Poll re 2020 Ohio CD Map

[muon2]

I think I like Torie's design for Cincinnati better, but that isn't horrible. It looks like you might have split off some of the black areas in northern Hamilton County from the Cincinnati district, though. That's absolutely out of the question and would constitute dilution (not in a VRA sense, of course, but it's definitely worse than splitting counties).

Forest Park could be swapped for the Evendale-Blue Ash piece which would increase erosity. That switch only increases the BVAP from 20.8% to 22.3% so it would be hard to make a antidilution argument for a 1.5% increase.

[Benj]

I think I like Torie's design for Cincinnati better, but that isn't horrible. It looks like you might have split off some of the black areas in northern Hamilton County from the Cincinnati district, though. That's absolutely out of the question and would constitute dilution (not in a VRA sense, of course, but it's definitely worse than splitting counties).

Forest Park could be swapped for the Evendale-Blue Ash piece which would increase erosity. That switch only increases the BVAP from 20.8% to 22.3% so it would be hard to make a antidilution argument for a 1.5% increase.

Seems like a significant difference to me. A 1.5% decrease in BVAP is much more important than a tiny bit of erosity.

[muon2]

Yes, I like your Cinci map better than mine Mike (it's less erose while keeping the burbs together, and that should take precedence over the size of the Hamilton County chop). I don't like your OH-6 however. It's too erose. Ditto to a lessor extent you OH-07. And it looks like you chopped the city of Columbus. That CD should all be in the city.  Did you do a number of the black percentage in OH-14 and OH-15? I like my OH-14 (your OH-12) better myself. I guess one source of variance is that you give higher weight to minimizing the size of chops than erosity, while I do the reverse. I prefer my approach of course.

My CD-6 and 7 could both be smoothed out for two additional county chops. I'll add that to the list of erosity tests to see if it can be justified.

My Columbus chop was predicated on keeping Dublin and Westerville intact in my CD-9. Columbus has to be cut since it extends out of Franklin, so I went for a nice shape that kept the minority areas largely together. I could wrap my CD-7 up through Reynoldsburg to New Albany and push my CD-8 to the west into more of Columbus proper. It wouldn't change the erosity measure much, but CD-7 would have an ugly peninsula to New Albany.

[muon2]

Yes, I like your Cinci map better than mine Mike (it's less erose while keeping the burbs together, and that should take precedence over the size of the Hamilton County chop). I don't like your OH-6 however. It's too erose. Ditto to a lessor extent you OH-07. And it looks like you chopped the city of Columbus. That CD should all be in the city.  Did you do a number of the black percentage in OH-14 and OH-15? I like my OH-14 (your OH-12) better myself. I guess one source of variance is that you give higher weight to minimizing the size of chops than erosity, while I do the reverse. I prefer my approach of course.

My CD-6 and 7 could both be smoothed out for two additional county chops. I'll add that to the list of erosity tests to see if it can be justified.

My Columbus chop was predicated on keeping Dublin and Westerville intact in my CD-9. Columbus has to be cut since it extends out of Franklin, so I went for a nice shape that kept the minority areas largely together. I could wrap my CD-7 up through Reynoldsburg to New Albany and push my CD-8 to the west into more of Columbus proper. It wouldn't change the erosity measure much, but CD-7 would have an ugly peninsula to New Albany.

Your OH-14 seems more erose than my CD-12 and has a chop that I avoid. But each to their own tastes.

[TJ in Oregon]

Here's my attempt.



One of the features I discovered while drawing this is that the NW can be drawn much cleaner without the lake CD and the Toledo one going west instead of east. This is not my preferred configuration but makes CDs 4 and 5 much cleaner.

I also don't the dearth of clean-looking options for where to put Coshocton County. It would look bad in any of the four options I have for it.



I selected Muon's treatment of Cincinnati except with the black areas in OH-1 moved to OH-2 as suggested. It's messier but cuts out the bridge through Warren County without screwing up Dayton. You may be able to tinker around with the lines and which suburbs are in which to make it somewhat cleaner, but I didn't bother too much without population numbers.



Here's a map neither party would really like but serves communities of interest well IMO. It may be a bit of a competitiveness gerrymander. The NE corner district is not nearly as nasty as most I've seen and avoids re-entering Cuyahoga County in the Cuyahoga Valley from Summit County, allowing the western suburbs to be kept together. As a result, it ends up with all of the suburbs in Summit County between Cleveland and Akron. I was able to preserve the Mahoning Valley CD by including Kent, Ravenna, and Alliance, and going down the Ohio Valley some. The Akron-Canton ends up centered more on Canton than Akron, but still has both cities and the suburbs between. These three create an extra county split above the minimum as the NE one is split with both the Akron one and the Mahoning one, which also share a county split with each other. However, I think it is worthwhile since it allows the Mahoning district to take the cities lying between the Mahoning Valley and the Akron-Canton area and the Mahoning Valley seat doesn't become a baconstrip.

The western Cleveland suburbs CD when you draw it this way has enough population for either Elyria or Lorain but not both, so I chose Elyria. When done this way, its height and width are within a couple miles' difference.

[muon2]

Here's my attempt.



One of the features I discovered while drawing this is that the NW can be drawn much cleaner without the lake CD and the Toledo one going west instead of east. This is not my preferred configuration but makes CDs 4 and 5 much cleaner.

I like the shape of your Toledo CD, but it probably has too much pop by about 36K. You either need to drop Defiance or add Paulding and drop Sandusky.

[jimrtex]

I'm not too familiar here with the geography but is Warren here serving as a bridge or is that cut helping make a suburban concentric circle around Cincinnati? Does the Warren/Montgomery border also delineate a difference in mentalities between Dayton and Cincy?

The 4-county area of Hamilton, Butler, Warren, and Clermont has slightly more than enough population for 2 districts (2.096).  In recent times the area has had 3 districts, but the two suburban districts have had to extend way out into rural areas to pick up enough population.  And it so happens that Montgomery+Greene is just short enough of one district by the amount of excess in the 4-counties.

So it makes sense to give the excess from the 4-counties to the Dayton district as opposed to rural-districts along the Ohio River or extending up towards Toledo.

The next question is how to create the two Cincinnati districts.  Hamilton County is barely more than one district, so basing the division on counties is logical.  The other possibility is to do an east/west split, splitting both Hamilton and Butler counties.  But that may mean splitting the city of Cincinnati, or including all of Cincinnati plus taking areas in western Butler County, and that is a pretty arbitrary cut, with the primary purpose to keep two districts based in a county with the population for one district.

Warren County is one of the fastest growing counties in Ohio (or in more precise terms, one of the few counties in Ohio that is growing at all).   And the growth is spillover from Cincinnati with the city of Mason in the SW corner having become significantly larger than Lebanon.   Warren is somewhat like Butler County, which has small cities like Hamilton and Middleton that has become absorbed in the greater Cincinnati area.

I-75 goes south from Dayton through the northwestern corner of Warren.   The leafier area of Dayton is to the south, and so the part of Warren that goes in the Dayton district is definitely suburban fringe of Dayton, though there must be people who also commute southward as well.

[muon2]

Well, my work is done I think. Below I believe is the best possible map that can be drawn. Well, one possible alternative is to have OH-07 kick OH-05 out of Franklin County entirely. I will try that later to see how the erosity looks.
 

I'll see if I can put together the muon2 erosity formula for your map sometime this weekend.

Towards checking the formula, are there chops which are really microchops of less than 3.9K (0.5% of a CD)?

[Torie]

The Clark County chop for OH-03 is 3,373, but should be 2,135 folks in 2020 population. However, the Warren population in OH-03 is an estimate since Warren is chopped. If OH-03's portion of Warren grows slower than the OH-02 portion, than the chop may be necessary (which Jimtex creditably suggests above is the case). However, for purposes of this exercise, it could be considered a micro-chop. The Portage chop is about 9K, and that one is accurate, since the OH-14 population is otherwise whole counties. I agree with the micro-chop rule by the way.

The OH-01 population by the way is accurate (806K for Hamilton based on 2020 numbers less the chop of 27). It does not have too many people.

On your chop rationale for the City of Columbus, that will never pass the smell test. If a CD can be placed entirely within a city, it should be as opposed to being chopped in order to avoid a smaller municipality chop. To do otherwise is the tail wagging the dog.

Here is a revised map of mine that gets the erosity down. The population numbers are off a bit, but twists can get it right, without another county chop I think. Jimtex's map below has way too much erosity in my opinion. Erosity has to be a key factor, or the rules simply are not going to seem plausible. That is the first thing folks see when looking at a map. It is more important than an extra chop or two I think, where there is a conflict between the two. (And in that regard, some credit should be given for straight rather than jagged lines, even if the outside the square test considers a jagged line CD relatively speaking equal in erosity to a relatively straight line CD.) But the map below I think keeps the chops to a minimum, unless I missed something - which is often the case.

[jimrtex]

This map includes refined county splits based on actual population of county subdivisions.  I changed the northeast a bit to keep Warren and Youngstown together and to avoid a split between Canton and Massillon.



Population methodology.  County populations are based on July 2012 Census Bureau estimates, and the April 2010 census, projected to 2020 assuming a constant percentage rate of growth.  County subdivisions are based on the  July 2011 Census Bureau estimates pro-rated to the 2020 county projection.  This underestimates population in higher growth areas, and overestimates it in low growth or declining areas.   But since only the intracounty distribution is affected and growth rates so low (or negative) it doesn't matter much to the overall distribution of the districts.

In the southwest, the split in Warren County is north of Lebanon, and clearly more under influence of Dayton than Cincinnati.  The most serious objection is that it splits Middleton which crosses the Butler-Warren county line.

The split of Franklin County has everything from Norwich around to Truro in the Springfield-South Metro district.

The Columbus district includes the fractured townships of Clinton and Franklin, the enclaves of Bexley and Whitehall, and the Upper Arlington-Marble Cliff-Grandview Heights peninsula.  About 2/9 of Columbus is in the North Metro district, along with the Dublin, Westerville, Gahanna, and the northern townships (or what is left of them): Washington, Perry, Sharon, Blendon, Plain, Miflin, and Jefferson.

The larger share of the population of Seneca County, including Fostoria and Tiffin is included with Toledo.  In a real map, a different configuration or even a different county might be split.  Likewise, most of Guernsey is included with the southern district, but Cambridge is included in the Northeast Central small cities district.

The Cleveland-East District includes Cleveland and everything east of the Cuyahoga river, plus the Brooklyn and Cuyahoga Heights peninsulas.  If Cleveland continues to decline, it might work out to include Lakewood and shift some of the southern areas out.

About 4/7 of the red district is in Cuyahoga County, with most of the the rest is in Lorain.  The portion from Medina is the most remote part from Cleveland, but that is partly because it is only 30,000 persons.  If you included part of Brunswick for example, it would have to be split.

The Summit-centered district is a combination of Akron COI, including Wadsworth and Kent, and far southern Cleveland suburbs like Brunswick, Richfield, Twinsburg, and Aurora.

The division of Mahoning around Youngstown is fairly tight.  It could be relaxed if the yellow district were brought northward into Geauga County, but I don't like the idea of splitting another county.  There is no claim of a single COI for the purple district, it is more of an aggregate of smaller COI that are grouped together based on geographical proximity.

Looking long-term the arrangement of districts in the northeast should be reasonably stable.  The orange district can expand westward, picking off older suburbs like Lakewood and Parma, while the red district sweeps eastward across Medina, and the slate district extends into Portage and Geauga counties.  The purple district the shifts southward fairly quickly.

If Ohio loses another district in 2030 it would probably be the lavender district, which doesn't have an anchor (either a large city or the edge of the state).  The Columbus districts won't need more more population, but will be shoved eastward by the western and southwestern districts.  The 5 northeastern districts will need about 300,000 more people, either from internal growth which is close to non-existent, or expanding westward from Canton.

[muon2]

The Clark County chop for OH-03 is 3,373, but should be 2,135 folks in 2020 population. However, the Warren population in OH-03 is an estimate since Warren is chopped. If OH-03's portion of Warren grows slower than the OH-02 portion, than the chop may be necessary (which Jimtex creditably suggests above is the case). However, for purposes of this exercise, it could be considered a micro-chop. The Portage chop is about 9K, and that one is accurate, since the OH-14 population is otherwise whole counties. I agree with the micro-chop rule by the way.

The OH-01 population by the way is accurate (806K for Hamilton based on 2020 numbers less the chop of 27). It does not have too many people.

On your chop rationale for the City of Columbus, that will never pass the smell test. If a CD can be placed entirely within a city, it should be as opposed to being chopped in order to avoid a smaller municipality chop. To do otherwise is the tail wagging the dog.

Here is a revised map of mine that gets the erosity down. The population numbers are off a bit, but twists can get it right, without another county chop I think. Jimtex's map below has way too much erosity in my opinion. Erosity has to be a key factor, or the rules simply are not going to seem plausible. That is the first thing folks see when looking at a map. It is more important than an extra chop or two I think, where there is a conflict between the two. (And in that regard, some credit should be given for straight rather than jagged lines, even if the outside the square test considers a jagged line CD relatively speaking equal in erosity to a relatively straight line CD.) But the map below I think keeps the chops to a minimum, unless I missed something - which is often the case.

I agree that the Clark chop is a microchop.

On the subject of Columbus, I can't make a CD that is solely in the city. It takes too many incursions to exclude all the enclosed pockets that are not the city - both unincorporated areas and other cities like Bexley and Whitehall. Bridges to all those pockets is a truly erose mess. Because of that unusual circumstance for Columbus, it can't exclusively enclose a CD in a reasonable way and it has to be split, splitting it to reduce erosity is reasonable.

I'll run erosity numbers but I think I'll find that you have a number of chops that actually induce extra erosity. For example that chop into Portage acts like a new county. It has a link to the rest of Portage which is cut for erosity, but it doesn't eliminate the cut links between Summit and Portage and between Geauga and Portage. The net is to add 1 to erosity. The Tuscarora chop at least eliminates a link to Stark while adding one to the rest of Tuscarora.

One technique to avoid these chops is to form compact apportionment regions first like jimrtex and I typically do. Then when the regions are cut it avoids creating excess chops.

[Torie]

A burb surrounded by the city counts as part of the city. That bit is easy. Here is a map with the least erosity of all, in which the Columbus CD is within exclusively the city (except for surrounded burbs). Some of the city is outside the CD, but nothing not city is within other than surrounded burbs. Maybe there are two or three weird county precincts or something, but to just chop the sh*t out of the city because there are a couple of odd precincts is a paradigm of the perfect being the enemy of the good. A precinct that is itself split I just don't count, and ignore the island within the city. The weird precincts create their own exception to all rules really. We are talking about a handful of people.

A lagniappe of the map is that it makes my Ohio-05 exclusively a rural CD, with just two CD's taking in the Columbus metro area basically. That's a beautiful thing. And it gives the Dems a chance in 2022 to take OH-07, as it rapidly trends Dem, creating a swing CD, as a nice little desert (OH-07 had a GOP PVI of 6.7% in 2008).

The more I think about it, the more important reducing erosity is. It's just job one, along with minimizing county chops, which is almost as important, but arguably not quite when it comes to metro unity issues. This map gets rid of the tri-chop of Franklin County.

[TJ in Oregon]

Torie, the other issue with that map is that, although Franklin County is nominally only split in half, the southern connection (near where the yellow sticks out in that odd shape) of the blue CD doesn't have a road through it. This might not matter, but it's something to be aware of.

[Torie]

Torie, the other issue with that map is that, although Franklin County is nominally only split in half, the southern connection (near where the yellow sticks out in that odd shape) of the blue CD doesn't have a road through it. This might not matter, but it's something to be aware of.

Well, if that bothers one, the OH-06 prong to the south can be added to OH-07, and OH-07 lose a couple of its NW city of Columbus precincts. The idea is to penetrate the city through only one "orifice" as it were, but that's a minor matter, if other issues seem more pressing. I could even get rid of the surrounded burbs that bothers Muon2, by building a bridge to them, losing some more NW Columbus precincts to OH-06, but the better policy is that surrounded burbs count as fungible with city precincts. That rule also reduces erosity too in most instances. It's all good.

Here is an alternative Franklin County chop map:

[muon2]

A burb surrounded by the city counts as part of the city. That bit is easy.

I don't see it as easy at all. From the point of view of representation either the CD is entirely within a city or not. The aggrieved parties are those in the non-city parts. Whitehall is about 18K and surrounded and is comparable in size to Evergreen Park, IL at 20K which is almost, but not quite surrounded by Chicago. I can say with certainty that the folks of Evergreen Park would rather be in a CD that includes a bunch of other suburbs, even if Chicago still makes up a majority, rather than in a CD which is only Chicago and EP. I would guess that the folks of Whitehall would feel likewise. Why should Grandview Heights be in better political shape than Bexley since it is surrounded as well when taken with Arlington Heights. In fact I could argue that a more compact shape would chop the eastern parts of your CD-6 in favor of GH and AH and Columbus to the north. I think the logic is clear, a CD that is not entirely within a city is what it is, regardless of whether the other parts are surrounded by a city or not.

You must be logged in to read this quote.

Torie, the other issue with that map is that, although Franklin County is nominally only split in half, the southern connection (near where the yellow sticks out in that odd shape) of the blue CD doesn't have a road through it. This might not matter, but it's something to be aware of.

Well, if that bothers one, the OH-06 prong to the south can be added to OH-07, and OH-07 lose a couple of its NW city of Columbus precincts. The idea is to penetrate the city through only one "orifice" as it were, but that's a minor matter, if other issues seem more pressing. I could even get rid of the surrounded burbs that bothers Muon2, by building a bridge to them, losing some more NW Columbus precincts to OH-06, but the better policy is that surrounded burbs count as fungible with city precincts. That rule also reduces erosity too in most instances. It's all good.

Here is an alternative Franklin County chop map:

This is a much preferred plan for Columbus. But cutting off the southern prong you have removed a county fragment that is used to compute erosity and solved the connectivity problem. If you included more of CD-6 towards Dublin and removed the parts where Columbus borders Fairfield along a numbered highway you would improve erosity even more. You might look at a similar situation where putting Berea in CD-10 and linking Strongsville to the rest within the county would make a less erose map.

You must be logged in to read this quote.

[Torie]

Hey, Mike, I forgot that you represent the burbs, you "save our suburbs" guy you.  (You are probably too young to remember to the SOS movement in Chicagoland aren't you?)  Anyway, we will have to disagree about surrounded burbs and the map drawing rules therefor in this context (particularly small ones), being the tail that wags  the shredding of the major city dog. But with Columbus you can cut into the city and remove them if that will self actualize you (creating more erosity by the way - bad!). I did cut off the southeast corner of Columbus, but it's part of the black portion of the city, so I reversed course on that part of the fix for TJ.

Below is your fix for the Cleveland area, and it's clearly superior. I thought Olmstead Falls had more black folks than it did, and forgot that I could unite Solon once the corner of Portage was bit off. So I did that as part of the exchange, and bit off two thirds of Lyndhurst to make the the east line of OH-11 smooth.

What do you think about the map overall, and my concept of making erosity control the single most important factor assuming the number of chops is contained in some manner?

[traininthedistance]

A burb surrounded by the city counts as part of the city. That bit is easy. Here is a map with the least erosity of all, in which the Columbus CD is within exclusively the city (except for surrounded burbs). Some of the city is outside the CD, but nothing not city is within other than surrounded burbs. Maybe there are two or three weird county precincts or something, but to just chop the sh*t out of the city because there are a couple of odd precincts is a paradigm of the perfect being the enemy of the good. A precinct that is itself split I just don't count, and ignore the island within the city. The weird precincts create their own exception to all rules really. We are talking about a handful of people.

A lagniappe of the map is that it makes my Ohio-05 exclusively a rural CD, with just two CD's taking in the Columbus metro area basically. That's a beautiful thing. And it gives the Dems a chance in 2022 to take OH-07, as it rapidly trends Dem, creating a swing CD, as a nice little desert (OH-07 had a GOP PVI of 6.7% in 2008).

The more I think about it, the more important reducing erosity is. It's just job one, along with minimizing county chops, which is almost as important, but arguably not quite when it comes to metro unity issues. This map gets rid of the tri-chop of Franklin County.

This is a lot better than your earlier map.  The only thing I really don't like is CD-10; certainly it should be possible to rotate things within 8, 9 and 10 to keep Elyria and Lorain together (and, as a bonus, Ashland/Mansfield in without introducing any more county chops or really upping the erosity at all.

I do like the ring around Columbus on CoI grounds, but one should make sure to keep road contiguity, and I'd definitely regard it as more of a luxury than a necessity.

[Torie]

That rotation will make OH-08 too erose. As OH-09 sucks up more of Wood County, and then some, OH-08 has nowhere to go to pick up its lost population, without getting substantially more erose, and wandering. As you can see, OH-04 is kind of boxed in between OH-08 and the Dayton and Columbus and Cinci metro areas, so it can't be pushed at all, without substantial map degradation.) Both OH-04 and OH-08, but particularly OH-08, were hell to draw. Again, I think erosity reduction along with chop minimization (with the possible exception as to chops of keeping metro areas together, to which I also give a high priority), is job one. I think it is what the public expects, and should have a right to expect.

Anyway, I did the best I could to get OH-09 out of Lorain County, and got about half of its population out, before running into a wall, as described above.

Yes, the map is much better than earlier drafts of mine. Thanks.

Well, I take that back in part. Ashland County is potentially playable. Below is another option, which makes OH-09 less erose, but OH-10, and OH-08 in particular (and OH-08 looks kind of nasty now with that choke point it has by virtue of being kicked out of Wood County), more erose. Close call. There should be some sort of overall erosity point system which dictates which version of the map is deemed superior.

[jimrtex]

A burb surrounded by the city counts as part of the city. That bit is easy.

I don't see it as easy at all. From the point of view of representation either the CD is entirely within a city or not. The aggrieved parties are those in the non-city parts. Whitehall is about 18K and surrounded and is comparable in size to Evergreen Park, IL at 20K which is almost, but not quite surrounded by Chicago. I can say with certainty that the folks of Evergreen Park would rather be in a CD that includes a bunch of other suburbs, even if Chicago still makes up a majority, rather than in a CD which is only Chicago and EP. I would guess that the folks of Whitehall would feel likewise. Why should Grandview Heights be in better political shape than Bexley since it is surrounded as well when taken with Arlington Heights. In fact I could argue that a more compact shape would chop the eastern parts of your CD-6 in favor of GH and AH and Columbus to the north. I think the logic is clear, a CD that is not entirely within a city is what it is, regardless of whether the other parts are surrounded by a city or not.

I think in the case of Bexley and Whitehall you could also be getting into VRA problems by not including them in a Columbus district.  They are much whiter than the surrounding areas of Columbus.  It would not be surprising to find that is part of their existing in the first place.   So end up either with discontiguous white spots, or corridors through black areas to link them to the outer suburbs.

The current Columbus (in Franklin) population is very close to the ideal size for a 2020 CD.  Were Columbus not growing, I could see making the CD coterminous with the city.  But it is growing at about the same rate as the rest of the county - Columbus annexes develop-able land in exchange for utilities, so it is still increasing housing units.

That would the only case that I would consider places like Whitehall and Bexley to be contiguous with all the other non-Columbus areas of the county.

But since Columbus must be split, I don't think it is logical to use all the split parts to connect to all the islands, and went with a more conventional split of Columbus.  Splitting the northern part of Columbus to place with the North Metro district fits well with my concept of making the South Metro district also extend to Springfield and Newark, and the fact that strongest growth in the Columbus area is to the north.

I included Franklin and Clinton townships within Columbus.  Clinton has very little population and the two archipelagos are fairly isolated.  Franklin and southwest Columbus are a real hodgepodge.   If all the Franklin fragments were connected, you would be including a lot of Columbus.

I can see including the Upper Arlington peninsula in the suburban area.  Upper Arlington was originally part of Perry Township, which was not a square Jefferson township, but a knife-shaped area from the Delaware County line southward.  Upper Arlington was the tip.  But if Upper Arlington is included with the suburbs, you would probably need to include the portion of Columbus immediately to its north in the city district, and cut off most of the connectivity.

If Upper Arlington were included, I would have no problem including Marble Cliff and Grandview Heights as well.

[traininthedistance]

That rotation will make OH-08 too erose. As OH-09 sucks up more of Wood County, and then some, OH-08 has nowhere to go to pick up its lost population, without getting substantially more erose, and wandering. As you can see, OH-04 is kind of boxed in between OH-08 and the Dayton and Columbus and Cinci metro areas, so it can't be pushed at all, without substantial map degradation.) Both OH-04 and OH-08, but particularly OH-08, were hell to draw. Again, I think erosity reduction along with chop minimization (with the possible exception as to chops of keeping metro areas together, to which I also give a high priority), is job one. I think it is what the public expects, and should have a right to expect.

Anyway, I did the best I could to get OH-09 out of Lorain County, and got about half of its population out, before running into a wall, as described above.

Yes, the map is much better than earlier drafts of mine. Thanks.

Well, I take that back in part. Ashland County is potentially playable. Below is another option, which makes OH-09 less erose, but OH-10, and OH-08 in particular (and OH-08 looks kind of nasty now with that choke point it has by virtue of being kicked out of Wood County), more erose. Close call. There should be some sort of overall erosity point system which dictates which version of the map is deemed superior.

You could always push the Toledo district further west: Fulton County is part of the Toledo metro.  I think further rotation is possible and arguably desirable, up to and including putting Sandusky in CD-10.  I would not necessarily be opposed to having Sandusky in any of those three districts, actually; like Ashtabula it is its own thing according to the Census Bureau and you put it where you need extra population.

Obviously I'd like to see Lorain County made whole, and would also consider Holmes (and to a lesser extent Wayne) to be a better cultural fit in 8.  But even just giving CD-9 Fulton in exchange for that portion of Seneca and maybe some of Wood ought to be an improvement on the erosity front.

[Torie]

Not satisfied yet, and going for more eh?  Greedy!

Anyway, you are not playing the anti-erosity game, so we are playing different games. Muon2 and I are trying to come up with tight rules, so that the communities of interest game is relegated to the dust bin of history. In that game, you hire shills to claim who is in love with whom, and where, and just who is beyond the pale, when in reality few in the public square give a darn - just the political gamesmen care (with the PVI data right on their little laptops).

So the idea is to have tight rules, along with procedural safeguards, and end up with something that looks easy on the eyes, such that it at least seems facially fair to Joe Six Pack, with which team that wins and which loses just a roll of the black box "formulaic" dice.

As to procedural safeguards, maybe with the point system, ala something akin to the Iowa system, present the map with the highest point number first, and either party can object, in which event  a second map with the second highest number of points is offered, as to which the other party (but not the first objecting party), could object, and then if both parties don't agree to the third map, you have a coin toss or something between the three maps, or a court picks one of the three. If the point counts are close enough between a set of maps, to make the game less itself "gameable," one map from the set of maps within the point range (maybe starting with the second map, so the highest point numbered map always gets considered), could be selected by a coin toss, so that neither party knows exactly which map will be offered as the second, and if need be, third, choices.

One needs to put some risk into going for the gold, so the players are more willing to take the bronze right off the bat if need be, rather then run the risk of ending up on the ropes and then crashing to the floor in a partisan knockout.

[traininthedistance]

Not satisfied yet, and going for more eh?  Greedy!

Anyway, you are not playing the anti-erosity game, so we are playing different games. Muon2 and I are trying to come up with tight rules, so that the communities of interest game is relegated to the dust bin of history. In that game, you hire shills to claim who is in love with whom, and where, and just who is beyond the pale, when in reality few in the public square give a darn - just the political gamesmen care (with the PVI data right on their little laptops).

So the idea is to have tight rules, along with procedural safeguards, and end up with something that looks easy on the eyes, such that it at least seems facially fair to Joe Six Pack, with which team that wins and which loses just a roll of the black box "formulaic" dice.

As to procedural safeguards, maybe with the point system, ala something akin to the Iowa system, present the map with the highest point number first, and either party can object, in which event  a second map with the second highest number of points is offered, as to which the other party (but not the first objecting party), could object, and then if both parties don't agree to the third map, you have a coin toss or something between the three maps, or a court picks one of the three. If the point counts are close enough between a set of maps, to make the game less itself "gameable," one map from the set of maps within the point range (maybe starting with the second map, so the highest point numbered map always gets considered), could be selected by a coin toss, so that neither party knows exactly which map will be offered as the second, and if need be, third, choices.

One needs to put some risk into going for the gold, so the players are more willing to take the bronze right off the bat if need be, rather then run the risk of ending up on the ropes and then crashing to the floor in a partisan knockout.

Well, yes, I consider metro area contiguity to be more important than erosity, though I do certainly put erosity quite high on my ideal "balancing test"- really, just behind VRA concerns, metro contiguity, and keeping municipalities whole (sub for counties, but only in truly rural areas).

As I mentioned upthread, I wish the Census Bureau put out NECTA-style determinations for the Mid-Atlantic and Midwest, where there isn't really any area for which the county is the lowest level of government.  If they did that, then it would be possible to develop a truly objective and nuanced standard for metro contiguity, and we could work from there.  In the South and West, okay, you'd still need to stick to counties, though hopefully there would be enough good sense and discretion for people to be willing to do things like split off the really empty parts of Riverside and San Bernardino.

I do not think that an entirely automated redistricting process is desirable.  Very stringent "fair" requirements, yes.  But tying everything to one particular number, absolutely not.

[muon2]

A burb surrounded by the city counts as part of the city. That bit is easy.

I don't see it as easy at all. From the point of view of representation either the CD is entirely within a city or not. The aggrieved parties are those in the non-city parts. Whitehall is about 18K and surrounded and is comparable in size to Evergreen Park, IL at 20K which is almost, but not quite surrounded by Chicago. I can say with certainty that the folks of Evergreen Park would rather be in a CD that includes a bunch of other suburbs, even if Chicago still makes up a majority, rather than in a CD which is only Chicago and EP. I would guess that the folks of Whitehall would feel likewise. Why should Grandview Heights be in better political shape than Bexley since it is surrounded as well when taken with Arlington Heights. In fact I could argue that a more compact shape would chop the eastern parts of your CD-6 in favor of GH and AH and Columbus to the north. I think the logic is clear, a CD that is not entirely within a city is what it is, regardless of whether the other parts are surrounded by a city or not.

I think in the case of Bexley and Whitehall you could also be getting into VRA problems by not including them in a Columbus district.  They are much whiter than the surrounding areas of Columbus.  It would not be surprising to find that is part of their existing in the first place.   So end up either with discontiguous white spots, or corridors through black areas to link them to the outer suburbs.

The current Columbus (in Franklin) population is very close to the ideal size for a 2020 CD.  Were Columbus not growing, I could see making the CD coterminous with the city.  But it is growing at about the same rate as the rest of the county - Columbus annexes develop-able land in exchange for utilities, so it is still increasing housing units.

That would the only case that I would consider places like Whitehall and Bexley to be contiguous with all the other non-Columbus areas of the county.

But since Columbus must be split, I don't think it is logical to use all the split parts to connect to all the islands, and went with a more conventional split of Columbus.  Splitting the northern part of Columbus to place with the North Metro district fits well with my concept of making the South Metro district also extend to Springfield and Newark, and the fact that strongest growth in the Columbus area is to the north.

I included Franklin and Clinton townships within Columbus.  Clinton has very little population and the two archipelagos are fairly isolated.  Franklin and southwest Columbus are a real hodgepodge.   If all the Franklin fragments were connected, you would be including a lot of Columbus.

I can see including the Upper Arlington peninsula in the suburban area.  Upper Arlington was originally part of Perry Township, which was not a square Jefferson township, but a knife-shaped area from the Delaware County line southward.  Upper Arlington was the tip.  But if Upper Arlington is included with the suburbs, you would probably need to include the portion of Columbus immediately to its north in the city district, and cut off most of the connectivity.

If Upper Arlington were included, I would have no problem including Marble Cliff and Grandview Heights as well.

I note that Torie mentioned keeping east Columbus together for black neighborhood integrity, but by that token Reynoldsburg should be in the CD as well. Since there is no VRA district possible in that area, I would shy away from that justification, unless it is applied consistently to the areas both in and adjacent to the city.

When a county needs to be split we can look at minimizing splits of internal structures like townships and munis. Ideally one would take the same approach with large cities that need to be split. In many cases cities have well defined planning areas and I would make those the units that should not be split without strong cause. Columbus seems to have some such planning areas, but they don't cover the entire city. Even so, I would suggest that they should be used, mixing and matching them with suburbs (internal or external) as needed.

[traininthedistance]

I should add that yes, I am aware that "communities of interest" is rather gameable, and as such we should be looking for objective proxies which are less gameable.  The things I am talking about- municipal/township boundaries, metro area lines, and urbanized areas, are all things which are in fact objective (for the latter two, we go with whatever the Census says). 

When a county needs to be split we can look at minimizing splits of internal structures like townships and munis. Ideally one would take the same approach with large cities that need to be split. In many cases cities have well defined planning areas and I would make those the units that should not be split without strong cause.

Yes, this is a good point.  I'd take it one further and posit that, within urbanized areas, keeping townships/municipalities/(wards or planning areas within large cities) together should even take priority over keeping counties together.

[Torie]

Well the problem is that we all try to cover our partisan motivations with little arguments that favor our side. It's just unavoidable. So a black box solution, with some flexibility to allow folks to agree on something they both think is better, even if the black box does not, seems the way to go.  And metro areas should be defined by density data, not the land included by the census bureau, since that tends to be far too crude, and take in a lot of empty land, or be arbitrary in their separation. Obviously the VRA needs to be adhered to, although tougher is this business about worrying about minority percentages even if the VRA in a given instance does not. Maybe the geography and anti chop rules serve as an adequate proxy there in most instances.

Anyway, back to the anti-erosity game, your suggested revision improved the erosity of OH-08, but makes that of OH-09 worse. I am trying to use in my mind the outside of an exact square approach, with the distance outside it in a CD (with such distance weighted as a ratio of the size of the square, so you don't have a bigger CD's ipso factor deemed more erose than smaller ones) weighted negatively on some exponential basis, as a function of the distance outside the square of the percentage of land area in the CD that lies outside the square. Anyway, in my mind, using that metric, I think your revision might get a slightly lower score, as the OH-09 square box either has to move west, or have all of the Fulton jut to the west deemed outside the square.

Where's muon2?