively few variants for the creation of subunits. I think that's the substance part of what I'm doing so far with WA.

To test changes to the five metrics I'd like to have a number of plans from different mappers. Then I compare the baseline metric to the proposed alteration. We can assess those results numerically and see if the proposed change is having a positive effect. Since the idea is to put out a set of plans, a positive effect is usually assessed by noting that undesirable plans that gamed the system have been excluded. I think that creating alternatives to test and the testing itself, as we did for UCCs, is the process part.

[muon2]

When I applied my technique to IL legislative districts, it worked better to build the senate districts then make the division into house districts. I would have thought the opposite. When creating districts, I like grouping whole counties in a whole number of districts just like we did with FL. If I use the house districts, I'm tempted to use an odd number of house districts in a region, and that leads to more chops for the senate. You want more chops?

Maybe what is driving you, is that you want an even number of seats in the Senate, and an odd number in the House. More likely, I am just in a state of confusion.

As a matter of fact, the system of fusion used in New York is a con. See the ballot for Hudson Ward 4 alderman as an example. It's not just you, there are millions more living in the same situation.

I think nesting is a bad idea. It seems like one approach - splitting, or joining should work. But the reality is that population concerns require some pretty ugly splits. If a house district has an ugly split, then joining it with another house district means that the senate district will also have the ugly split. If the reverse is done, an ugly senate split, will mean at least one of the house districts has an ugly split.

California has a modest requirement for nesting. I think it was expected that sometimes you might want to use a city boundary for a senate district, but that might not work so well for assembly districts. But what really caused the deviation was VRA requirements. For example, you can probably imagine where a minority-opportunity senate district, would divide nicely into a packed-assembly district, with the other assembly district being much whiter.

I agree with jimrtex on his observation about nesting. Dividing a compact district into two nested house districts generally results in less compactness. Joining two compact districts to form a senate district generally results in less compactness. The best compactness/lowest erosity comes when each chamber's plans are drawn independently of the other.

[muon2]

Here's Thurston using cities and towns, then filling in by school district. As strict as Kitsap is nesting precincts in school districts, Thurston is weak. I guess the Clerk there isn't as concerned about multiple ballots in a precinct.



There are 7 munis and 9 school districts:

Lacey (teal): 36,709
Olympia (green): 46,478
Rainier (forest green): 1,794
Tenino (tan): 1,695
Tumwater (purple): 17,371
Yelm (red orange): 6,848
Bucoda (brown): 562
North Thurston SD (cyan): 51,896
Olympia SD (light green): 19,363
Rainier SD (lime green): 2,600
Centralia SD (blue): 439
Rochester SD (dusty rose): 14,720
Tenino SD (beige): 8.046
Tumwater SD (lilac): 21,236
Yelm (orange): 15,948
Griffin (pink): 6,559

[muon2]

That light green thing near the top has an unfortunate shape. Is that due to geographic barriers?

The boundary between the precincts in light green and lilac is the actual SD boundary, too. There's not really any population in the western part of that light green area, it's all mountain. However, the Olympia SD (light green) actually includes the western part of state hwy 8 in the pink precinct, but not the part at the junction with US 101. I'm guessing that the western pink precinct is drawn to connect all the parcels along s.h. 8 even though that includes part of western Olympia SD.



You can see how poorly the precincts match here. Kitsap would probably have had an additional precinct with few if any residents to cover the separate SD along s.h 8 if they were in charge.

[muon2]

Here's Yakima. As with the others I used the cities and towns as separate subunits (but not places), then school districts for the remaining unincorporated areas. Precincts were assigned to the district with the most population, as best as I could tell from the map (I don't have the block data that jimrtex has.)



The Yakama Indian Reservation takes up a lot of the land in the county. Here are the subunits completely outside the reservation.

Naches town (brown) 795
Naches SD (peach) 7,331

Tieton city (dark blue) 1,191; 57.6% HVAP
Highland SD (slate blue) 4,410; 29.6% HVAP

Selah city (violet) 7,126
Selah SD (lilac) 14,460

Yakima city (dark green) 91,023; 33.5% HVAP
Moxee city (dark aqua) 3,308; 34.0%HVAP
West Valley SD (lime) 11,890
Yakima SD (green) 451
East Valley SD (aqua) 11,352

Union Gap city (tan) 6,047; 38.1% HVAP
Union Gap SD (yellow) 104

Zillah city (olive) 2,964; 35.0% HVAP
Zillah SD (khaki) 2,194; 25.0% HVAP

Sunnyside city (orange-red) 15,854; 75.9% HVAP
Sunnyside SD (salmon) 8,883; 56.4% HVAP

Grandview city (teal) 10,862; 73.6% HVAP
Grandview SD (turquoise) 3,390; 47.6% HVAP

Mabton city (deep pink) 2,286; 89.4% HVAP
Mabton SD (pink) 1,224; 63.6% HVAP

Bickelton SD (coral) 66; 27.3% HVAP

There's an interesting issue that arises here due to the Yakama Indian Reservation. It is served by 4 separate school districts (Mabton SD covers an unpopulated part of the reservation), but only Mount Adams is entirely within the reservation. The cities in the reservation are overwhelmingly Hispanic and less than 20% of the native population lives in those reservation cities. Here are the stats for the reservation subunits:

Harrah town (indigo) 625; 16.8% NVAP; 51.7% HVAP
Mount Adams SD (orchid) 4,257; 48.9% NVAP; 21.6% HVAP

Wapato city (dark red) 4,997; 6.5% NVAP; 79.4% HVAP
Wapato SD (dark salmon) 8,619; 24.2% NVAP; 39.8% HVAP

Toppenish city (medium blue) 8,949; 7.4% NVAP; 78.5% HVAP
Toppenish SD (light blue) 4,949; 20.6% NVAP; 44.1% NVAP

Granger city (green) 3,246; 1.5% NVAP; 83.6% HVAP
Granger SD (light green) 2,353; 11.9% NVAP; 44.1% HVAP

The reservation as a whole subunit, including munis has 29,978 people including a native population of 6,571. Here are some questions.

Should the reservation be kept together as a single subunit?
If the reservation is not a single subunit, should there be a scoring benefit to keeping the subunits that cover it together?
There is a large rural and small town Hispanic population in some of the subunits, so should these subunits be treated like the rural black counties of the South and maps be penalized that chop the cluster?

[muon2]

I don't get the connectivity issue.

The reservation should be treated like a hybrid county, just like I suggest in AZ. That addresses the first two points.

"There is a large rural and small town Hispanic population in some of the subunits, so should these subunits be treated like the rural black counties of the South and maps be penalized that chop the cluster?"

God no. We have the VRA. Leave it at that. What you are suggesting is moving the VRA towards the Florida law with minority opportunity districts.

We did spend some time when we looked at AL and we respected contiguous minority counties in the Black Belt the same way we did UCCs. The idea was that rural minority interests were on a par with urban interests. We also wanted to avoid hops over white areas just to link Black areas. One exception was that the AL Black Belt had a mandatory chop to deal with the fact that it extended all the way across the state, but otherwise we avoided chops to the Belt. We applied the same reasoning when we looked at LA.

Is that a bad idea to your mind now?

[muon2]

I don't get the connectivity issue.

The reservation should be treated like a hybrid county, just like I suggest in AZ. That addresses the first two points.

"There is a large rural and small town Hispanic population in some of the subunits, so should these subunits be treated like the rural black counties of the South and maps be penalized that chop the cluster?"

God no. We have the VRA. Leave it at that. What you are suggesting is moving the VRA towards the Florida law with minority opportunity districts.

I think there will have to be an objective way to draw minority opportunity districts. If you just try to "comply with federal law" or "comply with the VRA" you are guaranteed to end up in court.

My inclination is to start with counties that are 40%+ CVAP of the target minority, but exclude level 2 units that are less than 20%+ minority. Then repeat for 2nd level units, etc.

Conditionally, add areas that improve connectivity while maintaining an overall 40%+ CVAP. Determine the number of districts that will fit in the area (truncate down). Remove the surplus over that needed to create the whole number of districts. Divide the area into the number of districts.

To add to jimrtex's comment, in IL before the 2010 election there was concern that there would be a Pub gov. In anticipation there was legislation to specifically encourage coalition and crossover districts. The minority groups were and remain concerned going forward. If there is a question of selling the idea, minority interests will want to see protection of areas that don't rise to the level of the VRA but could significantly influence election outcomes.

[muon2]

Here's Yakima with the Yakama reservation kept whole, and then cities and towns as separate subunits elsewhere.



Naches town (brown) 795
Naches SD (peach) 7,287

Tieton city (dark blue) 1,191; 57.6% HVAP
Highland SD (slate blue) 4,410; 29.6% HVAP

Selah city (violet) 7,126
Selah SD (lilac) 12,460

Yakima city (dark green) 91,067; 33.4% HVAP
Moxee city (dark aqua) 3,308; 34.0% HVAP
West Valley SD (lime) 11,890
Yakima SD (green) 451
East Valley SD (aqua) 11,352

Union Gap city (tan) 6,047; 38.1% HVAP
Union Gap SD (yellow) 104

Zillah city (olive) 2,964; 35.0% HVAP
Zillah SD (khaki) 2,194; 25.0% HVAP

Sunnyside city (orange-red) 15,854; 75.9% HVAP
Sunnyside SD (salmon) 8,883; 56.4% HVAP

Grandview city (teal) 10,862; 73.6% HVAP
Grandview SD (turquoise) 3,390; 47.6% HVAP

Mabton city (deep pink) 2,286; 89.4% HVAP
Mabton SD (pink) 1,224; 63.6% HVAP

Bickelton SD (coral) 66; 27.3% HVAP

Wapato SD (dark salmon) 976; 35.2% HVAP

Toppenish SD (light blue) 1,982; 52.5% HVAP

Granger city (green) 3,246; 83.6% HVAP
Granger SD (light green) 1,813; 50.8% HVAP

Yakama Indian Reservation (orchid) 29,978; 21.8% NVAP; 54.8% HVAP

[muon2]

I don't get the connectivity issue.

The reservation should be treated like a hybrid county, just like I suggest in AZ. That addresses the first two points.

"There is a large rural and small town Hispanic population in some of the subunits, so should these subunits be treated like the rural black counties of the South and maps be penalized that chop the cluster?"

God no. We have the VRA. Leave it at that. What you are suggesting is moving the VRA towards the Florida law with minority opportunity districts.

I think there will have to be an objective way to draw minority opportunity districts. If you just try to "comply with federal law" or "comply with the VRA" you are guaranteed to end up in court.

My inclination is to start with counties that are 40%+ CVAP of the target minority, but exclude level 2 units that are less than 20%+ minority. Then repeat for 2nd level units, etc.

Conditionally, add areas that improve connectivity while maintaining an overall 40%+ CVAP. Determine the number of districts that will fit in the area (truncate down). Remove the surplus over that needed to create the whole number of districts. Divide the area into the number of districts.

To add to jimrtex's comment, in IL before the 2010 election there was concern that there would be a Pub gov. In anticipation there was legislation to specifically encourage coalition and crossover districts. The minority groups were and remain concerned going forward. If there is a question of selling the idea, minority interests will want to see protection of areas that don't rise to the level of the VRA but could significantly influence election outcomes.

At least when it comes to Congress, what you are doing is not gong to elect more minorities. And nobody has been pushing for what you are doing either. It's novel. The way to do it was what I suggested. In most places, getting more minorities elected, gets more Pubs elected too, unless you do something grotesque like that CD in Arizona going from Phoenix to Tucson to Yuma. And the minority politicians don't really care about districts unless they are actually performing. Absent that, and they are useless. Just ask Corrine Brown if you don't believe me.

I suspect we will have to see how SCOTUS rules on VA, and whether some guidance on performing minority districts emerges.

It is the minority interest groups who push to keep their communities intact. I'm not talking about minority politicians, whose goals may be more political. There was a great deal of testimony in IL to that effect in 2011, so I didn't think my question was so novel. I think the minority interest groups would complain mightily if a system was totally blind to their interests. Aren't we catering to some of those interests by keeping reservations intact as hybrid counties?

[muon2]

Ah, using Reservations as the nose in the tent. You're thinking like a lawyer now. I like it! But no, Reservations have separate governance mechanisms, and really are hybrid counties. And their impact is limited vis a vis giving them special consideration given their special status. Given those twin considerations, my hybrid mechanism is perfect Goldilocks. It gets the balancing test right, with no collateral damage. Thanks for your nose visiting the tent.

CA is going to need special consideration the way you are going, for whites, or it will be reverse discrimination. That is where this will end up in the end with all of this, and we don't need or want to end up there.

Not that it matters, or will change my mind, but where have "minority interests" weighed in, when it did not involve a "performing district" being on the line vis a vis the decision made?

The one clear example that I know is the substantial testimony from the Asian American Institute and their allies in IL during legislative redistricting hearings in 2011. Chinatown on the south side, Albany Park on the north side, and a cluster of neighborhoods in the the near north suburbs were presented as areas with Asian majority neighborhoods that should not be chopped as happened in 2001. None of the areas were going to qualify to be a performing district, but given Chicago's ethnic coalitions, they wanted to be preserved as an influential bloc.

[muon2]

To revisit one of my original questions, I'm going to look at my King CDs from two different plans. Here's the map of King subunits using cities first then filling in by school district.



This was my initial plan.


I've slightly adjusted the lines to conform with the subunits, removed the Kitsap macrochop, and I've zoomed on CD 10 showing the links cut by the boundaries.


The local links cut within King are in yellow and the regional links to Kitsap and Pierce are red following state highways and the ferry.

The erosity of the King districts becomes
CD 7 erosity = 10
King-Kitsap : 2 links (ferry)
King-Snohomish :  2 links (Snohomish is macrochopped and Shoreline links to two separate subunits)
King-King CD8 : 3 links (Shoreline and Seattle to Lake Forest Park, link to Mercer Island)
King-King CD10 : 3 links (from Seattle)

CD 8 erosity = 25
Snohomish-Snohomish : 6 links (not shown)
King-Snohomish : 2 links (Bothell and Woodinville)
King-King CD7 : 3 links
King-King CD10 : 11 links (in yellow)
Kittitas-Chelan : 1 link
Kittitas-Grant : 1 link
Kittitas-Yakima : 1 link

CD 10 erosity =18
King-King CD7 : 3 links (in yellow)
King-King CD8 : 11 links (in yellow)
King-Pierce : 4 links (in red)

Removing the duplicate cuts counted within King, the total erosity of the three is 36.

[muon2]

The other map I posted was a Dem pack of CD 10 designed to make a competitive CD 8. I thought it looked pretty erose, but without subunits it was technically less erose then my first version.



Connections to other counties don't change, so here's the recomputed erosity.

The erosity of the King districts becomes
CD 7 erosity = 10
King-Kitsap : 2 links (ferry)
King-King CD10 : 8 links (in yellow)

CD 8 erosity = 40
Snohomish-Snohomish : 6 links (not shown)
King-Snohomish : 2 links (Bothell and Woodinville)
King-King CD10 : 25 links (in yellow)
King-Pierce : 4 links
Kittitas-Chelan : 1 link
Kittitas-Grant : 1 link
Kittitas-Yakima : 1 link

CD 10 erosity = 35
King-Snohomish :  2 links
King-King CD7 : 8 links (in yellow)
King-King CD8 : 25 links (in yellow)

Removing the duplicate cuts counted within King, the total erosity of the three is 52. Using subunits the plan is significantly more erose the original offering.

[muon2]

Here's a detailed application for a Snohomish CD. The county is big enough to require a macrochop, so here's the subunit map using school districts except for those areas in incorporated munis which take precedence.



Because there is a macrochop there are three links to Skagit from separate Snohomish subunits, 1 link to Island by ferry, 1 link to Kitsap by ferry, and 6 links between Snohomish and King subunits.

One rule is that all units, or subunits when macrochopped, must be connected. In order to connect to the Skyhomish school district subunit in King, CD 8 has to include the Index and Sultan school districts in Snohomish. It can then use the Monroe school district to connect back to the rest of King. Here's an illustration with the school district boundaries shown with black lines, except for beige lines when they pass through an incorporated muni like Monroe or the part of Everett that is for its water reservoir.



I've used red and pink lines to show links between subunits across the county line, with red for those on the border between the two districts and pink for an internal link. I've used yellow lines for local links between the districts within Snohomish. Note that the Monroe SD was chopped, but the chop line along the river has no bridge so there is no link between those two pieces of Monroe SD.

To get the erosity for the dark blue CD 1 which is entirely within Snohomish, I add the 5 links to Skagit, Island and Kitsap, the 4 links between the Snohomish CD and King (two links to Shoreline are not on the detailed map), and the 6 yellow links cut within Snohomish. The total erosity is 15.

There are some open questions here. I used a brute force approach that created a link to an unincorporated SD if there was a connection to population in the SD. That ignored the idea of the node for the SD. I did that here to illustrate that this division of the subunits, typically put the SD node inside a muni subunit. The Monroe SD offices are in the city of Monroe, so any path to the SD node forces one through the city technically invalidating the path. That doesn't make much sense so I've looked at links to the largest population VTDs within those unincorporated subunits.

That brings up a second issue. The Woods Creek part of the Monroe SD is contiguous but disconnected from the rest of the SD, but only because the city is a separate subunit. In this illustration I allowed connections to it independent of the rest of the SD. It gets more complicated since the Maltby area is contiguous but disconnected as well, even keeping the city of Monroe in the SD for connection purposes. I show a link from the Northshore SD to the west of Maltby, but maybe it shouldn't exist since one can't get from Northshore SD to the Monroe SD offices without going through part of the Snohomish SD.