Texas Redistricting Competition.

[Torie]

"No seat shall be above 85% Hispanic"

This requirement forces fajita strips. It is not a VRA requirement to avoid that.

I think the VRA requires at least 10, and perhaps 11, Hispanic seats, i.e. seats that at least 50% HCVAP or performing Hispanic seats. My Pubmander had 11 such seats.

[Torie]

I can put up my TX Pubmander that I drew a couple of months ago, with Cinyc** giving me the 2020 POTUS numbers that are now available on the DRA, if the host of this thread deems that appropriate. The metric other than the damn thing being legal, was assessing the half life of the map (you know the last the decade thing), the care and feeding of Pub incumbents were possible, no matter how hideous they might be, and making dressing up the beast of the mander with the sartorial beauty of lines that please my artistic eye to the extent that does not unduly reduce the caloric content of Pub political gourmands with insatiable* appetites.

*I have read posts elsewhere yipping and yapping about 27R-11D maps. I lack the talent to satisfy that kind of appetite.

**He had the patience of Job as we went through about 15 iterations. It is a miracle that he choose not to cancel me out.

[Torie]

Here it is. I also when possible given the other metrics in play, tried to take cognizance of muni lines. That in the real world is often almost as important as county lines.

If you click the landmarks button, you will see where the Pub incumbents live.

https://davesredistricting.org/join/018eb38b-af4a-4029-bc3c-769262a56a79

Oh my TX-15 is 93% Hispanic, so my map is a fail given your 90% metric. Again, as long as a map is VRA compliant, that really should not be a metric. TX-28 is also a fail, and that CD was not gerrymandered at all, by design. Both CD's just rolled down the RG River. Between that and the next signs of much human life, is mostly rattlesnakes and lizards. Sad!

A couple of things about the VRA and Hispanics.  An Hispanic CD that is gerrymandered to make it Pub performing where there is a more compact Dem performing CD to be drawn depending on the facts is vulnerable under the VRA. Thus, for example, I have two Dem Hispanic CD's nested in Bexar County; I don't think an Hispanic Dem CD drawn by Pubs for partisan reasons that runs from Austin to San Antonio will cut it. A CD cannot count in TX as an Hispanic CD, if it is nether 50% HCVAP, nor Hispanic performing in practice. As to the latter metric, one needs to look at who will be voting in a Dem primary, if a Dem will be winning the General. Thus drawing my TX-18 needed to be drawn very carefully, since it is below 50% HCVAP, but it is triggered under Gingles because two 50% HCVAP CD's can be drawn in Harris County. TX-33 probably needs to be 50% HCVAP and thus the road bridge, and its demographics in a Dem primary.

[Torie]

The partisan scoring formula seems to turn on Dem packing, rather than maximizing the number of reasonably safe Pub CD's, plus whatever else can be put in play without putting the reasonably safe stable at risk. Whatever else in my map would be TX-07, and then the issue for a more marginal CD, is thinking about the trends going forward. For TX-07, do you jettison rather marginal former Pub high SES high rates of voting whites trending not so marginal Dem, for lower rate voting more Dem Hispanics but trending Pub, or not? I went with the Hispanics, jettisoning Bellaire and West  University Place and similar adjacent precincts, and embraced Alief and Bellaire West a couple of miles to the west.

In the real world, you also want to avoid CD's to the extent one can, where such a CD might nominate a Pub that is a very poor fit for higher SES urban whites, to the point that the CD might be put at risk. My Dem trending TX-3 for example is drawn the way it is for a reason. Keep the number of gun toting, MAGA hat wearing, rural hicks down to a point where they are put on ignore in Pub primaries! They are not wanted.

[Torie]

The partisan scoring formula seems to turn on Dem packing, rather than maximizing the number of reasonably safe Pub CD's, plus whatever else can be put in play without putting the reasonably safe stable at risk. Whatever else in my map would be TX-07, and then the issue for a more marginal CD, is thinking about the trends going forward. For TX-07, do you jettison rather marginal former Pub high SES high rates of voting whites trending not so marginal Dem, for lower rate voting more Dem Hispanics but trending Pub, or not? I went with the Hispanics, jettisoning Bellaire and West  University Place and similar adjacent precincts, and embraced Alief and Bellaire West a couple of miles to the west.

In the real world, you also want to avoid CD's to the extent one can, where such a CD might nominate a Pub that is a very poor fit for higher SES urban whites, to the point that the CD might be put at risk. My Dem trending TX-3 for example is drawn the way it is for a reason. Keep the number of gun toting, MAGA hat wearing, rural hicks down to a point where they are put on ignore in Pub primaries! They are not wanted.

Yeah I wasn't really sure what'd be the fairest way to have an objective metric. I'd figure that the formula would encourage as many R CDs in the sweet spot of R+15-25, since after that the amount of additional points you get for a redder districts reaches an asymptote (i.e. from a partisan standpoint, a R+25 and R+50 district are essentially the same).

There are certain decisions like primary electorate nominating someone who'd be a good fit for the district that are hard to measure, or how "trend proof" a given district is, since a lot of this is subjective (even if there is a consensus on Atlas).

Oh I see what you did now. That is outstanding. And I lost out because while in comparison, I an extra lean Pub CD, I also had other CD's that fell into likely but not totally safe Pub category, particularly as time goes by. I see now why you had +15 and an X exponent to 9/10th's to drive the asymptote. I am impressed. It works!

And yes, there is no way to put the rest of what I wrote into a scoring formula, even with consensus.

The main other issue is whether to somehow factor in muni chops, which might actually help to foster demographic coherency beyond its own merits.

[Torie]

"Do you know if there are any ways to calculate the number of chops?"

How do you calculate compactness and county splits now?

For this exercise, I would probably just count city chops, which show up on the DRA when you click the city button. It is a good discipline to take those lines into account, because I have learned over time that they really matter, and that at the margins giving up a little bit of partisan advantage to respect them is a good strategy.

In the highly unlikely event that you get a tie, you might either sum the total amount of population deviation from ideal, or the maximum deviation between the least populated and most populated CD as a tie breaker. My max also happened to be 1,000 for deviation, but I cut it down to 600, before submitting my map as the one change that I made, so my total is just under 1,200 for the second approach. But that I guess is all premature perhaps until we get the final census numbers into the VRA.

However, for purposes of Gingles, the HCVAP percentages are probably high enough to trigger Gingles for a second Hispanic CD in Houston, and one in the Metroplex, plus perhaps in the oil patch, although that is more problematical, since it is a Pub CD. It is a rather precarious high wire act, because to the extent TX-23 is deemed gerrymandered, and thus a candidate is elected that a majority of Hispanics there do not prefer, you have a VRA risk, even if the CD is over 50% HCVAP. Thus I was pretty careful to minimize anything that looked like a gerrymander around it.

I do know that if the Pubs draw an Austin to San Antonio CD that is intended to count as an Hispanic CD, and is drawn that way as a more effective Dem vote sink than a second nested Hispanic CD in Bexar County, that will generate an instant VRA lawsuit that has considerable merit, because that CD can in no way be considered compact and takes in two disparate Hispanic nodes far away from each other.  I doubt you can do that under the VRA when a more compact Hispanic CD is available taking in just one node. Although a closer case, you would have the same issue in the Metroplex, but there it is not possible to draw a performing Hispanic CD in Dallas County plus immediately adjacent territory. There you need to draw the road bridge to the NW corner of Ft. Worth.

When Gingles is triggered, entitling a minority to a CD, the grand unified theory of the VRA is that gerrymandering to effectively screw that minority out of being able to elect a candidate that a majority of them prefer, is a risky scheme. That is how I parse it anyway. So the best defense is to say, gerrymander, what gerrymander? There is no gerrymander that pertains to the subject real estate. You Dems are hallucinating!
 
One hypothetical example of that would be say, if TX-23 just wasn't Pub enough the way that I drew it, so instead one drew it to shed more of Bexar, or somewhat Dem areas in the Rio Grande area, and replaced the Hispanics with those in majority HCVAP Pub Ector County. Now you have a CD with two distinct and disparate Hispanic nodes that is unnecessary, and to exacerbate matters, my TX-11 ceases to be majority HCVAP. That would be the makings of a reasonably meritorious VRA lawsuit. I don't think there is much wiggle room to vary much from the way that I drew south/southwest Texas without assuming considerable VRA risk is my conclusion from all of this.

TX is the best thing ever to do another test drive of the VRA. One Pub defense, if they get cocky, is to argue that TX Hispanics are just too Pub now, and so it is time to deep six Gingles as it applies to them.

[Torie]

Would you put that blue blob on a map so I can figure out what it is?

Houston is tricky, and thus the lines that I drew (butt ugly on the Harris-Ft Bend County line), that manage to squeeze out a Pub leaning CD but no more. The requirement of a second performing Hispanic CD does reduce the efficiency of the gerrymander. But I don't see the need to give away the lean Pub CD to either comply with the VRA or avoid a dummymander. Do you have a different take, or think I did draw a dummymander in the sense that to squeeze out the lean CD, I put at risk one or more other CD's?

Thanks.

[Torie]

OK. Thanks. Opportunity districts are not required by the VRA. In my gerry, the first place I sliced into in Travis County to do a quad-chop of the county was low voting Hispanic precincts. Those are absolutely the best* to use in Pubmanders. Not legally required opportunity districts are just toxic. And drawing one would certainly reduce the score one would get from you.

*TX-07 loves them too - see above.

[Torie]

I revised my map to pump up my score in the Houston area. TX-14 was just too Pub, so Jasper County was transferred from TX-14 to TX-08, which makes the map look prettier as well. Not sure how you measure compactness, but maybe this change generates some points on that metric. It should.  So to effect the above, changes were made to TX-7, 8, 10, 14, 18 and 22. I missed a small Hispanic precinct on the border of TX-10 and TX-18, that needed to be moved to TX-18, so I took the opportunity to correct that error as well. The goal was to get TX-07 up to a 7.0% margin, while keeping all the adjacent Pub CD's at at least an 18% Trump margin, mostly by TX-08 sharing  its "excess" Trump margin a bit more "equitably,"  while avoiding any more county chops. Note that TX-08 and TX-14 now meet at both ends, entirely wrapping around the CD’s in Harris and Montgomery County.  

I think I got just about everything that was out there for TX-07 now, subject to the above constraints.


https://davesredistricting.org/join/0e1f2d77-378b-45f7-80cf-fe1208e14eb3

[Torie]

Thank you. I assume one is rewarded in your scoring system, all other things being equal, in having the same Trump plus margin in a bunch of CD's, rather than having them vary. Correct?

[Torie]

Leaderboard:

1. Torie(v2) (380 points, 26-12 map)
2. Torie (377 points, 26-12 map)
3. Thunder98 (346 points, 25-13 map)

My first map had a score change from what you posted above (quoted below). What happened? Never mind! That was the score of the map I did not draw.  

I see now that I got an extra point for compactness and two for partisanship. I am still not sure how you compute compactness. I assume I was rewarded for moving Jasper County, which is appropriate, if that was what did it.

I was interested in seeing how the score changed for each of the three prongs. Alas that is not currently possible for me. Once I see the changes of each prong, my next question if other than the partisan score, was to ask why of course.

"You attempted a 26-12 map, and your final score is 377:

40 points for compactness
9 points for county splitting
328 points for the partisanship of districts"

I might be useful in your master chart, to have the score for each prong and then the total, so the chart would have 4 columns, with each row being a given map of course.  Just a thought.

[Torie]

Here is my Pubmander TX CD contest map updated based on the 2020 census numbers.

https://davesredistricting.org/join/c6f7daa4-156b-4e7f-8d58-765965bb4976