Using Urban County Clusters To Guide Redistricting

[muon2]

The objective is to make districts from whole counties.

I strongly disagree with this.  The objective is to make fair, sensible districts, and using counties as a proxy for that is better thought of as merely a means to an end, rather than the end itself.  There are cases where other factors can and should override (such as, but not necessarily limited to, VRA concerns of course).

If there is not a strong objective of whole counties, then the districts are neither fair nor sensible.  

If whole counties are the only objective, then the districts are neither fair or sensible.  

If NO county cuts were made to this map:



It would be both fair and sensible.

No it wouldn't, because it would flount the VRA and obviously fail the Gingles test.  You may think that the state would defend it in court, but if they do so they are wrong, and I will eat all of my hats if the court doesn't concur.

I agree that this would be a tough case to defend. When I added up the numbers for 2008 and 2012 I find that this district is only D+1.1 despite 46.4% BVAP. There are 40% BVAP counties in this district that voted for Romney. 46.4% in AL is very different from the same percentage in Chicago or Cleveland where there is a large population of willing white crossover voters. It doesn't seem likely that one can make the case that this district provides sufficient opportunity at D+1 for the minority to elect their representative of choice as is required under section 2.

[jimrtex]

If NO county cuts were made to this map:



It would be both fair and sensible.

No it wouldn't, because it would flount the VRA and obviously fail the Gingles test.  You may think that the state would defend it in court, but if they do so they are wrong, and I will eat all of my hats if the court doesn't concur.

You're assuming that the SCOTUS is either fair or sensible.

The Supreme Court of Canada ruled that the right of effective representation trumps that of equal representation.  That is fair and sensible.  

Splitting counties dilutes and degrades representation: voters don't know who their representatives are, representatives don't know who their constituents are.  It may not be practical to have an office in the area, county residents have less opportunity to run for office since their potential political base is ripped asunder.   It will be more expensive to campaign since media markets will be split.  Representative may pay less attention to voters, since there are fewer voters within the district from the area.   That is neither fair, nor sensible.

The reason states strive for precise equality is because that lets them do redistricting porn.  The SCOTUS will look at a map, and if it has equal population districts, they don't have a basis for striking it down.

But we don't want to do redistricting porn.  The most sensible and fair way to do that, is to use whole counties.  If we use counties sometimes, and not others, the SCOTUS will knock us down quicker than you can say Bandemer.  Since we aren't a legislature, we can't make up our justification up as we go along, like they did in West Virginia.

And there is no reason whatsoever that a state might not adopt a different set of objectives than those adopted by the West Virginia legislature.  The US Constitution says that each legislature shall prescribe the manner of electing its representatives, subject to override by Congress.   Congress has said, "Ugh, states make districts".  We are thus in compliance with congressional directives, and certainly don't have to use the West Virginia objectives, nor does West Virginia have to use ours.

We don't have to preserve existing districts, nor protect incumbents, nor make as few changes as possible.  Disregarding political outcomes, and incumbent protection is a legitimate state objective.

Our process specifically provides for consideration of alternative plans.  If there is a whole county plan, with substantially better equality and compactness, it would eliminate my plan from consideration.  And our plan would be that chosen by the people themselves.

So a possible plaintiff might be able to demonstrate a whole-county plan that has less deviation.  If they failed to propose it during the actual redistricting process, they might be barred from even pursuing a case, given the substantive public openness of our process.

Our plan would better vindicate our legitimate state objectives:

Whole county districts;
Practicable equality and compactness; and
Popular selection.

Please state your VRA Section 2 claim.  I am willing to stipulate to the following results:

President (2012) Obama 176,276 (54.4%) Romney 145,687 (45.0%)

Senator (2010) Shelby 122,799 (51.5%) Barnes 115,534 (48.4%)
Governor (2010) Sparks 134,391 (55.9%), Bentley 104,801 (43.6%)

President (2008) Obama 171,770 (52.8%) McCain 151,372 (46.7%)
Senator (2008) Sessions 159,396 (50.1%) Figures 158,815 (49.9%)

And also that Obama (2012), Sparks (2010), and Obama (2008) carried Jefferson County, and that 49% of the 18+ black population is contained in 2 of the 7 (29%) districts.

[muon2]

Splitting counties dilutes and degrades representation: voters don't know who their representatives are, representatives don't know who their constituents are.  It may not be practical to have an office in the area, county residents have less opportunity to run for office since their potential political base is ripped asunder.   It will be more expensive to campaign since media markets will be split.  Representative may pay less attention to voters, since there are fewer voters within the district from the area.   That is neither fair, nor sensible.

This would seem to argue that one would measure the size of a chop as a fraction of the county size, not the size of a chop compared to the ideal district size.

[Brittain33]

I would love a one-sentence-each summary of what JIMRTEX, muon, and Torie each are proposing. This is impossible to follow as an outsider.

[jimrtex]

I would love a one-sentence-each summary of what JIMRTEX, muon, and Torie each are proposing. This is impossible to follow as an outsider.

Jimrtex one sentence summary: Ordinary citizens choose the plan.

There would be a state agency that would administer the process, preparing the census and other data, and facilitating public participation, but they would not be devising or selecting the plan.   The agency might be the legislative council, secretary of state, board of elections, state statistical agency, or a task-specific body.

The process would occur in stages:

I. Proposals would be made to divide the state into regions comprised of contiguous whole counties, with each region having a population equivalent to an integer number of districts, ideally one district per region.  Regions would be required to respect area of high urban concentration, to avoid excessive division of metropolitan areas, excessive linking of metropolitan areas, or consequent division of rural areas, be generally compact, and comply with Section 2 of the VRA.

Plans would be rated primarily on the amount of county splitting that would be necessary to adjust them to bring them into acceptable levels of equality.  In states like Iowa and West Virginia, it is possible to produce plans with no county splitting.  In Alabama, a minimal amount is necessary.  My proposed map would require splitting off about 36,000 persons from their county into another district.  This is 0.77% of the total state population.  So an Iowa or West Virginia can get to 100%, I was only able to get to 99.23% (I am playing two roles, one as advocate for a process, and providing an example of one possible proposal under that process).

There could be multiple submission deadlines.  Ideally, the process is somewhat collaborative, where someone could propose an improved version of a plan, or a different plan.

II. The administrator would select qualifying plans, and present them to a representative panel of the state's voters.  This panel would number in the 100s, and would be selected like a petit jury pool, at lot, chosen in such a way that all counties (and areas within counties) are represented.  It is impractical to have a full-blown plebiscite.  But you can get the same results with a representative sample.

Each panel member would rate a plan based on the district they would be place in under the plan.  Plans would be ranked on the composite.   Any redistricting plan will have some good districts, some fair districts, some bad districts, and some really awful districts.  So we want a plan with more good districts than bad,

After the best plan, as determined by the collective judgment of the panel, it would validated whether the the panel members in each district under the plan agree with the statewide consensus.  In areas where there is a significant difference, there would be a repeat of the first stage to see if an improvement for those areas can be made.

III. There would then be proposals made to refine the plan, splitting the counties necessary to make it come into the necessary degree of equality.  These decisions are more local in nature, and would be decided by the panel members in the respective regions.

[jimrtex]

Splitting counties dilutes and degrades representation: voters don't know who their representatives are, representatives don't know who their constituents are.  It may not be practical to have an office in the area, county residents have less opportunity to run for office since their potential political base is ripped asunder.   It will be more expensive to campaign since media markets will be split.  Representative may pay less attention to voters, since there are fewer voters within the district from the area.   That is neither fair, nor sensible.

This would seem to argue that one would measure the size of a chop as a fraction of the county size, not the size of a chop compared to the ideal district size.

Imagine that you needed 130,000 more persons for a district.  You have 3 counties with 50,000 or one with 200,000.

If you take the three smaller counties, you have an error of 20,000, which is around 3% of a district but 40% of the 3rd county.  If you take the large county, you have an error of 70,000, which is around 10% of a district, but only 35% of the county.   But you can't say that you were making the districts as equal as practicable, particularly when it is not an irrational argument to make to not split the 3rd county at all (about a 3% error).

This could be confusing for those making proposals.  You would first tell them to use whole counties, like in Iowa and West Virginia, but if that doesn't work out suggest a plan that would have greater inequality.

At the county level, all voters in a split county suffer some vote dilution.  In the larger county, there might be less individual harm, but it is spread over more people.  So the net effect would be about the same.  There might be a preference to splitting off X people from a more populous county.  So for my plan, it might be preferred to split Tuscaloosa, rather than Walker or Blount.  Election administrators in Tuscaloosa probably have more experience with complex elections with multiple legislative districts.  Ballots in Blount might be identical down for all offices above county commissioner.

But there are other considerations the voters who would be shifted into Blazer may work in Birmingham, or at least shop there.   But they might also have more affinity toward the rest of the Lion district.  I'd expect Cullman to prefer Blount be kept whole.   There might be more blacks in Tuscaloosa.

I'd leave the decision up to the panel members from the region, so they can apply their own subjective weighting.

[muon2]

I would love a one-sentence-each summary of what JIMRTEX, muon, and Torie each are proposing. This is impossible to follow as an outsider.

muon2 one sentence summary: Plans should be drawn to minimize both split counties and use of irregular shapes.

Mine is less about process and more about the plan. On the process I would leave it to the states to determine the body that would select the maps. My proposal would determine which maps are eligible for selection by that body.


I. A plan is constructed. All members of the public may construct and submit plans.

I.0. A plan must comply with all federal laws including the Voting Rights Act.

I.A. A state should be divided into a number of regions of connected whole counties. A connection is the shortest continuous path of numbered state or federal highways that links the administrative seats of two counties without passing through any other county. Each region must have a population equivalent to an integer number of districts such that population shifts between regions need not exceed 0.5% of the district population quota. Urban county clusters and minority county clusters must be contained within a single region.

I.B. Each region is divided into a number of districts. The number of county chops (see II.B) in a region cannot exceed one less than the number of districts in the region. The population of a district cannot deviate more than 0.5% from the population quota. Counties and county chops within a district must be connected (see I.A), and unconnected whole counties cannot be connected solely by means of a part of a chopped county. Microchops need only be connected by a local road to the rest of the district. Within a chopped county the fragments shall be made of whole county subdivisions to the extent practicable and the chops will be made to minimize the erosity (see II.C) of county subdivisions within the county based on local road connections.


II. The plan is scored.

II.A. The population range inequality from the largest to smallest district is determined.

II.B. The number of county chops in a plan is determined. A county chop occurs each time a county is split to be in more than one district. In short the number of chops is one less than the number of districts that serves the county. Microchops that are less than 0.5% of the district size do not add to the chop count. Splits of contiguous urban county clusters and minority county clusters (as would be defined for I.A) in excess of the minimum necessary add to the chop count.

II.C. The erosity in a plan is determined. The erosity is the number of county connections (see I.A) that are severed to form the districts in the plan. Chops within a county are connected if there is a local road that connects the fragments made by a chop. Those severed local connections within a chopped county add to the erosity.


III. A set of plans is passed on to the selection body.

III.A. A plan in the selection set must comply with construction rules in I.

III.B. A plan is added to selection set if it provides the lowest erosity for the number of chops or the lowest chop count for a given erosity among plans submitted. In the case of two plans with the same erosity and chop count that would qualify, the plan with the least inequality is added.

III.C. The political makeup of the districts are calculated for each plan in the set. One value will measure the polarity of the districts, that is the number of districts with a high partisan voting index. The other value will measure the partisan skew of the districts, comparing the difference in partisan outcome at the district level to the expected partisan outcome based on statewide results.


IV. The selection body picks a plan from the selection set with input from the public.

[Torie]

Mike, I still don't see why you need this regional stuff. At the end of the day, you count chops. It seems just so artificial to me. Yes, it is a useful tool to minimize chops, and all your rules can be reverse engineered.

You rule set does not seem to preclude more than one chop of an urban cluster (unless of course the urban cluster is most of the region).

You do not seem to exclude plans that hit the ball out of the park on chops, but are an epic fail on erosity, or visa versa. Maps that are all chop and no erosity, or all erosity and no chop, as it were, make it into the mix. Some thought should be given how to force those twisted sisters out of the mix.

The erosity measurement ball is still up in the air. This connectivity approach will need to pass the eye test, using examples. The burden of proof to me is on the one proposing this proxy. It either works reasonably well, or it doesn't. And how to count the road chops efficiently also escapes me. I know I can't really do it. How do you draw maps of the roads and make it work for you?  I mean, I can't test it myself, because I don't know where the roads are, or an efficient way to count them.

[muon2]

Mike, I still don't see why you need this regional stuff. At the end of the day, you count chops. It seems just so artificial to me. Yes, it is a useful tool to minimize chops, and all your rules can be reverse engineered.

You rule set does not seem to preclude more than one chop of an urban cluster (unless of course the urban cluster is most of the region).

You do not seem to exclude plans that hit the ball out of the park on chops, but are an epic fail on erosity, or visa versa. Maps that are all chop and no erosity, or all erosity and no chop, as it were, make it into the mix. Some thought should be given how to force those twisted sisters out of the mix.

The erosity measurement ball is still up in the air. This connectivity approach will need to pass the eye test, using examples. The burden of proof to me is on the one proposing this proxy. It either works reasonably well, or it doesn't. And how to count the road chops efficiently also escapes me. I know I can't really do it. How do you draw maps of the roads and make it work for you?  I mean, I can't test it myself, because I don't know where the roads are, or an efficient way to count them.

Last point first, if you use DRA zoom into the boundary of your district. On each boundary between two counties in different districts see if there's a state or federal highway that crosses the boundary. If there is it more than likely a connection and you can count it as cut. There are some rare cases where there is a state highway but none are part of a path between county seats and then you can ignore them. If you choose to count the total for each district separately you have to divide by two after you add them all together since each boundary segment is on two districts.

I have looked at length at the issue you note about the extreme points. I still haven't found a way to characterize the extremum for erosity, though I can put an upper limit on chops. My problem with erosity is that no matter how one measures it there are going to be necessary districts that are twistier than one would like due to real geography. That makes it hard to specify where the cutoff should be. I won't say that I've given up, but I'm searching in ever more remote corners for a solution.

I thought the sense was that UCCs were treated like any other chop and not as a hard rule. Did I get the sense of things wrong there?

I did say "should" for regions. Doesn't that count for something?

[jimrtex]

I don't think that there can be true public participation without access to state supported tools.

And I find it equally difficult for participation when the state has a monopoly on the tools, particularly if the tools are designed for expert use (as happened in IL in 2011). I liked your suggestion of a tool where users could point and click on whole geographic areas at a time. To be useful it would need to include the ability to turn on a layer and do the same for county subdivisions.

The key is that those making proposals understand the process model being used by the state.  While it might be useful to have access to lower level data for reference purposes, it should be understood that proposals at the first stage are to be regions of whole counties.

There should be batch input tools so that those using other tools can prepare plans than can be input, validated, and scored as well.

Clicking on counties is easy enough.  I don't know about the flow diagrams.  I do them on a sheet of paper, and then get the exact numbers using a spread sheet.  It would be possible to show the region connectivity network, and then have the user click on the links they want to keep.  Since each region must be connected, and N-1 links used, there can't be any loops, so links that would form loops, would be dimmed out.  The mechanics are simple enough, but whether a user could use it effectively is another thing.  It may be akin to the traveling salesman problem, and an optimal linkage proposed.

This format would permit transformation of the graph back into a map.   I don't think that your graph has that property, does it?

Graphs can have the properties you describe, but need not. I looked at some of those and found they introduced complexity that really didn't change the outcome.

I think it would be confusing to not be able to transform your graph back into a map.

For this map the links exist when there is a state highway that links the two counties without going through an intervening county. The only exception is for contiguous counties in the smae UCC which are presumed to be linked. Here that means Montgomery and Autauga are linked despite the highway cutting through a bit of Elmore. Before asking to add other county crossers, I would suggest asking whether one would really want to create a district that hung together solely based on that connection.

"links the two counties" is ambiguous.   "state highway" is ambiguous.  While it might be possible to define districts using linear features, where the population is migrated to the nearest highway, arterial, or collector street, it is much more practical to use areal features like counties.  Further to maintain logical consistency, you should separate parts of counties that are not directly linked by state highway.  While I did that for Grand Isle, I don't think that is a morass that you want to enter into in general.

Further, you open the process to ridicule when the bypass around Hurtsboro snips one census block off Macon County.

I will accept the charge of ambiguity. I've posted in the past at greater lengths about highway connectivity and since this post was already quite long I did not go into detail on my views on that point.

I have two comments about the Hurtsboro bypass. First is that a district that only hung together by way of that connection was likely to be erose. In fact I'm not sure any of the offerings relied on that connection so this seems like a point of form over function. Second, you have suggested local modifications to your boundary percent rule and I agree that local modifications can be in order. We both establish baselines, and as long as there is a mechanism for local adjustment I see no reason for ridicule of one baseline compared to the other.

I think I can explain to a random user why they can't connect Lawrence and Cullman directly. 

And I think I can explain why Okanogan and Whatcom can not be connected.

And I think I can explain why when they ask, "But what about Marengo and Green" or "Marengo and Dallas".

But many of the others are inexplicable.  I think I see the counties as blocks.  While the Russell-Bullock connection might not be super-strong, a glue joint would hold.  Using the highway is like saying that a wire between two blocks is holding them together.

While none of our offerings depended on the Russell-Bullock connection, it is conceivably possible.  If Alabama had 6 districts, then the Mobile district might extend to the southeastern corner, and then a Black Belt district could cross the state.  If we were doing state legislative districts, a Russell-Bullock district would not be rejected because the Hurtsboro bypass travels for a block inside Macon County.

And you may be mixing up, "it might not be a good idea" or "it would score lower" and "it is not permitted".

BTW, why are Macon and Russell connected?

I don't see how the following two are materially distinguishable, yet I don't know what my score is on the second map.

The score on the map seems pretty easy to compute to me. Since you show no chops, you only need to add up the set of cut links. The yellow line cuts 16, the red line 3, the blue line 2, the green line 5, the purple line 22 and the orange line cuts 3 links. That adds up to 51 total erosity. It's a task that's not much harder than counting chops in a plan.

I don't disagree that we are essentially measuring the same thing. That becomes especially true since you get rid of the problem of wiggly natural boundaries by measuring straight lines to county corners. It becomes a question of computational preference. This probably falls into the same category of my preference to measuring inequality by range instead of your preference to use standard deviation.

I have demonstrated that my measure can be done by computer.

Inequality among the districts of a state can not be measured by range, nor can it measure whether it is practicable to make the districts more equal.

[Torie]

You only count cut highways that connect county seats? Does the highway need to actually penetrate the city limits of the county seat if that is the case? These days, sometimes highways circumvent towns. Interstates often do that in particular. If two highways between two counties are cut (however one picks and chooses which highways count), does that count as two cuts?

What state I wonder would be the best one to test this out, once the rules are precisely known, that is one of the more likely to generate a map more erose than most states in order to keep the chops down?  Any suggestions?

[jimrtex]

Last point first, if you use DRA zoom into the boundary of your district. On each boundary between two counties in different districts see if there's a state or federal highway that crosses the boundary. If there is it more than likely a connection and you can count it as cut. There are some rare cases where there is a state highway but none are part of a path between county seats and then you can ignore them. If you choose to count the total for each district separately you have to divide by two after you add them all together since each boundary segment is on two districts.

Why do show Macon and Russell as connected?

Can we presume self-connectivity of counties under your rules?  Can the northwestern corner of Chambers sever the Tallapoosa-Randolph link if itself is not connected to the rest of the county?  Shouldn't the part of Chambers County northwest of the Tallapoosa River be considered a county in its own right, with a link to bother Randoplh and Tallapoosa, but not the main part of the county?

Given a cut set, how do you draw a map of the district?

[Torie]

You want to punish maps with counties that are not chopped because they are not internally linked, Jimtex?  The law you would write to implement your structure would be a veritable trip to read I suspect, running laps around the text in the IRS regulations governing ERISA. Beyond erosity, I am into KISS now. That is why I want to jettison Muon2 putting in his region stuff that in the end just does not matter legally, even if it is indeed an efficient and efficacious way to produce pareto optimal maps. Just put the trade craft pointers in a footnote or something.

And KISS by the way, is why I am open to Muon2's connectivity approach to erosity, if the ensuing maps pass the eye test. It may be but a proxy, but it also is in KISS country, and that is no small matter. Sometimes proxies are the cat's meow.

[muon2]

Last point first, if you use DRA zoom into the boundary of your district. On each boundary between two counties in different districts see if there's a state or federal highway that crosses the boundary. If there is it more than likely a connection and you can count it as cut. There are some rare cases where there is a state highway but none are part of a path between county seats and then you can ignore them. If you choose to count the total for each district separately you have to divide by two after you add them all together since each boundary segment is on two districts.

Why do show Macon and Russell as connected?

Can we presume self-connectivity of counties under your rules?  Can the northwestern corner of Chambers sever the Tallapoosa-Randolph link if itself is not connected to the rest of the county?  Shouldn't the part of Chambers County northwest of the Tallapoosa River be considered a county in its own right, with a link to bother Randoplh and Tallapoosa, but not the main part of the county?

Given a cut set, how do you draw a map of the district?

This old's eyes couldn't tell a sharp edged rectangle from a rounded edge (Bing maps). So I pulled up different mapping software just now where the shapes could not be confused (Mapquest). So yes, Macon would not connect to Russell without a local override.

I presume self-connectivity of any whole county. If a county is chopped in such a way that there is no local road connection between two fragments then there is no connection between those fragments for erosity.

If I made a one-to-one correspondence between the graph and the map then all contiguous links would exist with weights equal to their border length. That would permit transformations in both directions of the kind you suggest. My proposal replaces the length-based weight with a 1 or 0 based on road connectivity. Since the graph is a tool to measure the map there is no requirement that the transformation be bidirectional any more than a ruler need map back onto the object it measures.

You only count cut highways that connect county seats? Does the highway need to actually penetrate the city limits of the county seat if that is the case? These days, sometimes highways circumvent towns. Interstates often do that in particular. If two highways between two counties are cut (however one picks and chooses which highways count), does that count as two cuts?

What state I wonder would be the best one to test this out, once the rules are precisely known, that is one of the more likely to generate a map more erose than most states in order to keep the chops down?  Any suggestions?

I use the closest geographic point on a highway to the building that represents the seat of county government (courthouse, executive office, etc. depending on the state). Attempts to use whole munis ran into problems when I ran into large munis that were going to get chopped.

As I noted in my proposal

You must be logged in to read this quote.

so if there two or more possible paths only the shortest path counts.

I think you are suggesting a test state with no large counties but relatively few counties per district. The map should be able to have an erose solution with no chops. My analysis suggests states with between 9 and 12 counties per CD absent CDs that could be wholly nested in a county. That would tend to force cases with either chops or erose districts to stay within the deviation. AL should be the best test, along with LA and NM. WI and IN would also be worth a look since though they each have one large county (over a CD) they avoid any VRA issues.

[Torie]

"A connection is the shortest continuous path of numbered state or federal highways that links the administrative seats of two counties without passing through any other county."

I read the above as not counting a highway that links two counties which does not penetrate the municipal boundaries of both county seats. In order to "link" you need to penetrate is the way I would opine if a judge who needed to rule on its meaning.  Is that your actual intent? If it is, it needs to be written more clearly to carry out that intent. Whether that is the best rule or not to implement your erosity proxy is another matter to discuss.

And I take it from what you said, that there can be but one road severing between two counties. So you count the number of counties that append the perimeter of a CD, and subtract from that number, the number of such counties that do not have a road link severance however that might be defined. Is that right?

[muon2]

Consider the border between the red and purple districts in AL. I'll consider the county seat to the measured at the point where a state of federal highway is nearest the county courthouse, but selecting the board room of the commissioners works equally well. I'll go from east to west.

Cleburne-Randolph: The one link between counties uses US 431/AL 1 at the county line to connect Heflin to Wedowee.
Cleburne-Clay: AL 9 and AL 49 both can be used to link Heflin to Ashland. AL 9 is shorter and defines the link. AL 281 crosses the border but doesn't link to any state route in Clay to connect with Ashland.
Cleburne-Talladega: No state highway crosses this border so there is no link.
Calhoun-Talladega: There are a number of highways that cross the border and could be used to define a path between Anniston and Talladega. These include AL 21, I 20, US 78/AL 4 (at Eastaboga), and AL 77. AL 21 is on the shortest path and defines the link.

For this border segment there are 3 severed links in the cut set, so this part has an erosity of 3.

[Torie]

The counties are not labeled, the towns are not labeled, the courthouse point not identified, so thus your post was incomprehensible to me. Sorry. I am no further along in putting your proposed rule words next to how it actually works, that in fact works. And this is the sort of thing that runs up legal fees, just so you know Mike.

[muon2]

The counties are not labeled, the towns are not labeled, the courthouse point not identified, so thus your post was incomprehensible to me. Sorry. I am no further along in putting your proposed rule words next to how it actually works, that in fact works. And this is the sort of thing that runs up legal fees, just so you know Mike.

Sorry but DRA (Bing maps) does not have a zoom level that shows both the seats and the roads of interest at the same time on one screen. Courthouses are irrelevant in this example since all I need to show is that a path gets close. I know your facility with DRA and have full confidence in your ability to operate the zoom. Perhaps I'll have time in a few days to create a synthetic example that puts it all on one screen, but since I'll be in DC for the first part of next week it will wait.

[jimrtex]

And KISS by the way, is why I am open to Muon2's connectivity approach to erosity, if the ensuing maps pass the eye test. It may be but a proxy, but it also is in KISS country, and that is no small matter. Sometimes proxies are the cat's meow.

Simple version:

The state's objective is equipopulous districts that are comprised of whole counties, that are contiguous, and compact.  Note that not erose and compact are not the same.  Long skinny districts are not compact either.

Counties are compact.  Pairs of adjacent counties are ordinarily reasonably compact.  It is when you put many together that they may become non-compact.  A simple measure of compactness is their boundary length.  Erose districts have long boundaries because they weave in and out.  Long skinny districts also have long borders.   A 12x1 groups of Iowa counties has a perimeter of 26.  A 6x2 group has a perimeter of 16.  A 4 x 3 group has a perimeter of 14.  A an 8 x 2 group, with 4 notches:

XX
X
XX
X
XX
X
XX
X

Has a perimeter of 26.  A long perimeter means that a district is not compact, whether it is long and skinny or erose.

Imagine the following problem:

Farmer Van Zandt owns 67 pastures which he wishes to divide among his seven children.  Each of the 67 pastures can support a certain head of cattle.   He wants the total head of cattle that can be pastured by the seven children to be as equal as practicable.  The pastures of each of the seven children must be connected, and Farmer Van Zandt also wants to minimize the amount of fencing to separate the ranches of the seven children.

It is a simple junior high level problem.

Turn it Up

[muon2]

To extend the example of the pastures, suppose there is a deep ravine cutting across part of the pastureland that the cattle cannot cross. One could put up a fence, but there is no reason to do so since the cattle will not cross.

One significant difference in my proxy to a direct perimeter calculation is that I don't charge for unneeded fencing. For example, In AL there is no crossing of the Alabama river between Autauga and Lowndes. I assign no erosity when that segment is on the border. This has the effect of recognizing that significant natural water and mountain barriers are acceptable as borders even though they might elongate a district. It also solves the issue of when a body of water gets large enough to become a barrier. For instance, how does one distinguish between the Alabama river and Chesapeake bay in MD when state county boundaries extend to the middle of both bodies of water. A connection-based decision resolves that.

[traininthedistance]

To extend the example of the pastures, suppose there is a deep ravine cutting across part of the pastureland that the cattle cannot cross. One could put up a fence, but there is no reason to do so since the cattle will not cross.

One significant difference in my proxy to a direct perimeter calculation is that I don't charge for unneeded fencing. For example, In AL there is no crossing of the Alabama river between Autauga and Lowndes. I assign no erosity when that segment is on the border. This has the effect of recognizing that significant natural water and mountain barriers are acceptable as borders even though they might elongate a district. It also solves the issue of when a body of water gets large enough to become a barrier. For instance, how does one distinguish between the Alabama river and Chesapeake bay in MD when state county boundaries extend to the middle of both bodies of water. A connection-based decision resolves that.

I've been playing around with Virginia a little bit (don't have any maps to show yet, but soon), and the idea that natural water and mountain barriers ought to be followed, even if they result in elongated districts, bears heavily there- esp. with regards to the Eastern Shore and the Shenandoah Valley.

[jimrtex]

County borders are simple to measure.



Highway connectivity is harder to measure, and reduces connection to a true/false value.  The Census Bureau does not connect Randolph to Cleburne by highway.  St.Clair County, the green county to the west has two county seats.  And a road that nicks the corner of a county, by as little as a block, separates the county.  It is not simple.



Perimeter length comports well with simple common sense notions of compactness.

[jimrtex]

You want to punish maps with counties that are not chopped because they are not internally linked, Jimtex?

I want a consistent set of rules.   It is Muon who wishes to punish maps because a highway between two counties clips a corner of another county.  When that corner is itself not accessible from the rest of its own county, such that it can't be reached from its own county seat except through the other counties, treat it as a county in its own right. 

I don't understand what principle that Muon is trying to promote with his rule.  If it is to ensure that a representative can travel throughout his district by numbered US or state highway, then my addition is essential  A representative can not travel to the northwestern corner of Chambers except via Tallapoosa or Randolph.

[jimrtex]

That is why I want to jettison Muon2 putting in his region stuff that in the end just does not matter legally, even if it is indeed an efficient and efficacious way to produce pareto optimal maps.

Muon has not stated his region rule in a useful way.  He has restricted deviation between regions to a level he believes is acceptable.  This results in you abandoning any attempt to use whole counties, and instead using your Tuschopalooza of the 5th most populous county in the state.  You can hack and chop and saw in that county to get whatever result you want elsewhere.

Compare with my rule:  Regions with up to 5% deviation are permitted, but the objective is to reduce inequality as much as practicable using whole county rules.

Here is a result of the KISS approach.



This map complies with Wesberry v Sanders, Kirkpatrick v Preisler, Karcher v Daggett, and Tennant v Jefferson.

[jimrtex]

To extend the example of the pastures, suppose there is a deep ravine cutting across part of the pastureland that the cattle cannot cross. One could put up a fence, but there is no reason to do so since the cattle will not cross.

One significant difference in my proxy to a direct perimeter calculation is that I don't charge for unneeded fencing. For example, In AL there is no crossing of the Alabama river between Autauga and Lowndes. I assign no erosity when that segment is on the border. This has the effect of recognizing that significant natural water and mountain barriers are acceptable as borders even though they might elongate a district. It also solves the issue of when a body of water gets large enough to become a barrier. For instance, how does one distinguish between the Alabama river and Chesapeake bay in MD when state county boundaries extend to the middle of both bodies of water. A connection-based decision resolves that.

I've been playing around with Virginia a little bit (don't have any maps to show yet, but soon), and the idea that natural water and mountain barriers ought to be followed, even if they result in elongated districts, bears heavily there- esp. with regards to the Eastern Shore and the Shenandoah Valley.

In delineating UCC, I combined independent cities with their original counties.

In addition to the obvious ones:

Arlington County (formerly Alexandria County) includes Alexandria.
Spotsylvania County includes Fredericksburg.

Norfolk City (County) includes Norfolk, Portsmouth, and Cheasapeake.
Virginia Beach (formerly Princess Anne County).
Newport News (formerly Warwick County).
Hampton (formerly Elizabeth City County).
James City County includes Williamsburg.
York County includes Poquoson.
Suffolk (Nansemond County)

Dinwiddie includes Petersburg.
Prince George includes Hopewell.
Chesterfield includes Colonial Heights.
Henrico includes Richmond.

Southampton includes Franklin.
Campbell includes Lynchburg,
Montgomery includes Radford.
Carroll includes Galax.