Arizona takes first step to return redistricting process to Republican control

[Torie]

Well, there are still some legal strictures on how the lines can be drawn, presumably. However, even a non gerrymandered map holds the Dems to 2 seats, or 3 if one deliberately draws a second Dem seat in Phoenix. I really don't think the last commission followed the law, because beyond cooking the partisan baseline figures, they put having competitive districts at the top of the list, rather than, as the law provides, at the bottom, as a tie breaker, and Mathis explicitly stated she was going for competitive districts. I am surprised the Pubs were not more careful in building a record of that, and suing based on those grounds.

[Torie]

That's a fair point, but if they are looking to make the commission better, then they are doing it completely, 100% ass-backwards. It's pretty obvious they are just trying to return redistricting back to their control, which basically means going from maps that favored Democrats one decade to Republican-favored maps. I'd think that at least with the existing commission, the chances of getting a fair, competitive map are much better. I can't see anything good coming from this new proposal.

No, it is a joke, unless the parameters have judicially enforceable teeth, but the point is that the Pubs can take it all, and get their way, without a gerrymander, and really get nothing more, with a gerrymander (other than to make the one seat that would otherwise be lean Pub in Phoenix, safe Pub). In fact, AZ is probably the state with the most CD seats in the nation, where that is the case.

[Torie]

No, it is a joke, unless the parameters have judicially enforceable teeth, but the point is that the Pubs can take it all, and get their way, without a gerrymander, and really get nothing more, with a gerrymander (other than to make the one seat that would otherwise be lean Pub in Phoenix, safe Pub). In fact, AZ is probably the state with the most CD seats in the nation, where that is the case.

How about for legislative districts? Wouldn't their proposal still help a lot more with those than Congressional districts?

Perhaps, but I have not played with legislative districts. In theory, however, as a general rule, the small the district, the harder it is to gerrymander.

[Torie]

No, it is a joke, unless the parameters have judicially enforceable teeth, but the point is that the Pubs can take it all, and get their way, without a gerrymander, and really get nothing more, with a gerrymander (other than to make the one seat that would otherwise be lean Pub in Phoenix, safe Pub). In fact, AZ is probably the state with the most CD seats in the nation, where that is the case.

How about for legislative districts? Wouldn't their proposal still help a lot more with those than Congressional districts?

Perhaps, but I have not played with legislative districts. In theory, however, as a general rule, the small the district, the harder it is to gerrymander.

With compactness/jurisdiction splitting rules, I strongly disagree.  Look at FL House vs. FL Senate on FDF compliant maps.  Giant chamber+polite boundary rules = huge advantage for rural/exurban party.  Smaller chambers with compactness rules means both rural seats and urban seats have to penetrate further into the suburbs, keeping more districts competitive.  AZ senate (Dems got to 13/17 after 2012) vs. house is itself an example of this, as is CO senate vs. house where an officially neutral party installed soft D maps and the house is safe while the senate flipped in a wave.  Also contrast VA house, which is ironclad vs. VA senate which is not.  I'm sorry, but you're wrong on this one.      

Muon2 agrees with me. So there!    I am not saying smaller districts cannot be gerrymandered. It is just that if there is a sharp partisan division, one can with large districts, take bits out of the hostile zone, and make the opposition end up with nothing, or pack them into one districts, and leave the rest all safe for the other party, even if the partisan balance is only modestly in favor of the other party. Alternative, putting aside the VRA, a heavily Dem urban line could be chopped up in a way, that tips a bunch of districts outside it to the Dems, like  a color wheel, if one got the balance right. In Ohio, with some work, perhaps the Pubs could be held to 3 districts for example. With smaller districts, some of the seats need to be given away, and it is only no the edges of the two zones, that line drawing makes that much difference.

But no doubt Muon2 can be more eloquent on this than I. It is just that when he made the claim, it certainly made intuitive sense to me.

[Torie]

One way to look at the gerrymandering size argument is to consider the following points.

Imagine a state with as many districts as voters. Clearly each district is one voter and it is impossible to gerrymander.

Now consider a somewhat larger district, say with an average of 4000 people per district. That corresponds to the optimum size of a census tract. Using contiguous census blocks to build a district, it's hard to get very far away from a census tract and its nearest neighbors. To the extent that the state isn't filled with too many sharp political/demographic distinctions between adjacent census tracts, it is hard to attach a given area block-by-block far away from its start.

As districts get larger, one can reach across greater distances to group people intentionally to pack or dilute their voting strength. Of course there is a point of diminishing return, since once a district gets too large the gerrymanderer is forced to add populations that take away from the goal. In the limit of a single statewide district one can't gerrymander at all, just like the case of single voter districts.

As an example consider the IL delegations gerrymandered for the 2012 election. Obama got 58.6% of the two party vote in IL in 2012. The IL House (108K/district) was 71/118 for the Dems or 60.2%. The IL Senate (217K/district) was 40/59 for the Dems or 67.8%. Congress was 12/18 for the Dems or 66.7%, though it was drawn to have D PVI's of 13/18 or 72.2%. I might speculate that the Congressional districts were at the point of diminishing returns for Dem gerrymandering, though that may be more due to the limiting effect of the VRA, and the creation of 4 minority CDs.

The ultimate gerrymander in Illinois would be to have each CD have the same PVI in Illinois. It might be possible to get close to that (maybe let the Pubs have one CD in Southern Illinois), if the map is hideous enough (have snake districts running along highways from Chicago to the hinterlands). But yes, the VRA stops that in its tracks.

[Torie]

That's why it is just so wrong to trust human beings, rather than computers, in this endeavor. Just bring the humans back at the end to cross a few T's and dot a few "i's," so they can entertain the illusion that they were somehow significantly involved in the process. One thing I have learned, is just how hard it is to be truly non partisan about these things. Your biases tend to cause you to see what you want to see, and not see, what you don't want to see. So we need 00011100101111101 to do it.

I saw a movie about Alan Turing last night. We should call the model redistricting code the Turing Model Redistricting Code in his honor.

[Torie]

I would rather trust the public to draw the maps, crowdsourcing the process. Sure, a computer could draw one, too, but I'd give the public a chance to do better. Let a computer then be the judge, eliminating plans that are gerrymandered by focusing on neutral criteria. Humans can help to verify legal issues in submissions, like compliance with the VRA. At that point I'd trust a body representative of the public at large to select from a few plans that survive the computer's winnowing.

Yeah, yeah, yeah, but it's all window dressing and you know it. And over time, the computer programming will get better and better. The selection from the few plans, is where the T crossing and I dotting comes into play. I just like to point out when the emperor has no clothes. I have been that way all my life.

And don't you admit, as we do this more, and get more skilled, that it really is possible for us to find the best map, with a pretty high degree of certainty? It just takes us a lot more time, than it would take a computer, is all.

[Torie]

Well obviously one needs to agree on a set of parameters, and that is not easy, and may never be accomplished, and maybe in the end, different states will pick from a menu of options (but if it varies too much, then you have the vexatious problem of one state fearing unilateral disarmament vis a vis some other states, which is not good at all).

But if a set of parameters can be agreed upon in a model code, based on your system, it will be possible to rank maps in the way that we have discussed, with some potentially a tie, but not many I would think, given the odds are low that two different maps, will have the same chop and erosity scores, and then the same tie breaker scores or whatever, that will vary much at all except perhaps in very minor detail. But I understand your desire to make it seem like the emperor appears in sartorial splendor.

[Torie]

I think you might be repeating yourself there by and large. So let us step back. For each number of chops, the odds are high that there is but one map with the lowest erosity, and thus for each number of chops, there is but one map that reaches the pareto optimal frontier. Do you agree with that?

[Torie]

I think you might be repeating yourself there by and large. So let us step back. For each number of chops, the odds are high that there is but one map with the lowest erosity, and thus for each number of chops, there is but one map that reaches the pareto optimal frontier. Do you agree with that?

No.  It generally takes criteria beyond erosity to get only one map for a given chop score.

You think there are a bunch of maps that would tie on both?

There is presumably a set of boxes on a graph of chop vs erosity that represents the Pareto frontier. It is quite possible that more than one plan has a the same chop and erosity score to get into a box. Inequality, skew, or polarization can be used to break that tie to reduce the number of plans in that box.

Yes, indeed. And I have a longer list of tie breakers.

Alternatively some of those other metrics can be added to the chop score to spread out the plans into more boxes and make it less likely to have multiple plans in a box. Adding a metric to the chop score makes it more important in excluding plans than just using it as a tie breaker as we saw with the UCC scores.


Don't quite follow that. Remind me please "what we saw."  I can see how adding a metric, might cause a switch in which map makes it to the frontier, but I don't quite see what "makes it more important."

Anyway, in the end, even assuming there will tend to be quite frequently more than one map with the same chop and erosity score, if you add metrics, or have tie breakers, you still get down to one map, and so in the end, you just have a few maps to pick from, as you add chops in exchange for less erosity, with the number depending how high up  in chops that you go. So the menu for humans to pick from, will be limited, and very tightly constrained, which is the whole point of course.

[Torie]

OK, we are on the same page then, if not necessarily on just what maps get culled at the end, and how. It's been fun. 1000111001011001. Yes, I know, I still owe you comments elsewhere.

[Torie]

I assuming population equality within the 1% parameter for CD's.

[Torie]

I feel like there is going to need to be a push via referendum to reign in the legislature on this.

The Pub plot needs to be passed by the voters in a ballot initiative, since it changes the AZ Constitution. So the law being discussed merely gets the issue on the ballot.