Georgia 2020 Redistricting Discussion

[Mr.Phips]

I decided to give an uninvited "lecture" on RRH, I thought perhaps, however delusional, some might be interested in reading if here. So below is my copy and paste of what I posted.

A poster the other day suggested that the Georgia “Pubmanderers” draw 4 black CD’s in the Atlanta area, and dismantle the rural black SW corner CD. I thought I would test that suggestion out. The Atlanta part of the equation works out quite well, and may even hold up for a whole decade – maybe. The second prong of the equation, the Pub snatch of the SW corner CD, however does not. It does not because although it can be drawn, its lines would be illegal, and tossed by the courts, and it is not a particularly close case in my opinion. Below explains the basis for my opinion.

The first step is how many CD’s need to be performing black, before the VRA no longer mandates additional minority performing CD’s. If it is only 4 CD’s, then the SW corner CD can be dismantled, and the plan fully executed. If it is 5 CD’s, then the VRA is still in play to police the CD lines. The answer turns out to be 5, not 4 CD. Blacks are about a third of the population of Georgia now. Thus, out of the 14 CD’s, under the VRA their allocatable share of the 14 CD’s is 14/3 or about 4.7 CD’s, which must be rounded up to 5 CD’s, if 5 “compact” CD’s can be drawn that are 50% BCVAP. In point of fact, they can (particularly when the racial percentages in Georgia are updated from the DRA figures which are circa 2015).  And yes, one of the 5 CD’s is the rural SW corner CD; like this, nice and compact:

https://i.imgur.com/uMuXm9O.png

So, the map the Pub gerrymanderers’ may want (the map to the right):

https://i.imgur.com/rHHHwCY.png

is illegal under the VRA.

And thus, they will need to draw the map to the right above, and even with that map, if the SW corner CD were not a performing black CD, its lines had better not be gerrymandered in any way, to wit, where there are reasonable choices of adjacent territory to take in, one always chooses the more black, and more Democratic territory, which is what I did. Trump 2016 still carried it by a couple of points, but Trump 2020 probably lost it by about the same margin. If still not performing, arguably one might need to chop in to Macon, or duplicate the lines that I drew (the first map that I linked above) that get the CD up to 50% BCVAP. Trump 2016 lost that iteration of the SW corner CD by 10.5% by the way.

I have seen a number of maps drawn on RRH that I think are very vulnerable to being struck down under the VRA. If one is going to gerrymander in a way that puts in play minority voters to their detriment, the VRA’s byzantine rules need to thoroughly mastered on a very granular level.

The talk of dismantling black majority districts like GA-02,black plurality ones like FL-05, and even ones with a large black minority like MO-05 is interesting because a district held by an African American has never actually been dismantled.  Even dismantling MO-05 would cause quite the outrage among black leaders in the state.

[Torie]

Yes, none of this is defensible on policy grounds. Anyway, here is the Pubmander that might hold up, with a most unfortunate extra chop into Gwinnett, because I think the CD I labeled 8 is the most likely to trend the most to the Dems over time since 8 is dominated by Gwinnett, as Gwinnett and the northern bit of Fulton will probably keep surging to the Dems (along with the northern bit of Cobb, which is why of course they all are parked in different Pub designed CD's). 2022 is the last picture show for the Pubs in Georgia. After that, the state will be lean Dem heading to safe, and by 2032, a new map will most likely be the reverse, 5 Pub seats, and 9 Dem ones. It may even be whittled down to 4.

Georgia aside (as a canary in the mine), the Pubs had better figure out how to keep moving on up with people of color, particularly Asians and Hispanics, or they will be out of business. I don't think urban  metro area higher SES whites are likely to come back to the Pubs, as opposed to  continuing their exodus, unless the Dems soak them with taxes, and I don't see that happening except at income levels that only hit the top 5%, if that. Right now the Dems are chatting more about the top 2% to hit. Those folks are rich enough to not care that much about taxes, at least at levels that are at European levels.

I modified the map to make the gerrymander as efficient as possible, while being careful to avoid the VRA quicksands, taking cognizance of where the strong Dem trends are most likely to manifest themselves going forward. It takes work to transfer the north Georgia Pub sink elsewhere, sometimes up to a 100 miles away, I must say. (And Athens is most inconveniently located - it needs to be moved north to the SE corner of the state ASAP.) But somebody has to do it!

[lfromnj]

https://www.georgialawreview.org/api/v1/articles/7669-the-history-of-redistricting-in-georgia.pdf


Very interesting read I found, albiet a bit old.(download link for PDF)

[beesley]

Is the current GA-02 majority black?

[Torie]

Is the current GA-02 majority black?

Per the 2010 census numbers it was 52.3% black in population, and 49.5% BCVAP. The 2018 numbers are whited out as an option. The almost identical CD that I ended up drawing above was 50% black, and 47.1% BCVAP in 2010, and 51.3% black, and 49.5% BCVAP in 2018

[Torie]

Well I revised the map, for a good reason from the Pub perspective. Can anyone surmise what my concern was?

[Stuart98]

Gwinnett district wasn't safe enough, I'm guessing?

[Torie]

Gwinnett district wasn't safe enough, I'm guessing?

That was a minor concern, but now that minor concern has transferred to the yellow CD (and actually by virtue of that transfer of the vulnerability to the yellow CD become slightly more of a minor concern), so that is close to a wash but on the downside. Moreover,  that just short of a wash to the downside substantially increased erosity, so the changes in that quarter were a material net negative. So, while good thinking on your part, that was not what drove me to otherwise degrade the map when I moved the chess pieces around. There was something else out there that made me do it!  

[Stuart98]

Are you trying to get rid of MTG by forcing her to run in a district with an incumbent other than her? I'm... skeptical that would work.

[Torie]

Are you trying to get rid of MTG by forcing her to run in a district with an incumbent other than her? I'm... skeptical that would work.

I never gave that crazy women the slightest thought. Thanks for reminding me, not, about her. Why is Georgia such a hospitable place for crazy female Pub politicians? Although not in MTG's league by any means, the state also had the Stepford wife who was like Hal in that Sci Fi flick, as she kept mechanically in a disembodied monotone repeating the same lines in her Senate debate - radical extreme socialist, radical extreme socialist, radical extreme socialist ... zzzz.

[Torie]

Well, as a hint, here is a rather substantially revised map, because the revision above did not fully mitigate the risk. This map I think is the best one can do to minimize it, while keeping 9 quite safe Pub seats that should hold for the decade absent even more substantial plate tectonic shifts than occurred in the last decade.

[President Punxsutawney Phil]

https://davesredistricting.org/join/2eebf1aa-e053-4ea6-993a-4e95a1ce0d05
Behold...a 10-4 map

The Kemp seats voted for him by margins of 36, 54, 33, and 58 points respectively. Every other seat went to Abrams.

[President Punxsutawney Phil]

Non-partisan Georgia state senate map drawn with election data hidden.
Emphasis was put on avoiding crossing county lines.
There are 22 districts that voted for Stacey Abrams.
https://davesredistricting.org/join/9aa28a5c-843e-43a4-b43a-87d6ca292981

[Idaho Conservative]

Well, as a hint, here is a rather substantially revised map, because the revision above did not fully mitigate the risk. This map I think is the best one can do to minimize it, while keeping 9 quite safe Pub seats that should hold for the decade absent even more substantial plate tectonic shifts than occurred in the last decade.

not convinced the Gwinnett seat would hold

[Torie]

The quad chop of Gwinnett is a collateral damage of my concern, but a seat could be drawn to very likely hold for the decade without the quad chop, as my previous maps did with a tri-chop. Heck, in theory even one chop would have been enough if the two pieces were relatively equal in population (in fact the share of Gwinnett for the Pub GA-8 (the purple CD) is about half of its population give or take), and that would be my first inclination if the chop were clean and avoided unpleasant municipal chops (you might notice that in the Atlanta area in particular I made a real effort to respect municipal lines, arising out of the same concern). I kept twisting the clock around Atlanta for a very specific reason, and, hint 2, it was driven by the characteristics of one and only one CD. There was something about it that raised a red flag for me that moved me to call RNC headquarters, and say, hey, I am worried about this, I think it has the potential to be like the corona virus is to the human body, particularly one like mine.

[Tintrlvr]

The quad chop of Gwinnett is a collateral damage of my concern, but a seat could be drawn to very likely hold for the decade without the quad chop, as my previous maps did with a tri-chop. Heck, in theory even one chop would have been enough if the two pieces were relatively equal in population (in fact the share of Gwinnett for the Pub GA-8 (the purple CD) is about half of its population give or take), and that would be my first inclination if the chop were clean and avoided unpleasant municipal chops (you might notice that in the Atlanta area in particular I made a real effort to respect municipal lines, arising out of the same concern). I kept twisting the clock around Atlanta for a very specific reason, and, hint 2, it was driven by the characteristics of one and only one CD. There was something about it that raised a red flag for me that moved me to call RNC headquarters, and say, hey, I am worried about this, I think it has the potential to be like the corona virus is to the human body, particularly one like mine.

In your original map, were you concerned about the rapidly growing black population in GA-10? Including Rockdale in a Republican district was a bold choice, and Newton and Spaulding are also changing quickly. Long-term, I think there's a real chance that Walton starts to trend hard to the Democrats as the black population grows as well; the only adjacent suburbs to it these days are majority black.

Oconee County isn't in exactly the same situation (it's an Athens suburb, not Atlanta, and there is unlikely to be a surge in the black population there any time soon), but it is definitely trending Democratic as well due to spillover from Athens.

All in all, I could see GA-10 on the previous map being a concern by 2030. Taking out Rockdale and Oconee definitely helps with the long-term risk.

[Torie]

The quad chop of Gwinnett is a collateral damage of my concern, but a seat could be drawn to very likely hold for the decade without the quad chop, as my previous maps did with a tri-chop. Heck, in theory even one chop would have been enough if the two pieces were relatively equal in population (in fact the share of Gwinnett for the Pub GA-8 (the purple CD) is about half of its population give or take), and that would be my first inclination if the chop were clean and avoided unpleasant municipal chops (you might notice that in the Atlanta area in particular I made a real effort to respect municipal lines, arising out of the same concern). I kept twisting the clock around Atlanta for a very specific reason, and, hint 2, it was driven by the characteristics of one and only one CD. There was something about it that raised a red flag for me that moved me to call RNC headquarters, and say, hey, I am worried about this, I think it has the potential to be like the corona virus is to the human body, particularly one like mine.

In your original map, were you concerned about the rapidly growing black population in GA-10? Including Rockdale in a Republican district was a bold choice, and Newton and Spaulding are also changing quickly. Long-term, I think there's a real chance that Walton starts to trend hard to the Democrats as the black population grows as well; the only adjacent suburbs to it these days are majority black.

Oconee County isn't in exactly the same situation (it's an Athens suburb, not Atlanta, and there is unlikely to be a surge in the black population there any time soon), but it is definitely trending Democratic as well due to spillover from Athens.

All in all, I could see GA-10 on the previous map being a concern by 2030. Taking out Rockdale and Oconee definitely helps with the long-term risk.

You sir are almost there when you refer to Rockdale, Spaulding and Newton Counties.  (I was suspicious of Oconee for the reasons you stated, and that is why when I moved the chess pieces around the final time that I put it in the Cyan CD, but that is not what propelled my last set of changes.)   You are so so close, but alas still not there, as to what my energizer bunny was. In short, you did not guess the CD I was worried about, and why, even though you were in the zone as to counties involved in the drama.

Hint 3: Why in my penultimate set of changes did I remove Fayette from the lime green CD, which CD  was overall totally safe from the specter of an "invasion" of blacks and educated whites or Asians into Fayette eroding away its Pub pad to the point that it might slip into the twilight zone of a competitive CD? Heck, that was the only "useful "work" that lime green was doing; it was like 70%ville for the Pubs. In the final set of changes, lime, in lieu of the brown CD doing it, was instead given the larger task of "pacifying" northern Cobb from its expected invasion of "undesirables," so lime was still pulling its weight just fine (65% Trump 2016, which is just about the perfect % for dealing with the Cobb problem, with no excess Pub waste.

But now the most Pub place of all, the northwest Georgia corner and the brown CD, isn't doing any "useful" work at all! It's on welfare and a breeding ground for Pub crazies, that the Dems can run ads about nationwide. The damn place in my last map gave Trump 2016 80% of its vote, and it's basically illegal for blacks and educated whites to live there, so it will still be on the Pub team even if close to all of the other 434 CD's slip away.  Why on earth did I do that? Am I a Dem mole or something?

[Sol]

You could just be explicit--I can barely parse that post.

[Torie]

I apologize for my irritating people. I wanted to give whomever a chance to surmise what my concern was. Under the VRA, when you have a black pack issue, you had better be extra cautious that your map in that area cannot be characterized as gerrymandering the black pack CD's. So one draws CD's as if race (or partisan preferences) was not an issue. Fayette is part of the Atlanta metro ring, and the southern edge of the metro area CD's had better coincide reasonably with the metro edge, with no county chops that appear gratuitous. Thus, for example, carving out strong Pub precincts in the Southern counties was an absolute no-no from the get-go, and one can see in all three maps I avoided southern tier county chops. However, given the dangerously high black percentage of the green CD, that was not enough. So, I first added Rockdale County to the green CD, but still too high a black percentage. Therefore, as seen in draft 3,  I felt compelled to add Fayette County to the green CD, the missing county from the southern tier of metro Atlanta, which got the percentage down to a less dangerous percentage, leaving no southern edge metro counties behind, and the green CD nice and compact.

The fact that map draft 3 reduced the green CD's  chop into DeKalb to one much smaller in size, and the size of the chop of the blue CD in Gwinnett, demonstrated just how important it was to make the change, to reduce the VRA risk. However, map draft 3  in turn necessitated another chop into Gwinnett by a 4th CD (GA-10) to Pub up my GA-08, but gerrymandering white areas where black CD's are not in play, or otherwise gerrymandered, is far less of a risk under the VRA. And with all 4 black CD's, I also avoided municipal chops to the extent possible, where the precinct lines allowed me too.

Whether you agree with me or not is fine. But that was the issue as to which I wanted to share my thinking. I don't think the VRA is very well understood by many of those drawing maps, including parsing its ambiguities and how it may well evolve. That is understandable. It is  a very complex legal issue. So I thought I would make my little contribution to the public square, for better or worse.

[Torie]

Well I think I am done now. Gwinnett and Cobb each apparently have townships the precincts in which can be identified by the precinct name, so I adjusted the lines in Gwinnett and Cobb to hew to those (only one township chop in one place), and it actually makes the lines cleaner and more pleasing to my eye. The lines on the DRA map to which precinct lines do not hew are apparently villages or something.

Not much change in anything, except a bit to the better given the purpose of the map: both GA-04 and GA-01 drop a point in their black percentage, and GA-08 actually goes up a fraction of a point in Trump 2016's percentage. And I suspect in Gwinnett one of the biggest Dem trends in 2020 was in Berkshire township (a fair number of higher SES whites there maybe), so putting all of Berkshire in GA-01 may have increased the Trump 2020 percentage in GA-08 by close to a point from what it was.

[Torie]

The codicil. Assuming the 4 black performing CD's in the Atlanta metro are drawn the way one would draw them given the VRA, free of gerrymandering otherwise, and we get rid of the gerrymandering of the white CD's, below is what one might find. The Gwinnett based CD is gone for the Pubs out of the box, GA-05 is unlikely to hold for the decade, and GA-12 may also fall to the Dems before the decade is out. So the legal gerrymander for the Pubs, buys it one CD right away (and an outside chance of even 2 CD's), and one to two more over the course of the decade that might otherwise fall. So the Pubs might actually for the short term pick up a seat in a "fair map," on a relatively short term lease.

Yes, I know, some might not consider the map "fair." That is OK. And maybe it is not the best map that can be drawn. But I suspect most would consider it reasonable, at least in general perhaps, if not here.  

[muon2]

The codicil. Assuming the 4 black performing CD's in the Atlanta metro are drawn the way one would draw them given the VRA, free of gerrymandering otherwise, and we get rid of the gerrymandering of the white CD's, below is what one might find. The Gwinnett based CD is gone for the Pubs out of the box, GA-05 is unlikely to hold for the decade, and GA-12 may also fall to the Dems before the decade is out. So the legal gerrymander for the Pubs, buys it one CD right away (and an outside chance of even 2 CD's), and one to two more over the course of the decade that might otherwise fall. So the Pubs might actually for the short term pick up a seat in a "fair map," on a relatively short term lease.

Yes, I know, some might not consider the map "fair." That is OK. And maybe it is not the best map that can be drawn. But I suspect most would consider it reasonable, at least in general perhaps, if not here.  

The map looks nice, but one question I have is whether you tried to take into account the full Atlanta metro as defined by the UCC, or was it impossible due the VRA constraints. With 2018 data, the UCC supports 7.3 CDs. You have 6 nested, and 4 CDs that split the remaining 1.3 in the UCC. By the UCC rules that's 1 pack and 3 chop points of penalty.

The use of the UCC may not be so academic as it seemed last decade. If the suburbs are realigning, then keeping more of them together may be beneficial to capturing the trend over the coming decade.

[muon2]

Here's my quick take on a division of the Atlanta UCC using the 2018 data.



The UCC has 7.3 CDs of population and the closest 7 CD arrangement of whole counties drops Coweta and Walton.

The overall population range is less than 1%. The 5 NW counties naturally form 2 CDs (CD6 and CD 7) with a deficit of 4700 people, and I wouldn't be surprised to see enough growth to cover that by 2020.

In the other counties the DeKalb chop became a macrochop and there is one additional macrochop in Fulton beyond what would be required by population.

CDs 1-4 are Black VRA districts with BCVAP of 51.7%, 53.7%, 50.7% and 52.7% respectively.

CD 5 follows the Atlanta city line in Fulton and the old township lines in Gwinnett with one chop. It wouldn't have been competitive in 2016, but has been trending more competitive with the suburban realignment.

CD 6 is completely nested in Cobb and is a competitive district.

CD 7 is uncompetitive R.

[Torie]

Here is my effort that avoids cover and pack penalties under the Muon2 rules, ignoring the VRA. It probably is compliant under the VRA however, given the 2020 election returns, and  when we get the racial data that will be used to assess VRA compliance. The Pubs will not be drawing this map! Btw, last night the DRA added the 2020 shape files, which changed the populations a bit in the lines that I had drawn for some reason, since it of course does not include the 2020 census population data. Another oddity is that the village and city lines on the map disappeared. I see that we did the exact same chop into Cobb. Great minds think alike.   I quite like my nested CD's, particularly on the north end of course, but the pink CD is well - butt ugly! It can probably be improved by playing with the lines between it and the SW corner CD up to a point without getting into VRA trouble.



Something like this perhaps:

[muon2]

Here's what I put together using the Atlanta UCC plan from earlier today. I stuck with whole counties, and brought the range down to 0.63%. Including the Atlanta UCC there are 6 county chops total which only 3 above the minimum required given the permissible range. There is one extra UCC
chops and no pack penalty.

The counties were grouped to minimize erosity by eye. I'd have to find an old erosity map of GA counties to see if there might be a grouping with lower erosity.

I didn't look at political data while drawing the plan, much like some independent commissions (eg. CA) might do. (I don't find averages including data before 2016 very indicative of current performance anyway). When they put up the 2020 PVI numbers I'll be more interested in drawing maps to better meet the political metrics of skew and polarization.