Weighted Voting For Congress

[traininthedistance]

So New Jersey is kind of a natural for three districts, given the whole North Jersey/Central Jersey/South Jersey division that people map onto the state.  Of course, not calling them exactly that feels really weird to type, and there are more judgment calls than one might think as to what, exactly, counts as Central Jersey.  Anyhoo:



District 1: NEW JERSEY SOUTH.  Population 2,211,987 (deviation -718,644).  Obama 60.7%, Dem 57.8%.  This district is pretty easy for me; it's basically all of the state that cares more about Philadelphia than New York.  Mercer is a bit of a borderline case: it often gets put in Central Jersey, and parts of it probably do identify more with NYC, but historically Trenton was either considered its own thing or closer to a satellite of Philadelphia, and this district is pretty darn underpopulated as it is.  Some folks want to put Ocean in the south as well, but they're just flat-out wrong.  Safe D.

District 2: NEW JERSEY CENTRAL.  Population 3,005,097 (deviation +74,466).  Obama 52.1%, Dem 50.3%.  Aside from putting Mercer in the South, this is just straightforwardly the most expansive definition of Central Jersey that's out there, which is appropriate for this exercise given the extent to which North Jersey really does have the lion's share of the population otherwise.  Of course one could argue that "Central Jersey" really is just the crappier parts of North Jersey.   Anyway, Hunterdon and Union could both plausibly go into the North, but that would stretch even the super-wide range proffered in the ground rules.  Tossup.

District 3: NEW JERSEY NORTH.  Population 3,574,810 (deviation +644,179).  Obama 59.7%, Dem 57.4%.  This district is actually only 51.6% white (53.6% VAP), with Hispanics the largest minority in the low 20s.   Not much left to say about its boundaries.  Safe D.

[traininthedistance]

Georgia is, for all its zillions of counties, really easy to do.  And even with the mandate for coarse equality it all ends up being well within plus or minus 10 percent.



District 1: SOUTH GEORGIA.  Population 3,013,994 (deviation -215,224).  Obama 45.7%, Dem 48.0%.  36% Black (34% Black VAP).  I used the Atlanta media market as the dividing line here: everything south of it is in 1, everything within it (plus the few peripheral northern counties in other markets such as Rome) in 2 and 3.  Likely R, there's an opening for a Blue Dog in a good year here, but it's probably a narrow one.

District 2: ATLANTA.  Population 3,365,297 (deviation +136,079).  Obama 61.4%, Dem 58.2%.  40W/38B/13H (44W/37B/11H VAP), so min-maj.  These five counties were the original Atlanta metro area in 1950 and I imagine they're still considered to be the core of it today.  Obviously the exurbs spill far out into District 3 by now.  Safe D.

District 3: NORTH GEORGIA.  Population 3,308,362 (deviation +79,144).  Obama 36.6%, Dem 35.9%.  Pretty self-explanatory.  Safe R.

[traininthedistance]

I have a question related to population equality. For example, If the Boston metro is kept intact, and there is contiguity, then the only division is to separate Worcester and the counties to the west from the rest of the state. That creates MA-Bay with 1.5 times the state's quota and MA-Berkshires with 0.5 times the quota. Is that acceptable? Given the nature of the exercise in weighting, it seems to me that it should be.

I was thinking of Massachusetts as a state where strict contiguity might not be required, because of the extreme concavity caused by Rhode Island.  You have to choose between (1) population imbalance; (2) Using all of Norfolk, which would violate community of interest; (3) splitting Norfolk which violates the rule on splitting counties; or (4) Having a non-contiguity between Worcester and Bristol.

I feel like splitting counties in MA could plausibly be acceptable due to the fact that county government in New England is weak/nonexistent and town government is strong.  That region of the country is a special case.  I'd personally rather split counties than go non-contiguous here, but obviously this is your thought-experiment and not mine.

I haven't yet checked out whether doing that along NECTA boundaries (or even divisions within the Boston NECTA) would actually lead to districts that are closer to equal; the Boston metro is so dominant that it might not even help much.

Another possible point in favor of splitting Norfolk is that Norfolk County is, itself, noncontiguous.

[traininthedistance]

New York, New York.  The deviations get pretty high here, for obvious reasons.



District 1: LONG ISLAND.  Population 2,832,868 (deviation -1,042,747).  Obama 53.2%, Dem 55.2%.  Nassau and Suffolk.  A natural pair.  Way underpopulated, and there is absolutely nothing you want to, or even can do, about it- Queens needs to be in one of the two NYC districts, and adding it would overcorrect things even worse, and there's no compelling COI reason to go noncontiguous (or cross the Sound, which is functionally going noncontiguous).  Tilt D.

District 2: BROOKLYN-QUEENS.  Population 4,735,422 (deviation +859,807).  Obama 77.4%, Dem 79.5%. Obviously very diverse, min-maj with no dominant group: 32W/25B/23H/16A (by VAP it's 33W/23B/21H/16A).

NYC is obviously getting split into two districts here that average out to being slightly overpopulated; Queens + Bronx might actually have lower deviations (and would be contiguous), but in terms of COI the two LI boroughs belong together, I think.  Piece of evidence #1: the other three boroughs are part of the New York Public Library system, but Brooklyn and Queens are separate.  Piece of evidence #2: isn't it obvious that we should have one district for the Knicks and another for the Nets?  Anyway, Safe D.

District 3: NEW YORK, NEW YORK. Population 3,439,726 (deviation -435,889).  Obama 81.4%, Dem 77.9%.  Again, min-maj, but more Hispanics and less blacks and Asians: 35W/19B/36H (39W/19B/33H by VAP).  The other three boroughs; perhaps it's not fair to Brooklyn that those people just get to be called New York, but "Manhattan-Bronx-Staten Island" is a mouthful.  It looks noncontiguous on the map, but I'm gonna say that the Staten Island Ferry counts as the requisite transportation link here.  Safe D.

District 4: NEW YORK HUDSON VALLEY AND NORTH.  Population 3,941,721 (deviation +66,106).  Obama 56.2%, Dem 57.5%.  Kind of a mishmash, hence the uninspiring name: northern NYC suburbs, Hudson Valley, the Capitol Region, and the North Country.  I went by media markets and metro areas, as usual; the biggest judgment call was putting the North Country and Watertown in here rather than the Binghamton area, but this configuration had modestly lower deviations.  Eh, could go either way on that.  Lean D.

District 5: NEW YORK CENTRAL AND WESTERN.  Population 4,4428,337 (deviation +552,722).  Obama 53.5%, Dem 52.7%.  Utica, Syracuse, Ithaca, Rochester, Buffalo.  Western NY is more of an identity than whatever got thrown in District 4, at least.  Tossup.

[traininthedistance]

In that case I will proceed under the assumption that the 0.67 to 1.33 of a state's quota are soft limits. If there are good CoI reasons I can run from 0.18 to 1.62 of the national average so that no district is smaller than the smallest state, and no district is larger than the smallest state with two districts.

I still tend to think that equality is not so important, since this is an exercise for a system with weighted votes. Census groups and other factors should have more weight. For example, I don't like splitting the Newark metropolitan division within the New York metro in NJ. I would group Hudson, Bergen and Passaic together as a district and keep all of the Newark division in the central part of NJ. The resulting three districts would be NJ-Palisades (pop 2041K), NJ-Raritan (pop 4539K), and NJ-Pinelands (pop 2212K). They are not as equal in population as train's version, but do a better job of matching neutral definitions of CoI.

Speaking as someone who grew up in the Newark division...no, just no.  The divisions in urban northern NJ are not quite meaningless, but they simply are not the relevant cleavage at this scale. I can guarantee you that anybody who grew up in NJ would rather put Essex, Morris, and Sussex in with Bergen/Passaic than with Middlesex/Monmouth.  Maybe use them if NJ had four districts rather than three, or make the central district maximally underpopulated instead, but not that.

[traininthedistance]

In that case I will proceed under the assumption that the 0.67 to 1.33 of a state's quota are soft limits. If there are good CoI reasons I can run from 0.18 to 1.62 of the national average so that no district is smaller than the smallest state, and no district is larger than the smallest state with two districts.

I still tend to think that equality is not so important, since this is an exercise for a system with weighted votes. Census groups and other factors should have more weight. For example, I don't like splitting the Newark metropolitan division within the New York metro in NJ. I would group Hudson, Bergen and Passaic together as a district and keep all of the Newark division in the central part of NJ. The resulting three districts would be NJ-Palisades (pop 2041K), NJ-Raritan (pop 4539K), and NJ-Pinelands (pop 2212K). They are not as equal in population as train's version, but do a better job of matching neutral definitions of CoI.

Speaking as someone who grew up in the Newark division...no, just no.  The divisions in urban northern NJ are not quite meaningless, but they simply are not the relevant cleavage at this scale. I can guarantee you that anybody who grew up in NJ would rather put Essex, Morris, and Sussex in with Bergen/Passaic than with Middlesex/Monmouth.  Maybe use them if NJ had four districts rather than three, or make the central district maximally underpopulated instead, but not that.

I guess what caught my eye is the split between Essex and Union which I cannot fathom. If my combination isn't to your taste I would suggest the solution is to move Middlesex/Monmouth/Ocean south and leave the Newark division all by itself. My best friend from Princeton would tell me that the Jersey Shore was more southern than northern, though that was back in the 70's and 80's.

BTW my plan for NY is almost the same as yours. I would leave Watertown (Jefferson and Lewis) with the west. The economic tie works better with Syracuse instead of the Hudson valley. As for names I used the largest borough for the two city pieces and so I get NY-Long Island (pop 2833K), NY-Brooklyn  (pop 4735K), NY-Manhattan (pop 3440K), NY-Hudson (pop 3579K), NY-Ontario (pop 4791K).

Yeah, Essex-Union is probably the most questionable part of that plan; if that's considered to be a deal-breaker, I'd probably rather just put Union up in the North and live with the extra-high deviations. 

My primary rationale for keeping Watertown in 4 rather than 5 was that the Watertown media market includes St. Lawrence County, which I also obviously wanted to keep with the rest of the North Country.  So I just ended up treating Watertown-North Country as one big block.

[traininthedistance]

I was editing my post above while you were typing. It seems to me that if one starts with the Newark division as a district (Essex, Hunderdon, Morris, Somerset, Sussex, Union) with isolated Warren and maybe Middlesex you get a population of 3332K, and by placing Monmouth and Ocean in the Pinelands you get 3419K there. The remainder is my Palisades district with 2041K and all are within the 2/3 to 4/3 of the state quota. Newark wouldn't have to choose between Monmouth and Bergen, and if Middlesex is still an issue, move it south as well but with greater inequality.

Well, I don't think that one should necessarily start with the Newark division as a district; it really doesn't have any meaningful identity to New Jerseyans in the way that "North Jersey" or "Central Jersey" does.  

It's completely insane to me, BTW, that the Newark district still exists, but the Edison, NJ division  (which consisted of Middlesex, Monmouth, Ocean, and Somerset, and mapped quite well onto the core of what people recognize as Central Jersey) was mostly folded into the "main" NYC district, which it shares with nearby places such as White Plains, and whose only connection with the rest of the district is the Outerbridge Crossing.  Not that there necessarily shouldn't be a Newark district, but I'd get rid of it way before I got rid of the Edison district.

Assuming that the Newark division is automatically more important and meaningful than what used to be the Edison division, and building a plan around that rather than local sentiment, is basically just GIGO to me.

[traininthedistance]

My bias in this exercise is to create a robust set of data for jimrtex' exercise. To get that data I want to resist preconceptions that I might bring to the process, so I'm starting with neutral divisions as determined outside the exercise. The Census groupings of metro areas adjusted by jimrtex' work on UCCs form a big part of that.

...

On the subject of UCCs, any reduction of the Atlanta metro below the UCC seems arbitrary. The UCC by itself is about the same population as the metro Boston population I described earlier, so it falls within the consideration that it is smaller than any state that gets two districts. The test of population weighting works better if districts aren't pushed too hard to be equal.

Well, these plans are supposed to be subject to local plebiscite, and what I'm saying is that local plebiscite will prefer, I can guarantee you, a larger North Jersey than your "Palisades" group.  I'm a fan of neutral criteria as well but I do want to make sure there's a fail-safe if they spit out something that's inappropriate to local sentiment, and not just take a 100% algorithmic approach.

Obviously there are other neutral criteria we can look at- for instance I'm leaning heavily on media markets for states outside of New Jersey (and within NJ as well, it's part of the reason I consider South Jersey set in stone).  Another piece of evidence we could use is area codes, which don't line up exactly with counties but can be illustrative all the same:



Some of the larger county cuts (I'm mostly thinking of how the southern tip of Ocean is in a Southern area code but the bulk is in a Central code) do pass the smell test.  Worth noting that this would actually be evidence for putting Warren in Central Jersey.

I'm gonna say that if we actively want larger deviations to test things, the Central district should just be the old Edison division (Middlesex, Monmouth, Ocean, Somerset) and North should be the overpopulated one.  That would be fine by me and probably fine for most Jerseyans.

As for Atlanta I have no objecting to drawing a larger metro district, but I suspect that the residents of those exurban counties would object.  Another option for Atlanta, actually, would be to draw the inner core district even tighter, getting rid of Cobb and Gwinnett, maybe adding a couple counties to the south, and turn it into a full-on VRA district.

[traininthedistance]

Anyway, one more proposed map for now and then I'm taking a break: Pennsylvania.



District 1: GREATER PHILADELPHIA.  Population 4,009,011 (deviation +833,416).  Obama 66.5%; roughly 21% black.  You could also call it "Southeastern Pennsylvania" or "Delaware Valley", I prefer to use city names for major metro cores.  Anyway, putting the five counties together is a no-brainer here.  Safe D.

District 2: NORTHEASTERN PENNSYLVANIA.  Population 2,690,334 (deviation -485,261).  Obama 51.9%.  As always, I'm going along media markets for most of these lines; the Scranton-Wilkes-Barre market (which includes Williamsport, the Poconos, and coal country) is the core of this district.  Reading and the Lehigh Valley are included, as well as Tioga and Pike, which get their broadcasts from out-of-state (Elmira and NYC).  Tossup.

District 3: CENTRAL PENNSYLVANIA.  Population 2,805,334 (deviation -370,261).  Obama 43.0%.  The spine of the "T", Pennsyltucky, whatever you want to call it.  The Harrisburg and Johnstown markets, as well as two small Northern Tier counties that are closer to Buffalo.  Really the only close judgment call I waffled on was whether to put them- McKean and Potter- here or in District 2; I chose this so as to not stretch the definition of "Northeastern".  (Tioga could go here too, I guess, but that might be a bit erose.)  Safe R.

District 4: WESTERN PENNSYLVANIA.  Population 3,197,700 (deviation +22,105).  Obama 50.7%.  Pittsburgh, Erie, and points in between (including one county that's part of the Youngstown, OH area).  Tilt R.

[traininthedistance]

That's ok, but I'm not wild about linking the lower Susquehanna to parts across the mountains to the west. I put everything from State College west with Pittsburgh and despite the higher population it's only a little over the 4/3 recommendation and less than the population of the largest state with only one district.

Since the lower Susquehanna region is smaller than the city of Philly, I went with that as a stand-alone county-city. That left me linking the whole Susquehanna valley together and adding the Scranton region as well. The economic connections (eg. I-81) make more sense to me and figure into a lot of my choices. That leaves the Philly and Allentown metros (minus Philly) as the remaining piece. As you may note, I don't like directional designations for the districts, so here I went with major river valleys: PA-Allegheny 4455K, PA-Susquehanna 3299K, PA-Delaware 3423K, and PA-Philadelphia 1526K.

And yes, I'm planning a map of all 22 states with divisions when I'm done.

I'm... deeply not okay with severing Philly from its region at this level of coarseness (and, yes, I know Reading is technically in the metro, but everyone thinks of it as the five counties with Reading being a Trenton-esque satellite, and countless organizations both public and private treat SEPA as just the five counties).  I guess I also don't really see the issue with south-central PA going west; perhaps Johnstown would prefer to be in the Pittsburgh district but I'm skeptical that combining Harrisburg-Lancaster-York with the Scranton region is actually any better.  They are Steelers fans in Harrisburg, after all.

[traininthedistance]

Here's my take on Tennessee:



District 2:  West Tennessee.  Population 3,260,832 (+87,780).  64W/27B/6H.  VAP 67W/25B/5H.  49.8% McCain, 51.1% DEM.  Tossup - Tilt D.

District 2:  East Tennessee.  Population 3,085,273 (-87,779). 88W/6B/3H.  VAP 90W/5B/3H.  64.8% McCain, 59.3% GOP.  Safe R.

Never thought I could make a 90% White district containing more than 3 million people in the South.

Tennessee's a dificult state to do because of its elongated shape and relatively even distribution of its surprisingly urban population.  A Nashville/Memphis district and a Mountain district probably highlights the political cleavages in the state the most (which, ATST, shows a strong adherence to the community of interest standard).

I could have gone for a Nashville + far-flung exurbs district, but results in the other district being a very strange U-Shape that is visullay unappealing and puts Memphis and Chattanooga in the same district :/

This map also probably represents the best Democrats can hope to get in Tennessee; a Blue Dog-type would easily win Tennessee West while Tennessee East could elect a true fire-breather if they wanted to.

I've mostly been drawing districts with deviations too low for muon2, but Tennessee is a state where I might actually be inclined to push that envelope: I feel like you want to stick to the canonical Grand Divisions if you can, and so I might just put East TN (the most populous division) by itself, and live with an oversized West+Central.  And, yes, I know that doing such a thing would be an R gerrymander... I don't care.

ETA: Not really "gerrymander", actually, Tennessee is Republican enough that if it has two districts they should probably just both be solid R. 

[traininthedistance]

Most of these are pretty good.  The nitpicks/questions I can think of just off the top of my head:

*Obviously I'm still very much not a fan of your PA, and also prefer a tighter Atlanta (if only for VRA purposes more than anything else).
* Indiana's a little jagged; I wonder if the north district could push south a little bit without disturbing COIs horribly?
* Putting NoVA and the far Southwest seems pretty bad to me; I can see how it's somewhat of a chain where NoVA bleeds into the northern edge of the Shenandoahs and the southern edge of same bleeds into the Southwest, but the final result is kind of a mess.  I'd guess I'd probably prefer giving the SW to the southern district, and giving NoVA the Northern Neck instead.
* In NC, we've got three major metros (Triangle, Triad, and Charlotte), and I'd prefer they each get their own district (which is why the giant Eastern NC doesn't actually bother me).  Probably put the Triad in with Blue Ridge rather than in with Charlotte?  
* I'm more or less fine with Wisconsin, but I also wonder about putting Madison in with Milwaukee rather than Sheboygan/Fox River/Green Bay.  I guess that's more of a judgment call along the lines of what you articulated with Missouri, where either E/W or N/S are reasonable.

I can't think of anything else that really bothers me; obviously one could quibble over a county or two here or there but the basic shapes seem good.

[traininthedistance]

Thanks.

Your first dot point raises an interesting question. I didn't use race as a CoI, but one certainly could. If I did, it would make a more compact Atlanta district by picking out those counties with significant black population (eg Fulton, Gwinnett, DeKalb, Clayton, Rockdale, Newton) within the metro. Would that do better at gaining a plebiscite? If so, perhaps something like the Minority County Cluster should be applied.

That same point also raises your concerns with PA. Yet if VRA concerns come into play, it would seem that the best result for minority candidates would be a district with Philly by itself where blacks make up a plurality of the population.

For IN I looked at those areas that were tied to Chicago as much as Indianapolis. Certainly the NW corner, South Bend and the western edge fit. Purdue seems to look as much to either center, so I kept it with the north. Ft Wayne is a border line case but feels more like a northern IN city from my experience. The guidelines seemed to make it clear that shape didn't matter that much for this exercise.

I agree that VA was a challenge and it would have been much easier with three districts since then NoVa could sit by itself, but population needs require more. The SW is the least like the rest of central and southern VA, and has the natural connection along I-81 to the Shenandoah Valley.

NC is often described in terms of the mountains, piedmont, and coastal plain. I found it hard to go against that. If you think a plebiscite would fail if the Triad was in with Charlotte, then I could rethink it, but I'm also thinking that Asheville would rather just be with the mountain counties.

I've spent a lot of time in WI and have family there to boot. People often see the political angle, but economically the cities along Lake Michigan are generally older industrial ones you can find all along the Great Lakes. Once you get inland the cities bear more resemblance to small town Midwestern communities with agricultural roots, though Madison has grown quite large due to the Capitol and university.

Yeah... PA, GA, and MI all have similar issues of how tightly to draw the core, but I don't think I have the time to really sort through my thoughts on that tonight.  Hold that thought, I'll get back to it soon.

Your explanation of WI sounds convincing and I'l happy to go along with it.

As for Virginia, that's pretty much the explanation I was expecting, and I guess my rejoinder would be that, for all the SW is somewhat dissimilar to the rest of south and central VA, it's far more dissimilar to NoVA.  I'll also refer again to media markets, which I've been leaning on for some of the maps here, and which seem to indicate that perhaps Roanoke and Lynchburg shouldn't be separated:



Taking those regions (and their concordance with metro areas, UCCs, etc.) as building blocks, it seems easiest to sort them roughly as thus:



Deviations 865,177.  One could possibly put the Charlottesville area in the north, which would lower the deviations.

On a similar note, the North Carolina media markets do cross the Piedmont/mountains a couple times, and might actually suggest putting Charlotte in with Asheville and having a tiny Triad district instead... which might be too tiny in fact, haven't plugged it in yet.  Eh, I dunno.

[traininthedistance]

When you say SV, do you mean Santa Clara County? If so, it's an inseparable part of the Bay Area; despite the Census Bureau's definition of the San Francisco metropolitan area as separate from the San Jose metropolitan area, there is no definition of the Bay Area that excludes Santa Clara County.

If shoehorning San Jose into a non-Bay Area district is off-limits (as it should be) there are then only three potential options:

1) Draw a district consisting only of the Central Coast
2) Connect the Central Coast to the San Joaquin Valley
3) Split the Central Coast between the north and the south, connecting the northern part to the Bay Area and the southern part to greater Los Angeles

The first solution is obviously unfeasible from a population standpoint. The second solution might keep the Bay Area and the Los Angeles area intact, but it connects two areas that are not connected either culturally or by major roads. That leaves the third option, which involves connecting the parts of the Central Coast that are in Northern California (Monterey County and points northward) with the Bay Area, while connecting the parts that are in Southern California (San Luis Obispo County and points southward) with greater Los Angeles. This line is basically non-negotiable; lumping in Salinas with a Southern California district is straightforwardly wrong, as is grouping San Luis Obispo with a Northern California district.

I haven't looked to see whether the numbers work by splitting the Central Coast like this. If they do not, then the only conclusion is that California simply does not work with this number of districts and this level of acceptable deviation.

A ten-second look makes it pretty clear that, to comply with your Option 3, all muon would need to do is shift San Luis Obispo south- and such a move would actually decrease variance anyway.  

EDIT: I misread where San Francisco itself was, which complicates matters.  But presumably removing San Luis Obispo would render that split of the Bay Area less objectionable?

(Yes, muon, I still owe you my thoughts on Philly/Atlanta/Detroit.  Comin' soon.)

[traininthedistance]

The Edison division can become the Jersey Shore district and is within range. To fix the southern district I note that Warren county isn't part of the NY UCC, but is part of the Allentown UCC and like South Jersey is oriented towards PA. Shifting that would be enough, but looking at population, I can move Hunterdon as well which is only loosely connected to the Newark division. If one wanted to further smooth the populations, they could shift Sussex as well. Here's the NJ map I come up with.



Palisades 4003K; 2008 pres: D 60.7%, R 38.5%, O 0.8%
Jersey Shore 2340K; 2008 pres: D 50.2%, R 48.6%, O 1.1%
Delaware (NJ) 2449K; 2008 pres: D 58.7%, R 40.0%, O 1.2%

I quite like this map, FWIW.  The biggest question is of course whether Hunterdon and Warren would like it.  And putting Sussex in the Delaware district, while a little weird by current CoI standards, would do a good job of reconstituting the old West Jersey:

 

Though, while Warren is part of the Allentown metro and thus is nominally oriented toward PA, it's not like the county is entirely separate from the NYC orbit: the eastern portion is more NYC exurbs and the Allentown connection comes in mainly through Phillipsburg; and Allentown is part of the New York CSA after all.

[traininthedistance]

Since I abhor dull names for this exercise, I prefer geographic names that would be identifiable by residents of the state even if they are not inclusive of the entire area. Naming the district after a famous person from the district would beat a dull name here. Directional names should only be used when they are part of a specific geographic feature, and city or county names should be reserved for districts that comprise only the city or county in question. My revised GA plan based on the above CoI would be as follows.



Kennesaw Mt: 4253K, Obama 34.7%, McCain 64.4%
King: 2393K, 46.1% BVAP, Obama 68.6%, McCain 30.8%
Ocmulgee: 3041K, Obama 45.6%, McCain 53.8%

I certainly think this would be an acceptable Georgia; and quite possibly preferred if we're taking the VRA into account.

Regarding what I was going to say earlier: I noticed that the Philly, Detroit, and Atlanta metros all were somewhat similar in size- 7th, 11th, and 12th in current rankings, and if you remove the New Jersey portion of Philly they're very close indeed.  All are in between one and two districts at this size, ideally.  And, in addition, all of the three have large black populations that support between 1 and 3 VRA districts at our current 435-district size.

So it would seem to make sense to treat them somewhat similarly on these maps, and either draw them all with expansive borders, or all with close-in borders.  But there are differences as well: Philly's one black-majority district is the result of a much smaller AA population than Georgia's three districts; and there is of course the accident of very different historical political boundaries:

*Philly has the most expansive city boundaries, being a combined city-county; Atlanta is very penned-in and takes up about 10 percent of the metro area; Detroit is in the middle...
*In terms of central county sizes, Wayne is the most expansive, and Fulton again is the smallest, both relatively and absolutely.

There's also the larger philosophical question of how much animosity there is between the (largely AA) central cities, and the (largely white) suburbs, and whether we want to be encouraging the suburbs to identify as a separate unit from the central cities, or as one unified metro area.  Obviously my bias is toward a rapprochement between city and suburb and a recognition that urban areas don't always end at the exact town lines (and that the suburbs need the central city just as much as the city needs the suburbs), which I hope explains my horror at cordoning off Philly in particular- at least, whatever can't be simply explained by the fact that Philadelphia County is really underpopulated, beyond jimrtex's original guidelines.

I guess, when looking at those three areas, I'd say that it's somewhat tricky to figure out how to draw the lines when the metros are obviously too large for one and too small for two districts, so at the end of the day while I'd like to try and be consistent in my principles you have to look at the differences between them, both in terms of their internal composition and their neighbors.  The PA portion of the Philly district is not too overpopulated, and is surrounded by other metros, (especially if you slice off Berks, which is legitimately separate), so I'd go with maximum inclusivity.  I'm more willing to draw tightly around Wayne County, though, because it's a tougher judgment call as to which suburban counties to include, and also in recognition of the sad fact that the city/suburb animosity is particularly keen there. 

As for Georgia, my original map went for a bit of a middle ground (which also happened to minimize deviations) since Fulton/DeKalb/Clayton would be too small and the entire metro would be too large.  Though, Atlanta is also the metro with the highest AA population and thus the one case where the VRA could plausibly be triggered (the Detroit and Philly black populations just aren't quite large enough I think), so upon further reflection your most recent map, with its black-plurality Atlanta & south core, is also a solid option.  At this population size there are very few genuine opportunities for minority districts, so it makes sense to take the opportunity here.  (Where else even is there?  I guess NYC, Chicago, LA, South Texas, and Miami?  And none of those are quite comparable, excepting possibly Miami-Dade.)

Bah, this is all quite a meandering hash.  Hope I made a little bit of sense at least.

[traininthedistance]

Thanks for the comments, since you can see I really do try to integrate good ideas. I went back and forth on Sussex, since my memories from 30 years ago had virtually no NYC exurban area out there. If you think that there's still a good case to keep the Delaware Water Gap with the rest of the new West Jersey (Morris is obviously too much in the Newark metro) I can move it. The fact that the NYC metro is so large and has to be split, along with the fact that the NYC CSA extends into PA, makes it easier to split the western counties from Newark.

As you can see I came to agree with you on Philly. What do you think of the rest of my PA solution? Allentown may not be an ideal fit for the Lower Susquehanna, but I liked it better there since it allowed me to keep the mountain counties together.

I'm in the process of creating another full US map, this time with PVIs for all the districts, so I appreciate the input.

Thanks.

I think I still prefer putting Allentown with Scranton-Wilkes-Barre and the rural Northeast; and south-central PA with the rural Appalachians... but that bothers me a lot less than the Philly split did.  I guess my main pieces of evidence for splitting PA that way would be that Allentown's metro does cross Blue Mountain and include Carbon County, while both Allentown and the Poconos have some extra-peripheral NYC ties; whereas south-central PA and the Alleghenies are (mostly) linked in opposition to those other areas by being Steelers country.  It's not an airtight case, I'll admit; the Steelers connection breaks down in eastern Lancaster and Berks is a plausible connection between Dutch Country and the Lehigh Valley. 

I feel like that's something that would make a good plebiscite; hold SEPA and the West constant, and have people vote on which way to split the other two districts. 

[traininthedistance]

Unlike GA where there were core counties with 40%+ and 50%+ BVAPs, Wayne county MI only reaches 39.2% BVAP and I can't justify keeping it apart with its small population. I also can't see adding only one of the adjacent counties from a CoI view since if either Oakland or Macomb is added, so should the other. With that in mind, I took jimrtex's suggestion of shifting the Lansing UCC to the east to get this map.



Mackinac 3400K
Huron 2620K
St Clair 3864K

Yeah, in the end that's probably the best way to do the Detroit area.  I had forgotten just how small Wayne actually was; remembering that I think this more expansive three-county district would be preferred.

[traininthedistance]

Here's version 2 of my draft plan. In addition to the population I have included the PVI of each district in square brackets, with positive values for D PVIs and negative numbers for R PVIs.

Looks quite nice.

I guess my remaining quibbles would be:

1) I prefer jimrtex's Florida; crossing the Everglades like you do in the Fort Lauderdale-Naples district seems like a poor idea, and I like that he keeps Miami-Dade mostly by itself (obviously Monroe is stuck behind it) on VRA grounds.
2) Virginia still rankles.  Whatever problems one has in splitting Roanoke from the Shenandoahs pales in comparison to the silliness that is mashing the southwest and NoVA together.
3) What I said about PA, though that is much more minor than the issues with FL and VA.

FTR I agree with your rationale in putting Calvert with the DC district rather than the Baltimore one.

[traininthedistance]

Of course Monroe would best fit with Collier geographically, but that would add further to the Tampa overpopulation. That leaves it going with the east side counties.

Does the Hispanic CoI in Miami-Dade justify overpopulation of the Tampa district?

Does it?  Monroe really only has road connections to Miami-Dade; the border with Collier is unpopulated Everglades.

I suspect that I'd prefer an overpopulated Tampa district (or chopping off Hernando) in this case, not just because of the Hispanic CoI but also because of the dichotomy in FL between the Gulf and Atlantic coasts.

[traininthedistance]

So I think I can stipulate the following:

Avoiding the cross-peninsula district and preserving the Tampa UCC would justify a district at 1.35 of the quota, but no more.

Polk is sufficiently independent of both Tampa and Orlando that there is no compelling reason to place it with either.

The Hispanic CoI does not require a stand-alone district for Miami.

The revised guidelines suggest that some consideration be given to population equality, so why not keep M-D with Broward, and place Polk and the other inland counties in the Okeechobee district. The result reduces the difference in population between the largest and smallest districts.



Appalachicola, 3467K, O'08 41.8%
Tampa Bay, 4586K, O'08 49.7%
Cape Canaveral, 3693K, O'08 51.0%
Okeechobee, 2738K, O'08 54.1%
Everglades, 4317K, O'08 62.1%

What are the Hispanic #s on the Everglades district?  I assume it's plurality but not majority?

[traininthedistance]

Edit: Left Portsmouth out of Tidewater in Plan 3, Winchester city out of Southern Virginia Plan 2 and 3.

I think the first of your three Virginias is clearly the best- it seems to me like Richmond and Tidewater probably ought to stay together.

I generally approve of your compromise PA as well; perhaps a few rural counties (such as Clearfield, Elk, and the like) could also be shifted from the West to the Northeast to lower deviations if necessary.