Non Partisan Redistricting.

[muon2]

Counties within a region must be contiguous - that is, each county in a region must be connected to at least one other county in the region, that pair of counties connected to another county in the region, and so on, until all counties are connected.

A county is considered to be connected with another county if they are adjacent (share a boundary) and it is reasonably easy to travel directly between the two counties.  

"Reasonably easy" lacks the specificity that otherwise is used to set up the parameters for the public. I would still put a more restrictive definition on for links, and then provide the specific map showing amended links to reflect the specific goals of the state.

I would use the following description from my earlier post with the italicized addition in keeping with the notion of strong definitions.

link: Two nodes generally form a link if there is a continuous path between those two nodes that uses only non-seasonal numbered state and federal highways or non-seasonal regular ferry service that does not pass through any county other than the counties of the two linked nodes.

connected: Two counties are connected if there is a link between the population nodes of the counties. Connected counties are by definition contiguous.

region: A single county or group of connected counties.

Then add your text describing the exceptions to the general definition:

You must be logged in to read this quote.

[muon2]

The links between counties are weighted.  Most links are ordinary links, and have a weight of three.  

Links between counties spanned by urban areas (continuous areas of dense settlement defined by the Census Bureau) have a weight of five.   This is to provide an incentive to keeping these areas within the same district.   There are three such county-spanning spanning urban areas in Washington: Seattle urbanized area: Snohomish-King-Pierce; Wenatchee urbanized: Chelan-Douglas; and Kennewick-Pasco urbanized area: Benton-Franklin  

A few links have a weight of one.  These indicate connections which are somewhat circuitous.  For example, Pierce and Lewis have a quite lengthy boundary but it is remote from their population centers of Tacoma and Centralia.  While it is possible to drive between the two counties, most persons would use I-5 through Olympia and Thurston County, and only use the more circuitous route when visiting Mount Rainier.  The lower weighting provides a disincentive to bypass intermediate counties.

If you are going to weight links, l like the idea of recognizing links that connect urbanized areas as stronger than other links. I don't like the class of lesser links, since they will inherently suffer from a certain arbitrary standard. A simpler model is to give urban links a weight of 2 and the others a weight of 1. I suspect that doesn't really change the results in any substantial way, but does lend better and simpler definition to the process.

[jimrtex]

Counties within a region must be contiguous - that is, each county in a region must be connected to at least one other county in the region, that pair of counties connected to another county in the region, and so on, until all counties are connected.

A county is considered to be connected with another county if they are adjacent (share a boundary) and it is reasonably easy to travel directly between the two counties.  

"Reasonably easy" lacks the specificity that otherwise is used to set up the parameters for the public. I would still put a more restrictive definition on for links, and then provide the specific map showing amended links to reflect the specific goals of the state.

You indicated that the general public could not understand complex mathematical or legalistic definitions.  The explanation that I was giving would be directed at members of the general public who were looking at the map.

Interested persons could be directed to the more technical standards, as well as the hearings the previous year that established the base map as well as connectivity weights.

Any adjacent counties which are not "connected" are exceptional.  The map provides a complete visual representation of which adjacent counties are counted, and which are not connected (exceptions).

[muon2]

Counties within a region must be contiguous - that is, each county in a region must be connected to at least one other county in the region, that pair of counties connected to another county in the region, and so on, until all counties are connected.

A county is considered to be connected with another county if they are adjacent (share a boundary) and it is reasonably easy to travel directly between the two counties.  

"Reasonably easy" lacks the specificity that otherwise is used to set up the parameters for the public. I would still put a more restrictive definition on for links, and then provide the specific map showing amended links to reflect the specific goals of the state.

You indicated that the general public could not understand complex mathematical or legalistic definitions.  The explanation that I was giving would be directed at members of the general public who were looking at the map.

Interested persons could be directed to the more technical standards, as well as the hearings the previous year that established the base map as well as connectivity weights.

Any adjacent counties which are not "connected" are exceptional.  The map provides a complete visual representation of which adjacent counties are counted, and which are not connected (exceptions).

We may have to agree to disagree here. The only way I can make your last statement work is to allow all adjacent counties be connected that have any road connection between them. That make "exceptional" precisely those counties with no road connection at all. That seems like a minimum threshold to then allow connections between counties with no road connection. With my rule I can say that the amendments to a specific rule were made by some third party and leave it at that.

Personally I think the public would be highly perplexed that your map links Okanogan to Lincoln but not Grant, since one must go through Grant to get to Lincoln. You used a much more complex method to make that determination than my simple rule. I don't see any functioning rule that doesn't either put both links in or keep both out, and I doubt that any panel in WA would choose one but not the other without an intent to gerrymander.

[jimrtex]

This is one possible entry.



Total: 39 enclosed links, weight 115.

Region score: 7, 4 regions plus 3 single-district regions.

And this is another, which rearranges the eastern districts.



It scores higher for connectivity, but has greater population deviation.

Scoring:

Paciific: 99.93% of ideal.  10 counties.  9 enclosed links.  Weight 27.  A strung out district, but actually fairly concentrated in the outer Puget Sound area.

Southeast: 99.21% of ideal.  10 counties.  15 enclosed links.  Weight 45.

Northeast: 100.54% of ideal.  7 counties.  10 enclosed links.  Weight 30.  7/10 of the district is in Spokane.

Central: 700.32% of ideal (100.05% per district).  6 enclosed links.  Weight 20.

Total: 40 enclosed links, weight 122.

Region score: 7, 4 regions plus 3 single-district regions.

[jimrtex]

"Reasonably easy" lacks the specificity that otherwise is used to set up the parameters for the public. I would still put a more restrictive definition on for links, and then provide the specific map showing amended links to reflect the specific goals of the state.

You indicated that the general public could not understand complex mathematical or legalistic definitions.  The explanation that I was giving would be directed at members of the general public who were looking at the map.

Interested persons could be directed to the more technical standards, as well as the hearings the previous year that established the base map as well as connectivity weights.

Any adjacent counties which are not "connected" are exceptional.  The map provides a complete visual representation of which adjacent counties are counted, and which are not connected (exceptions).

We may have to agree to disagree here. The only way I can make your last statement work is to allow all adjacent counties be connected that have any road connection between them. That make "exceptional" precisely those counties with no road connection at all. That seems like a minimum threshold to then allow connections between counties with no road connection. With my rule I can say that the amendments to a specific rule were made by some third party and leave it at that.

Personally I think the public would be highly perplexed that your map links Okanogan to Lincoln but not Grant, since one must go through Grant to get to Lincoln. You used a much more complex method to make that determination than my simple rule. I don't see any functioning rule that doesn't either put both links in or keep both out, and I doubt that any panel in WA would choose one but not the other without an intent to gerrymander.

You are looking for a mathematical rule.   I am looking for guidelines that allow a reasonable level of human judgment.

Desirable attributes of districts are:

Contiguity;
Equipopulous;
Compact;
Convenient;
Community of Interest.

They are ordered in roughly the ease with which they may be objectively measured.

Contiguity has an almost absolute definition if we ignore disjoint counties.  There are reasonable measurements of population equality.  But sometimes these are abused to gerrymander.

There are mathematical definitions of compactness - but often they do not lead to reasonable conclusions.

Convenience is subjective.  I interpret it to mean the ease by which a representative or the representative may move about the district.  Counties that are adjacent but not connected are not convenient.  It is not convenient to travel between Mount Vernon and Onak.  The fact that it is inconvenient has nothing to do with  the fact that the highway crosses into both Whatcom and Chelan counties.   It is convenient to travel between Everett and Wenatchee.  That the route cuts a remote corner of King County is not an inconvenience.

It is much more convenient to travel between Omak and Moses Lake by going across Douglas County than going via Grand Coulee, totally disregarding which counties the highway may touch upon.  This does not mean that Okanogan and Grant may not be in the same district, but rather that some other counties must be included.   It is really no more remarkable than saying Benton and Asotin counties may not be in the same district unless you include some other counties.

It is a direct route between Davenport and Omak.  The route does not cross substantive parts of other counties, and comes nowhere near to Republic, Moses Lake, or East Wenatchee.

There are 92 adjacency relationships in Washington.  Adjacency is a necessary condition for connectivity.  There are 73 connectivity relationships in Washington based on my map.  19 of 92 pairs with adjacency, but not connectivity may be considered exceptional (not necessarily rare, but certainly not the norm).

The public would have to contemplate that Grant and Okanogan were not connected, to become perplexed by it.

[jimrtex]

This is one possible entry.



Total: 39 enclosed links, weight 115.

Region score: 7, 4 regions plus 3 single-district regions.

And this is another, which rearranges the eastern districts.



It scores higher for connectivity, but has greater population deviation.

Scoring:

Paciific: 99.93% of ideal.  10 counties.  9 enclosed links.  Weight 27.  A strung out district, but actually fairly concentrated in the outer Puget Sound area.

Southeast: 99.21% of ideal.  10 counties.  15 enclosed links.  Weight 45.

Northeast: 100.54% of ideal.  7 counties.  10 enclosed links.  Weight 30.  7/10 of the district is in Spokane.

Central: 700.32% of ideal (100.05% per district).  6 enclosed links.  Weight 20.

Total: 40 enclosed links, weight 122.

Region score: 7, 4 regions plus 3 single-district regions.

And another.  With very good population equality but low connectivity.



Paciific: 99.93% of ideal.  10 counties.  9 enclosed links.  Weight 27.

Southeast: 99.98% of ideal.  9 counties.  12 enclosed links.   Weight 33.

Northeast: 99.98% of ideal.  6 counties.  5 enclosed links.  Weight 15.

Central: 700.11%.  Implied splits of Snohomish, King, Pierce, and Thurston counties.   8 enclosed links.  Weight 33.

Total: 34 enclosed links, weight 99.

Region score: 7, 4 regions plus 3 single-district regions.

The low connectivity score would provide a rationale for outright rejection.

[muon2]

This is one possible entry.



Total: 39 enclosed links, weight 115.

Region score: 7, 4 regions plus 3 single-district regions.

This has a population range of 0.51%. It seems like a reasonable plan to go forward.

You must be logged in to read this quote.

[/quote]
The southeast and northeast regions are single district and have variances in excess of 0.5%. Because of those districts the range is 1.33% at best. This places the burden on the state to defend the population variance, and given your preceding plan, it would seem to be likely to fail in court.

You must be logged in to read this quote.

The range between regions is 0.18%, but it could be as low as 0.09% after the central region is split. Nonetheless the rejection of this plan could be justified in court based on the logic of the WV case.


Here's an entry that I think would add some spice to our discussions. I've traded two points from whole county districts for high connectivity and low deviations.



North: 199.90% of ideal (-686), 11 counties, 11 enclosed links, weight 33.
West: 400.08% of ideal (+522), 8 counties, 8 enclosed links, weight 24.
South: 299.97% of ideal (-228), 14 counties, 16 enclosed links, weight 46.
East: 100.06% of ideal (+392), 6 counties, 8 enclosed links, weight 24.

Total 43 enclosed links, weight 127. This beats the other plans if I understand your method.
The range between regions is 0.18%, but with detailed splits of the regions it could be as low as 0.11%, so it's almost as good as the third plan with low connectivity.

Note also that the King piece of Stevens Pass could be excluded from a north district since it falls in a different region. Does that get fixed in the second round, and if so how does one defend the regions as a means of reducing county splits?

[jimrtex]

This is one possible entry.



Total: 39 enclosed links, weight 115.

Region score: 7, 4 regions plus 3 single-district regions.

This has a population range of 0.51%. It seems like a reasonable plan to go forward.

The absolute range is meaningless.  Equality should be measured using deviation from the mean for individual districts, and some composite measurement such as rms (I'm not sure how to do this for plans with multi-district regions.   They should not be afforded a presumption that all districts in the region will have identical populations),

I'm assuming that Congress set a 1% limit, or that Washington set a 1% limit prior to initiation of the actual redistricting process.  The courts should give more deference to such a standard than they gave to New Jersey (Karcher) where the state attempted to rationalize deviation after the fact.

And this is another, which rearranges the eastern districts.



It scores higher for connectivity, but has greater population deviation.

Scoring:

Paciific: 99.93% of ideal.  10 counties.  9 enclosed links.  Weight 27.  A strung out district, but actually fairly concentrated in the outer Puget Sound area.

Southeast: 99.21% of ideal.  10 counties.  15 enclosed links.  Weight 45.

Northeast: 100.54% of ideal.  7 counties.  10 enclosed links.  Weight 30.  7/10 of the district is in Spokane.

Central: 700.32% of ideal (100.05% per district).  6 enclosed links.  Weight 20.

Total: 40 enclosed links, weight 122.

Region score: 7, 4 regions plus 3 single-district regions.

The southeast and northeast regions are single district and have variances in excess of 0.5%. Because of those districts the range is 1.33% at best. This places the burden on the state to defend the population variance, and given your preceding plan, it would seem to be likely to fail in court.

It is arguably more compact, and if selected by the citizen panel, a better reflection of communities of interest.  The permitted deviation of 1% for a single district, and square root of N x 1% for a region was set in advance of the actual process.

And another.  With very good population equality but low connectivity.



Paciific: 99.93% of ideal.  10 counties.  9 enclosed links.  Weight 27.

Southeast: 99.98% of ideal.  9 counties.  12 enclosed links.   Weight 33.

Northeast: 99.98% of ideal.  6 counties.  5 enclosed links.  Weight 15.

Central: 700.11%.  Implied splits of Snohomish, King, Pierce, and Thurston counties.   8 enclosed links.  Weight 33.

Total: 34 enclosed links, weight 99.

Region score: 7, 4 regions plus 3 single-district regions.

The low connectivity score would provide a rationale for outright rejection.

The range between regions is 0.18%, but it could be as low as 0.09% after the central region is split. Nonetheless the rejection of this plan could be justified in court based on the logic of the WV case.

After the first round, all plans would be published.  It would likely be recognized that it was possible to create higher scoring plans.

After the second round of submissions, a central body would determine which proposed plans would be considered (voted on) by the representative panel of citizens.  The central body might be required to choose at least N (3?)  "high-scoring" plans, and also submit all other plans that score as well or higher.

So in this hypothetical example, the central body would not so much "reject" this plan, but not choose it, and because of its low connectivity score would not be submitted to the panel of citizens.

Here's an entry that I think would add some spice to our discussions. I've traded two points from whole county districts for high connectivity and low deviations.



North: 199.90% of ideal (-686), 11 counties, 11 enclosed links, weight 33.
West: 400.08% of ideal (+522), 8 counties, 8 enclosed links, weight 24.
South: 299.97% of ideal (-228), 14 counties, 16 enclosed links, weight 46.
East: 100.06% of ideal (+392), 6 counties, 8 enclosed links, weight 24.

Total 43 enclosed links, weight 127. This beats the other plans if I understand your method.
The range between regions is 0.18%, but with detailed splits of the regions it could be as low as 0.11%, so it's almost as good as the third plan with low connectivity.

Note also that the King piece of Stevens Pass could be excluded from a north district since it falls in a different region. Does that get fixed in the second round, and if so how does one defend the regions as a means of reducing county splits?

In scoring multi-district regions, I did not count enclosed links involving counties that were divided.  In essence, the connectivity relationship is between whole counties.  I believe this will also tend to favor concentrating multi-county regions on larger metropolitan areas since high connectivity scores can be achieved in large more-remote districts.

North: 199.90% of ideal (-686), 11 counties, 11 enclosed links, weight 33.

Snohomish must be split.  At this point we don't need to decide whether one whole district is formed in Snohomish, with the small remnant linking the northwestern and northeastern parts of the district, or we do a more even split of Snohomish, linking the two parts separately with the rest of the district.

In either case, the three Snohomish links are not counted.

***North: 199.90% of ideal (-686), 11 counties, 8 enclosed links, weight 24.

West: 400.08% of ideal (+522), 8 counties, 8 9 enclosed links, weight 24 27.

King must be cut.  But in addition, Kitsap must be cut.  So we lose 3 links, but you had missed one.

***West: 400.08% of ideal (+522), 8 counties, 6 enclosed links, weight 18.

South: 299.97% of ideal (-228), 14 counties, 16 enclosed links, weight 46.

There is a required cut of Pierce, but also one of Clark.  So we lose 3 links, and weight of 7.

***South: 299.97% of ideal (-228), 14 counties, 13 enclosed links, weight 39.

East: 100.06% of ideal (+392), 6 counties, 8 enclosed links, weight 24.

No change to East.

***East: 100.06% of ideal (+392), 6 counties, 8 enclosed links, weight 24.

Total 4344 enclosed links, weight 127130. This beats the other plans if I understand your method.  The range between regions is 0.18%, but with detailed splits of the regions it could be as low as 0.11%, so it's almost as good as the third plan with low connectivity.

***Total 35 enclosed links, weight 105, so relatively slow scoring.

In addition, its regions score is 5 (4 regions, 1 single-district region). 

Note also that the King piece of Stevens Pass could be excluded from a north district since it falls in a different region. Does that get fixed in the second round, and if so how does one defend the regions as a means of reducing county splits?

I value adherence to county boundaries much more than literal transportation continuity.

[muon2]

Note also that the King piece of Stevens Pass could be excluded from a north district since it falls in a different region. Does that get fixed in the second round, and if so how does one defend the regions as a means of reducing county splits?

I value adherence to county boundaries much more than literal transportation continuity.

So what you are advocating is a hierarchical system where each factor is evaluated separately before one gets to the next factor. That's quite different than a multivariate system with factors of relatively equal weight that are consider in toto. Based on proposals floated in IL and OH for reform based on a hierarchy, and my analysis of them, I have to lean towards a multivariate approach. The discussion of the CA maps and possible reform also support the concept of maps in a multivariate space.

Of course if you are a voting member viewing the submissions you are free to define which variables should take precedence, but others may want different variables to take precedence or create a blended score.

[muon2]

In scoring multi-district regions, I did not count enclosed links involving counties that were divided.  In essence, the connectivity relationship is between whole counties.  I believe this will also tend to favor concentrating multi-county regions on larger metropolitan areas since high connectivity scores can be achieved in large more-remote districts.

I get the motivation, but this doesn't make sense to me. Determination of the county to be split, and how it is split can remove different internal links, and in some cases even create them. My examples in the TN thread illustrate some of this. A blanket removal of the links ignores the ability of some counties to split more cleanly than others.

[jimrtex]

Note also that the King piece of Stevens Pass could be excluded from a north district since it falls in a different region. Does that get fixed in the second round, and if so how does one defend the regions as a means of reducing county splits?

I value adherence to county boundaries much more than literal transportation continuity.

So what you are advocating is a hierarchical system where each factor is evaluated separately before one gets to the next factor. That's quite different than a multivariate system with factors of relatively equal weight that are consider in toto. Based on proposals floated in IL and OH for reform based on a hierarchy, and my analysis of them, I have to lean towards a multivariate approach. The discussion of the CA maps and possible reform also support the concept of maps in a multivariate space.

Of course if you are a voting member viewing the submissions you are free to define which variables should take precedence, but others may want different variables to take precedence or create a blended score.

It is not actually a hierarchy in this case.

What I am proposing is an approach that depends on collaboration and decomposition.   This would probably be easier to see in the case of the legislature.  Imagine we were drawing 98 house districts for Washington.  We'll create senate districts by combining house seats - as is done in Ohio.  King County is entitled to 28 House seats.  If we don't simply split off King County as a separate map, redistricting becomes too complex to permit direct citizen involvement.

Consider how legislative redistricting was done in Ohio.  They did a partial decomposition based on the requirement that whole house districts be formed in larger counties.  But they largely ignored the (seemingly apparent) requirement that other districts be formed from whole counties.

In Ohio, I did a full decomposition based on counties, and I suspect that you did also.  I then did districting in each apportionment area.  In a limited number of cases, I adjusted my county groups, because I didn't like how the house districts came out.  You probably did this as well (I suspect more extensively).

This was possible because you and I had full control of the complete map.  I don't think this would be possible with a large public body.  So I would be opposed to making adjustments across county boundaries at a second stage.

My approach has the following advantages:

(1) Will likely permit more flexible population ranges, which will in turn permit more actual recognition of community of interests.
(2) Creates a simpler process that would allow direct citizen involvement.

Disadvantages:

(1) Limited look-ahead may produce poorer final districts.  For example, someone from Okanogan might look at your northern district, and decide it wouldn't be too bad to be in a district with Sultan or Monroe.  But when the final map is drawn, they may be in a district with Edmonds; or conceivably a district that stretches across the northern part of the state, linked by a piece of Snohomish (around 40,000 persons).

Skykomish is in a King County Council district with which it does not have direct road contact.  It is of course imperative that it be in a King County Council district, as opposed to a Snohomish Council district.

[muon2]

Note also that the King piece of Stevens Pass could be excluded from a north district since it falls in a different region. Does that get fixed in the second round, and if so how does one defend the regions as a means of reducing county splits?

I value adherence to county boundaries much more than literal transportation continuity.

So what you are advocating is a hierarchical system where each factor is evaluated separately before one gets to the next factor. That's quite different than a multivariate system with factors of relatively equal weight that are consider in toto. Based on proposals floated in IL and OH for reform based on a hierarchy, and my analysis of them, I have to lean towards a multivariate approach. The discussion of the CA maps and possible reform also support the concept of maps in a multivariate space.

Of course if you are a voting member viewing the submissions you are free to define which variables should take precedence, but others may want different variables to take precedence or create a blended score.

It is not actually a hierarchy in this case.

What I am proposing is an approach that depends on collaboration and decomposition.   This would probably be easier to see in the case of the legislature.  Imagine we were drawing 98 house districts for Washington.  We'll create senate districts by combining house seats - as is done in Ohio.  King County is entitled to 28 House seats.  If we don't simply split off King County as a separate map, redistricting becomes too complex to permit direct citizen involvement.

Consider how legislative redistricting was done in Ohio.  They did a partial decomposition based on the requirement that whole house districts be formed in larger counties.  But they largely ignored the (seemingly apparent) requirement that other districts be formed from whole counties.

In Ohio, I did a full decomposition based on counties, and I suspect that you did also.  I then did districting in each apportionment area.  In a limited number of cases, I adjusted my county groups, because I didn't like how the house districts came out.  You probably did this as well (I suspect more extensively).

This was possible because you and I had full control of the complete map.  I don't think this would be possible with a large public body.  So I would be opposed to making adjustments across county boundaries at a second stage.

My approach has the following advantages:

(1) Will likely permit more flexible population ranges, which will in turn permit more actual recognition of community of interests.
(2) Creates a simpler process that would allow direct citizen involvement.

Disadvantages:

(1) Limited look-ahead may produce poorer final districts.  For example, someone from Okanogan might look at your northern district, and decide it wouldn't be too bad to be in a district with Sultan or Monroe.  But when the final map is drawn, they may be in a district with Edmonds; or conceivably a district that stretches across the northern part of the state, linked by a piece of Snohomish (around 40,000 persons).

Skykomish is in a King County Council district with which it does not have direct road contact.  It is of course imperative that it be in a King County Council district, as opposed to a Snohomish Council district.

That's fine, but I can't reconcile it against statements like "I value adherence to county boundaries much more than literal transportation continuity," or your earlier statement that a mechanism to compare erosity in plans with different region counts wasn't needed since minimizing county splits was more important. This implies a hierarchy where the first goal is to maximize regions and the second is to maximize regions made of one district. Transportation continuity is secondary. I value them more equally, and on that we seem to disagree.

In OH I did use apportionment regions as you did. But I also found that reducing the number of regions could sometimes give me a better outcome for the other variables in the contest. I did not have the fewest counties split in the contest, yet I had the best score overall. I think those tradeoffs are important, and I want to explicitly solicit those tradeoffs.

On population range I only want flexibility insofar as it will not place the burden to justify a map on the state. Where I get mathematical is precisely to see how the legal structure might be established to defend a challenge to a map that did not have exact population equality. I agree that a consequence of my approach is to manage the translation from a defensible legal/mathematical framework to an understandable public framework.

[jimrtex]

What I am proposing is an approach that depends on collaboration and decomposition.   This would probably be easier to see in the case of the legislature.  Imagine we were drawing 98 house districts for Washington.  We'll create senate districts by combining house seats - as is done in Ohio.  King County is entitled to 28 House seats.  If we don't simply split off King County as a separate map, redistricting becomes too complex to permit direct citizen involvement.

Consider how legislative redistricting was done in Ohio.  They did a partial decomposition based on the requirement that whole house districts be formed in larger counties.  But they largely ignored the (seemingly apparent) requirement that other districts be formed from whole counties.

In Ohio, I did a full decomposition based on counties, and I suspect that you did also.  I then did districting in each apportionment area.  In a limited number of cases, I adjusted my county groups, because I didn't like how the house districts came out.  You probably did this as well (I suspect more extensively).

This was possible because you and I had full control of the complete map.  I don't think this would be possible with a large public body.  So I would be opposed to making adjustments across county boundaries at a second stage.

My approach has the following advantages:

(1) Will likely permit more flexible population ranges, which will in turn permit more actual recognition of community of interests.
(2) Creates a simpler process that would allow direct citizen involvement.

Disadvantages:

(1) Limited look-ahead may produce poorer final districts.  For example, someone from Okanogan might look at your northern district, and decide it wouldn't be too bad to be in a district with Sultan or Monroe.  But when the final map is drawn, they may be in a district with Edmonds; or conceivably a district that stretches across the northern part of the state, linked by a piece of Snohomish (around 40,000 persons).

Skykomish is in a King County Council district with which it does not have direct road contact.  It is of course imperative that it be in a King County Council district, as opposed to a Snohomish Council district.

That's fine, but I can't reconcile it against statements like "I value adherence to county boundaries much more than literal transportation continuity," or your earlier statement that a mechanism to compare erosity in plans with different region counts wasn't needed since minimizing county splits was more important. This implies a hierarchy where the first goal is to maximize regions and the second is to maximize regions made of one district. Transportation continuity is secondary. I value them more equally, and on that we seem to disagree.

Maximizing regions, particularly those with one district is consistent with a goal of minimizing county splits.   My proposed scoring was: Number of regions + number of single county regions + number of single district regions.  That does not imply a hierarchy,

We are using counties as proxies for community of interest and transportation continuity.  We know that this is not absolutely true.  We know that the western part of Jefferson County can not be reached directly from Port Townsend.   We know that Skykomish can not be directly reached directly from Seattle.

I am willing to ignore these inconsistencies with reality in order to take advantage of the value of the abstraction of using county boundaries: (1) simplicity; (2) greater range of population variation; (3) less likelihood of abuse.

By suggesting that the boundaries could be fixed up, you are abandoning the use of counties.

On population range I only want flexibility insofar as it will not place the burden to justify a map on the state. Where I get mathematical is precisely to see how the legal structure might be established to defend a challenge to a map that did not have exact population equality. I agree that a consequence of my approach is to manage the translation from a defensible legal/mathematical framework to an understandable public framework.

My presumption is that setting out a equality standard in advance, along with a procedure will reduce the burden on the state.

My important state interests are:

(1) Simplification.
(2) Active participation by ordinary decent citizens.

[muon2]

What I am proposing is an approach that depends on collaboration and decomposition.   This would probably be easier to see in the case of the legislature.  Imagine we were drawing 98 house districts for Washington.  We'll create senate districts by combining house seats - as is done in Ohio.  King County is entitled to 28 House seats.  If we don't simply split off King County as a separate map, redistricting becomes too complex to permit direct citizen involvement.

Consider how legislative redistricting was done in Ohio.  They did a partial decomposition based on the requirement that whole house districts be formed in larger counties.  But they largely ignored the (seemingly apparent) requirement that other districts be formed from whole counties.

In Ohio, I did a full decomposition based on counties, and I suspect that you did also.  I then did districting in each apportionment area.  In a limited number of cases, I adjusted my county groups, because I didn't like how the house districts came out.  You probably did this as well (I suspect more extensively).

This was possible because you and I had full control of the complete map.  I don't think this would be possible with a large public body.  So I would be opposed to making adjustments across county boundaries at a second stage.

My approach has the following advantages:

(1) Will likely permit more flexible population ranges, which will in turn permit more actual recognition of community of interests.
(2) Creates a simpler process that would allow direct citizen involvement.

Disadvantages:

(1) Limited look-ahead may produce poorer final districts.  For example, someone from Okanogan might look at your northern district, and decide it wouldn't be too bad to be in a district with Sultan or Monroe.  But when the final map is drawn, they may be in a district with Edmonds; or conceivably a district that stretches across the northern part of the state, linked by a piece of Snohomish (around 40,000 persons).

Skykomish is in a King County Council district with which it does not have direct road contact.  It is of course imperative that it be in a King County Council district, as opposed to a Snohomish Council district.

That's fine, but I can't reconcile it against statements like "I value adherence to county boundaries much more than literal transportation continuity," or your earlier statement that a mechanism to compare erosity in plans with different region counts wasn't needed since minimizing county splits was more important. This implies a hierarchy where the first goal is to maximize regions and the second is to maximize regions made of one district. Transportation continuity is secondary. I value them more equally, and on that we seem to disagree.

Maximizing regions, particularly those with one district is consistent with a goal of minimizing county splits.   My proposed scoring was: Number of regions + number of single county regions + number of single district regions.  That does not imply a hierarchy,

The hierarchy is implied when you use this score before considering a connectivity score or the degree of population equality. I suggest a Pareto test instead. It's easy to understand and implement.

Mathematically adding a point for one-region districts may be a desired policy but it is not consistent with the goal of minimizing county splits. Maximizing regions alone minimizes county splits. Both my plan and yours have four regions and for 10 districts require 6 county splits, some of which will involve multiple splits of the same county, such as in King.

You must be logged in to read this quote.

Perhaps I thought you planned on having the districts truly connected in the final plan. From your statement above I understand that you would not. Unless the disconnected parts are in the same county, and there is no other way to arrange the districts in that county district I would insist on their connection. Joining disconnected but contiguous pieces is a key weapon in the gerrymanderer's arsenal. I find that a connection requirement is therefore one of the best defenses against gerrymandering. I cannot concede it merely for some level of simplicity. This goes back to why I would disallow the Snohomish-Chelan link. It avoids this outcome but preserves the simplicity of whole counties.

On population range I only want flexibility insofar as it will not place the burden to justify a map on the state. Where I get mathematical is precisely to see how the legal structure might be established to defend a challenge to a map that did not have exact population equality. I agree that a consequence of my approach is to manage the translation from a defensible legal/mathematical framework to an understandable public framework.

My presumption is that setting out a equality standard in advance, along with a procedure will reduce the burden on the state.

My important state interests are:

(1) Simplification.
(2) Active participation by ordinary decent citizens.
[/quote]
But I think one must also consider the decisions of the court and not a purely hypothetical case. A congressional plan that results in a population range in excess of 1% is unlikely to survive, even with state interests involved. In a plan that involves a measurable factor like region count or connectivity if another plan can achieve the same goal but with lower population range, the court has said you must take the plan with lower population range. It is only when the goal cannot otherwise be achieved that a higher range may be used.

I think it would be hard to convince the court that simplification and public participation justifies a higher range. I could argue other ways to achieve those goals without relaxing population equality as much. Successful arguments have involved more concrete claims, such as WV arguing that their congressional plan kept counties whole and minimized the number of people shifted from one district to another.

[jimrtex]

Another map (based on 2, but could start from 1).

[jimrtex]

In scoring multi-district regions, I did not count enclosed links involving counties that were divided.  In essence, the connectivity relationship is between whole counties.  I believe this will also tend to favor concentrating multi-county regions on larger metropolitan areas since high connectivity scores can be achieved in large more-remote districts.

I get the motivation, but this doesn't make sense to me. Determination of the county to be split, and how it is split can remove different internal links, and in some cases even create them. My examples in the TN thread illustrate some of this. A blanket removal of the links ignores the ability of some counties to split more cleanly than others.

Consider a county that could be its own district, Snohomish almost.  San Joaquin in California.

We create a district consisting of that single county.  It is ridiculous to claim that it does not represent a community of interest or is not compact.

Yet it has maximum possible erosity and zero connectivity by our measures.  This suggests that these measures don't scale well, especially when building districts from whole counties.  If we have a two-county legislative district in Iowa, it has the maximum possible non-compactness and is almost twice as long as it is wide.  But it would be silly to claim it non-compact or erose.

Here is a first stage map for the Washington House (98 districts):



There are only 3 instances where there is a connected triangle of counties.   And two of those are in the central regions where I didn't do a very good job of minimizing the regions.  Both regions have a total surplus of two districts, and Douglas is the only split county with a population less than the target size.  Otherwise, the regions have N-1 internal links where there is N counties.  County connectivity/erosity don't provide any information.

Small districts in urban areas are likely to be less compact under any measure.  Growth along highways and railroads is more sustainable so that communities of interest may be more linear.  Non-developable areas may be relatively larger in size so that settlement has to skip over and around them.  Compactness measurements suitable for such areas may not be comparable to statewide measures.

So in a sense it doesn't matter how we measure erosity for urban areas.

Finally, your measurement of erosity was not the same as I used for my three proposed maps.

My rule was not to count any link for a county that would be split (Snohomish, King, Pierce, Thurston), or in your map (Snohomish, King, Pierce, Kitsap, and Clark)

An alternative scoring would be to only count boundaries that were actually crossed by districts.  That would add Lewis-Thurston, Thurston-Pierce, Pierce-Kitsap, Pierce-King, King-Snohomish, and King-Kittitas.

Plan 1: 45 enclosed links, 137 weight. (115 original measure)
Plan 2: 46 enclosed links, 144 weight. (122)
Plan 3: 40 enclosed links, 121 weight. (99)
Muon: 44 enclosed links, 130 weight. (105)

But why should we get connectivity points for Snohomish when it is very close to a district in its own right?

[jimrtex]

Maximizing regions, particularly those with one district is consistent with a goal of minimizing county splits.   My proposed scoring was: Number of regions + number of single county regions + number of single district regions.  That does not imply a hierarchy,

The hierarchy is implied when you use this score before considering a connectivity score or the degree of population equality. I suggest a Pareto test instead. It's easy to understand and implement.

Mathematically adding a point for one-region districts may be a desired policy but it is not consistent with the goal of minimizing county splits. Maximizing regions alone minimizes county splits. Both my plan and yours have four regions and for 10 districts require 6 county splits, some of which will involve multiple splits of the same county, such as in King.

Your plan splits 5 counties, mine splits 4.   If we accept that our goal is to form districts of whole counties, then the number of split counties is more relevant than the number of split counties.

You must be logged in to read this quote.

Perhaps I thought you planned on having the districts truly connected in the final plan. From your statement above I understand that you would not. Unless the disconnected parts are in the same county, and there is no other way to arrange the districts in that county district I would insist on their connection. Joining disconnected but contiguous pieces is a key weapon in the gerrymanderer's arsenal. I find that a connection requirement is therefore one of the best defenses against gerrymandering. I cannot concede it merely for some level of simplicity. This goes back to why I would disallow the Snohomish-Chelan link. It avoids this outcome but preserves the simplicity of whole counties.
[/quote]
Alternatively, it reflects a midwesterners sense of geography where the highway system was developed after the counties were defined, and the terrain was more forgiving and county commissioners made sure that there was a highway to each neighboring county.

Chelan County was split off from Kittitas County because in winter it was easier to travel from Wenatchee to Ellensburg via Seattle or Spokane where the railroads met.  Wenatchee developed as a railroad town, since trains had to be split to get over Stevens Pass.

Stevens Pass connects Everett and Wenatchee and directly led to the development of Wenatchee and creation of Chelan County.   The King-Snohomish boundary was an arbitrary line drawn from Puget Sound eastward to the Cascades

You must be logged in to read this quote.

My presumption is that setting out a equality standard in advance, along with a procedure will reduce the burden on the state.

My important state interests are:

(1) Simplification.
(2) Active participation by ordinary decent citizens.
[/quote]
But I think one must also consider the decisions of the court and not a purely hypothetical case. A congressional plan that results in a population range in excess of 1% is unlikely to survive, even with state interests involved. In a plan that involves a measurable factor like region count or connectivity if another plan can achieve the same goal but with lower population range, the court has said you must take the plan with lower population range. It is only when the goal cannot otherwise be achieved that a higher range may be used.

I think it would be hard to convince the court that simplification and public participation justifies a higher range. I could argue other ways to achieve those goals without relaxing population equality as much. Successful arguments have involved more concrete claims, such as WV arguing that their congressional plan kept counties whole and minimized the number of people shifted from one district to another.
[/quote]

If you had made this assessment prior to Tennant vs Jefferson, you would have based it Karcher (see for example the explanation provided by the Iowa legislative council in 2012).

That assessment was wrong.

The Supreme Court wants to get out of the business of micro-managing redistricting.  They realize that Justice Harlan was right, but can't figure out a way of saying so.

The referendum in Maryland was not active participation of the voters in the redistricting process.  It was faux democracy.

Direct and active participation of citizens in the redistricting process is a legitimate State goal.  Article I, Section 2 says that Representatives are to be "chosen every second Year by the People of the several States".  It is entirely consistent that the People not only choose the Representatives, but that they choose the districts from which the Representatives are elected.

It is arguably more legitimate than one of the goals cited by West Virginia, which was to avoid pairing incumbents - in effect preserving the choices made years ago, rather than making a new choice every second Year.

For the citizens to make an active and meaningful choice as to which districts are used - they must have real plans to choose among.  A slightly larger range in population, permits creation of real alternatives, without infringing in any meaningful way on one man one vote.

It is the tiny variation in population that makes active and meaningful participation of citizens in the redistricting process practicable.

It would be reasonable to permit rejection of a plan, if there is an alternative that would by simple changes (counties moved between two regions) improve either connectivity and/or population equality without reducing the other).

Your burden is to provide an alternative method of achieving the goal of active and meaningful public participation and greater population equality.

[muon2]

Maximizing regions, particularly those with one district is consistent with a goal of minimizing county splits.   My proposed scoring was: Number of regions + number of single county regions + number of single district regions.  That does not imply a hierarchy,

The hierarchy is implied when you use this score before considering a connectivity score or the degree of population equality. I suggest a Pareto test instead. It's easy to understand and implement.

Mathematically adding a point for one-region districts may be a desired policy but it is not consistent with the goal of minimizing county splits. Maximizing regions alone minimizes county splits. Both my plan and yours have four regions and for 10 districts require 6 county splits, some of which will involve multiple splits of the same county, such as in King.

Your plan splits 5 counties, mine splits 4.   If we accept that our goal is to form districts of whole counties, then the number of split counties is more relevant than the number of split counties.

We had a lengthy discussion a little over a year ago on this board as to how to count county splits. The consensus in the end is that each piece of a district in a county, including whole counties counts as a piece of a district. It means each chop into a county is the same whether it chops into an already chopped county. That way a three-way split of one county should count the same as two counties each split two ways. I recognize that is not how the OH competition viewed this issue, but since it was the consensus of the board, I'm going to use it. Mathematically it implies that the region count is equivalent to the number of county pieces.

[jimrtex]

Your plan splits 5 counties, mine splits 4.   If we accept that our goal is to form districts of whole counties, then the number of split counties is more relevant than the number of split counties.

We had a lengthy discussion a little over a year ago on this board as to how to count county splits. The consensus in the end is that each piece of a district in a county, including whole counties counts as a piece of a district. It means each chop into a county is the same whether it chops into an already chopped county. That way a three-way split of one county should count the same as two counties each split two ways. I recognize that is not how the OH competition viewed this issue, but since it was the consensus of the board, I'm going to use it. Mathematically it implies that the region count is equivalent to the number of county pieces.

In Washington we must make 4 cuts, two for King, and one each for Pierce and Snohomish, regardless of how any counties are connected.

If our goal is to achieve Iowa-like redistricting then 4 would be ideal.

It appears that this is not possible.  My 4th plan illustrates that 5 is possible.  But if 6 are used, then one of the extra two should be in King - based on the notion that the maximum number of cuts in a county should be round(entitlement) + 1, and to the extent possible avoid splitting smaller counties.   Note a standard of preferring splitting of larger counties in codified in Iowa statute (42.4.2).

Since King is fairly close to three districts, one may create 4 reasonable-sized fragments, which is desirable compared to the alternatives such as splitting Kitsap County (as in your plan), or separating Douglas and Chelan as in my Plan 4.  It also reduces the number of split counties.

I doubt that the discussion of a year ago ever confused split counties and county splits.

The rules of the Ohio contest were simply reflecting the terms of the Ohio Constitution that provides that as many whole house districts be formed in a county as possible.  As such, the rules did not count those whole house districts as being split.  If one follows the Ohio Constitution, then it really doesn't matter how those are counted.

[jimrtex]

This is the process that would be used to define connectivity.

(1) State law or regulations define default connectivity relationship.

Possibilities are that all adjacent counties (except point adjacency) are presumed connected.  Alternatively, boundaries with limited adjacency (eg adjacency index less than 5%) could be presumed to be not connected.   Counties where the boundary is substantially marine might be presumed connected regardless of the boundary length.

(2) State law or regulations define means of connection.  

This might include numbered federal and state highways, but it might also incorporate performance standards such as speed limits and access controls.

(3) State law or regulations define end points of connection.  

This might be based on population centers, or might be based on a distributed population model.  I suspect it would be trivial to incorporate Google or Mapquest routing on a batch basis between all census tract pairs between two counties.

(4) State law or regulations define local bodies for making initial determination of (non)-connectivity.   This might be for example, the local county governing bodies.

(5) State law or regulations define weighting.   These might diminish near-adjacency or long distance connections (perhaps connections where the travel time is greater than 1-1/2 hours).  And enhanced weighting for urban areas that span county boundaries.

(6) Local bodies make initial determination.   The burden would be on the local body to justify a value different than the default.  Local bodies would first identify routes between the counties, and then determine if any are reasonably direct.   Direct means non-circuitous, and not overly dependent on other counties, for a substantial proportion of one or both counties.

Factors to consider would be whether direct route is seasonal, or whether it passes substantially through other counties, particularly through areas of concentrated population.  In such cases, the route is indirect.  Factors not to be considered: Direct route is outside state, or passes through other counties to only a limited extent.

(7) Central body considers decisions of local bodies.   If the two local bodies disagree, then the central body resolves the issue.  The central body could also override the local decisions.

[muon2]

Your plan splits 5 counties, mine splits 4.   If we accept that our goal is to form districts of whole counties, then the number of split counties is more relevant than the number of split counties.

We had a lengthy discussion a little over a year ago on this board as to how to count county splits. The consensus in the end is that each piece of a district in a county, including whole counties counts as a piece of a district. It means each chop into a county is the same whether it chops into an already chopped county. That way a three-way split of one county should count the same as two counties each split two ways. I recognize that is not how the OH competition viewed this issue, but since it was the consensus of the board, I'm going to use it. Mathematically it implies that the region count is equivalent to the number of county pieces.

In Washington we must make 4 cuts, two for King, and one each for Pierce and Snohomish, regardless of how any counties are connected.

If our goal is to achieve Iowa-like redistricting then 4 would be ideal.

It appears that this is not possible.  My 4th plan illustrates that 5 is possible.  But if 6 are used, then one of the extra two should be in King - based on the notion that the maximum number of cuts in a county should be round(entitlement) + 1, and to the extent possible avoid splitting smaller counties.   Note a standard of preferring splitting of larger counties in codified in Iowa statute (42.4.2).

Since King is fairly close to three districts, one may create 4 reasonable-sized fragments, which is desirable compared to the alternatives such as splitting Kitsap County (as in your plan), or separating Douglas and Chelan as in my Plan 4.  It also reduces the number of split counties.

I doubt that the discussion of a year ago ever confused split counties and county splits.

The rules of the Ohio contest were simply reflecting the terms of the Ohio Constitution that provides that as many whole house districts be formed in a county as possible.  As such, the rules did not count those whole house districts as being split.  If one follows the Ohio Constitution, then it really doesn't matter how those are counted.

I found one thread here where Torie and I reviewed different ways to count splits. You can see he initially went with method one, but after looking at the implications over the next page and a half of the thread went to method two. I interpreted your comments as favoring method one compared to while my count is based on method two.

Let me expound more on the traditional ways to count county chops. One presumes that each county has an ideal configuration. For small counties that configuration is to be entirely within one district. For large counties that configuration is to have the maximum number of whole districts within the county and the remainder in a single district.

There are two ways to count the traditional configuration. Most common is to assign it zero chops for both small and large counties. That treats the baseline with an ideal of zero. The other way is to count the total number of districts in each county so that small counties start at one and large counties start at the number of whole districts that can be placed within plus one for the remainder. This starts with a baseline ideal equal to the number of counties plus the maximum number of districts that can be placed entirely within counties.

For any district which is not ideal one counts all the districts in the county, and in the first method of counting one subtracts the districts entirely within and in the second method one does not. Note that in the first method one jumps from zero to two and it is not possible to have a county with one fragment. In the second method one can never have zero pieces in a county so there is no jump of two for the first chop.

As I noted the first method is preferred since it has an ideal of zero, and it puts a premium on avoiding the first chop. It does allow one to go below the number in my aforementioned formula by chopping more than two districts into a county. The second one has the feature of treating two counties each split in two the same as one county with no split and one with a three-way split.

Suppose there are five counties for four districts. The counties are each 80% of the population for a district. They are arranged as a central county A and four counties that wrap completely around A: B, C, D, and E in clockwise rotation. Imagine two ways of mapping: I) one where A is whole and shares into B, which shares to C, which shares to D, which ends at E; II) the other divides A into four slices and leaves the other four counties whole. The first method of chop counting scores I) as 6 chops (2n-2) and II) as 4 chops. The second method counts both as 8 chops.

[jimrtex]

We had a lengthy discussion a little over a year ago on this board as to how to count county splits. The consensus in the end is that each piece of a district in a county, including whole counties counts as a piece of a district. It means each chop into a county is the same whether it chops into an already chopped county. That way a three-way split of one county should count the same as two counties each split two ways. I recognize that is not how the OH competition viewed this issue, but since it was the consensus of the board, I'm going to use it. Mathematically it implies that the region count is equivalent to the number of county pieces.

In Washington we must make 4 cuts, two for King, and one each for Pierce and Snohomish, regardless of how any counties are connected.

If our goal is to achieve Iowa-like redistricting then 4 would be ideal.

It appears that this is not possible.  My 4th plan illustrates that 5 is possible.  But if 6 are used, then one of the extra two should be in King - based on the notion that the maximum number of cuts in a county should be round(entitlement) + 1, and to the extent possible avoid splitting smaller counties.   Note a standard of preferring splitting of larger counties in codified in Iowa statute (42.4.2).

Since King is fairly close to three districts, one may create 4 reasonable-sized fragments, which is desirable compared to the alternatives such as splitting Kitsap County (as in your plan), or separating Douglas and Chelan as in my Plan 4.  It also reduces the number of split counties.

I doubt that the discussion of a year ago ever confused split counties and county splits.

The rules of the Ohio contest were simply reflecting the terms of the Ohio Constitution that provides that as many whole house districts be formed in a county as possible.  As such, the rules did not count those whole house districts as being split.  If one follows the Ohio Constitution, then it really doesn't matter how those are counted.

I found one thread here where Torie and I reviewed different ways to count splits. You can see he initially went with method one, but after looking at the implications over the next page and a half of the thread went to method two. I interpreted your comments as favoring method one compared to while my count is based on method two.

Let me expound more on the traditional ways to count county chops. One presumes that each county has an ideal configuration. For small counties that configuration is to be entirely within one district. For large counties that configuration is to have the maximum number of whole districts within the county and the remainder in a single district.

I disagree that it is ideal to have the maximum number of whole districts within larger counties.

But it is a requirement under the Ohio Constitution for house districts, and in the current interpretation of the Texas Constitution for the Texas House.

The reapportionment provisions are very much about apportionment, rather than districting.

Under the Texas Constitution, there are three types of districts:

(1) Multi-county single-member districts.  Districts made of (smaller) whole counties with a total population approximately equal to the ideal population for one representative.
(2) Single-county multi-member districts.  Districts made of a single (larger) county that elect one or more representative.
(3) Floterial districts.  Districts made of whole counties apportioned one representative on the basis of the surplus population from one or more large counties and zero or more small counties.

The Texas Supreme Court has ruled that nothing in the Texas Constitution prevents election from single-member (sub)districts in multi-member counties, and it is practically required in cases where there is a significant racial minority (and possibly political minorities).  When Texas first started using single-member districts, they were designated 1A, 1B, 1C, etc. recognizing that they were really subdistricts of the district specified in the constitution.

On the other hand use of floterial districts for election purposes violated OMOV.  If county A was entitled to 1.6 representatives, and county B was entitled to 0.4 representatives, then County A would elect its own representative, and the voters of county A and B collectively would elect the representative from the floterial district.   County B voters would represent only 20% of the electorate, though its population would entitle them to be 40% of an ordinary district.  Meanwhile, the same number of voters in County A would represent 25% of the electorate of its own district and the same 20% of the floterial district.

As a consequence, the Texas Supreme Court has ruled that an area representing the surplus population of the large county(-ies) may be attached to the smaller counties.  This violates the Texas Constitution which requires districts to be made of whole counties.  But it violates less than the alternative which would be to completely ignore the whole county provisions.

So there is a hierarchy.  Multi-whole-county single-member districts and multi-member single-county districts are preferred.  Floterial (apportionment) districts are permitted only to resolve OMOV issues, and then must be formed into districts with only one crossing county lines.

As a consequence, counties with population equivalent to 10 or more districts are not seen as having a surplus, because they can always be formed into districts within the 5% safe harbor.

Despite this interpretation of the Texas Constitution, there is typically an exception or two necessary to comply with OMOV.  In 2000, both Galveston and Jefferson counties had a population equivalent to 1.8 representatives.  The only county that they could be paired with was Chambers which had a population equivalent to 1.2 representatives.   But only one could be paired.  Ultimately, Galveston was paired with Chambers, and Jefferson was paired with a part of Orange, which was split.

In 2010, Ellis County had a population equivalent to about 0.9 representatives, but there were no potential partners with a population between 0.05 and 0.15 representatives.   A small fragment of Henderson was added to Ellis.

But in both instances, an additional floterial district could be created that would avoid splitting a smaller county.  In 2000, Galveston and Harris counties could have been combined in a floterial district with one district crossing the Harris-Galveston line, and in 2010, Dallas and Ellis counties could have been combined in a floterial district, with an Ellis County district including a sliver of Dallas County.

In these two instances, Harris and Dallas would have had an additional "chop", but this is preferable to chopping a smaller county.  So a focus on the chop count is misplaced if takes the focus away from what is being chopped.

[muon2]

I found one thread here where Torie and I reviewed different ways to count splits. You can see he initially went with method one, but after looking at the implications over the next page and a half of the thread went to method two. I interpreted your comments as favoring method one compared to while my count is based on method two.

Let me expound more on the traditional ways to count county chops. One presumes that each county has an ideal configuration. For small counties that configuration is to be entirely within one district. For large counties that configuration is to have the maximum number of whole districts within the county and the remainder in a single district.

There are two ways to count the traditional configuration. Most common is to assign it zero chops for both small and large counties. That treats the baseline with an ideal of zero. The other way is to count the total number of districts in each county so that small counties start at one and large counties start at the number of whole districts that can be placed within plus one for the remainder. This starts with a baseline ideal equal to the number of counties plus the maximum number of districts that can be placed entirely within counties.

For any district which is not ideal one counts all the districts in the county, and in the first method of counting one subtracts the districts entirely within and in the second method one does not. Note that in the first method one jumps from zero to two and it is not possible to have a county with one fragment. In the second method one can never have zero pieces in a county so there is no jump of two for the first chop.

As I noted the first method is preferred since it has an ideal of zero, and it puts a premium on avoiding the first chop. It does allow one to go below the number in my aforementioned formula by chopping more than two districts into a county. The second one has the feature of treating two counties each split in two the same as one county with no split and one with a three-way split.

Suppose there are five counties for four districts. The counties are each 80% of the population for a district. They are arranged as a central county A and four counties that wrap completely around A: B, C, D, and E in clockwise rotation. Imagine two ways of mapping: I) one where A is whole and shares into B, which shares to C, which shares to D, which ends at E; II) the other divides A into four slices and leaves the other four counties whole. The first method of chop counting scores I) as 6 chops (2n-2) and II) as 4 chops. The second method counts both as 8 chops.

We had a lengthy discussion a little over a year ago on this board as to how to count county splits. The consensus in the end is that each piece of a district in a county, including whole counties counts as a piece of a district. It means each chop into a county is the same whether it chops into an already chopped county. That way a three-way split of one county should count the same as two counties each split two ways. I recognize that is not how the OH competition viewed this issue, but since it was the consensus of the board, I'm going to use it. Mathematically it implies that the region count is equivalent to the number of county pieces.

In Washington we must make 4 cuts, two for King, and one each for Pierce and Snohomish, regardless of how any counties are connected.

If our goal is to achieve Iowa-like redistricting then 4 would be ideal.

It appears that this is not possible.  My 4th plan illustrates that 5 is possible.  But if 6 are used, then one of the extra two should be in King - based on the notion that the maximum number of cuts in a county should be round(entitlement) + 1, and to the extent possible avoid splitting smaller counties.   Note a standard of preferring splitting of larger counties in codified in Iowa statute (42.4.2).

Since King is fairly close to three districts, one may create 4 reasonable-sized fragments, which is desirable compared to the alternatives such as splitting Kitsap County (as in your plan), or separating Douglas and Chelan as in my Plan 4.  It also reduces the number of split counties.

I doubt that the discussion of a year ago ever confused split counties and county splits.

The rules of the Ohio contest were simply reflecting the terms of the Ohio Constitution that provides that as many whole house districts be formed in a county as possible.  As such, the rules did not count those whole house districts as being split.  If one follows the Ohio Constitution, then it really doesn't matter how those are counted.

I found one thread here where Torie and I reviewed different ways to count splits. You can see he initially went with method one, but after looking at the implications over the next page and a half of the thread went to method two. I interpreted your comments as favoring method one compared to while my count is based on method two.

Let me expound more on the traditional ways to count county chops. One presumes that each county has an ideal configuration. For small counties that configuration is to be entirely within one district. For large counties that configuration is to have the maximum number of whole districts within the county and the remainder in a single district.

I disagree that it is ideal to have the maximum number of whole districts within larger counties.

I think in the above referenced discussion, we came to that conclusion as well. I started assuming that counting method one was preferred (as in the OH competition), but as we looked at some of the results that could create in CA, the preference for counting method two emerged. Counting method two favors neither large nor small counties. I think you want to favor small county integrity, and one way we batted around to do that while keeping chops neutral for both types of counties was this:

There should be as few county splits as possible, and when a county is split in two one fragment should be as small as possible.

We were unable to come to an agreement on a hard percentage as a cutoff for a split, but the sense is that it moved maps away from relatively even splits of small counties.

[jimrtex]

I've rescored the plans:

Plan 1



Region Score: 7 (4 regions, 3 single-district regions)

West: 10 counties, 99.93%, 9 links, 27 weighted.
Northeast: 9 counties, 99.95%, 14 links, 42 weighted.
Southeast: 8 counties, 99.80% 10 links, 26 weighted.
Central, 12 counties, 700.32%, 3 split counties, 9 county fragments.
   King (east and south) 77% , Chelan, Kittitas, Douglas 23%, 3 links, 11 weighted.
   Snohomish (most) 100%. 0 links.
   King (Seattle) 100%, 0 links.
   King (north) 94%, Snohomish (remnant) 6%, 0 links.
   Thurston 38%, Kitsap 37%, Pierce (west) 25%, 0 links.
   Pierce (central) 84%, King (far south) 16%, 0 links.
   Clark 63%, Cowlitz, Lewis, Skamania 28%, Pierce (southeast) 9%, 3 links, 9 weighted.

Total 39 links, 115 weighted.
Split Counties: 3 (King, Pierce, Snohomish)
County Fragments: 9 (King 4, Pierce 3, Snohomish 2)
Mean County Fragment: 382,901 (56.84%)

Plan 2



West: 10 counties, 99.93%, 9 links, 27 weighted.
Northeast: 7 counties, 100.54%, 10 links, 30 weighted.
Southeast: 10 counties, 99.21% 15 links, 45 weighted.
Central, 12 counties, 700.32%, 3 split counties, 9 county fragments.
   King (east and south) 77% , Chelan, Kittitas, Douglas 23%, 3 links, 11 weighted.
   Snohomish (most) 100%. 0 links.
   King (Seattle) 100%, 0 links.
   King (north) 94%, Snohomish (remnant) 6%, 0 links.
   Thurston 38%, Kitsap 37%, Pierce (west) 25%, 0 links.
   Pierce (central) 84%, King (far south) 16%, 0 links.
   Clark 63%, Cowlitz, Lewis, Skamania 28%, Pierce (southeast) 9%, 3 links, 9 weighted.

Total 40 links, 122 weighted.
Split Counties: 3 (King, Pierce, Snohomish)
County Fragments: 9 (King 4, Pierce 3, Snohomish 2)
Mean County Fragment: 382,901 (56.84%)

Plan 3

Plan 3 has an improbable hook into eastern Washington.



West: 10 counties, 99.93%, 9 links, 27 weighted.
Far East: 6 counties, 99.98%, 6 links, 18 weighted.
Southeast: 9 counties, 99.98% 12 links, 33 weighted.
Central, 14 counties, 700.11%, 3 split counties, 9 county fragments.
   Snohomish (east) 78%, Chelan, Okanogan, Ferry, Lincoln, Adams 22%, 5 links, 15 weighted.
   King (north) 72%, Snohomish (west) 28%, 0 links.
   King (west) 63%, Kitsap 37%, 0 links.
   King (Seattle) 100%, 0 links.
   King (south) 53%, Pierce (east) 47%, 0 links.
   Pierce (west) 62%, Thurston 38%
   Clark 63%, Cowlitz, Lewis, Skamania 28%, Pierce (southeast) 9%, 3 links, 9 weighted.

Total 35 links, 102 weighted.
Split Counties: 3 (King, Pierce, Snohomish)
County Fragments: 9 (King 4, Pierce 3, Snohomish 2)
Mean County Fragment: 382,901 (56.84%)