Redistricting Competition?

[OBD]

Here's a fun idea that I just had. As we're now 6 days out from the presidential election, what if we had some kind of gerrymandering contest?

Here's the premise - we all draw maps of a certain state (preferably one with 10+ districts and one that's competitive, e.g. Georgia), then attempt to draw as many Biden or Trump districts as possible. When the 2020 election data is uploaded to DRA, we can determine who drew the best gerrymander in terms of districts won, with bonus points for things like compactness and security.

Thoughts?

[OBD]

This seems fun, also we should clarify whether or not we're following the VRA, because that changes things a lot. (count this as yes, I'm participating)

The VRA is not required, as it's very subject to interpretation. That said, maybe we make minority representation a tiebreaking factor in addition to the other DRA metrics.

What states do people want to do, and do people here want to maximize Biden or Trump districts? As this should be done before Election Day for maximum fairness (and all map links and images should be posted here before then), please comment quickly!

I think one of the following would be a good choice, for competitiveness and availability of 2016 data. We could also do all of them and have winners for each! Also, 2010 district numbers and populations will be used.

- Georgia
- Florida
- Texas
- Arizona
- North Carolina
- Michigan

[OBD]

I could try. I wouldn’t do too well but could be fun

Great!

Also, if people could say whether they prefer to max out Biden or Trump seats as well, that would be great. I personally vote for maximizing Biden seats.

[OBD]

Let's make it a bit more interesting, maximize seats opposite of how the state voted in 2016 and let's keep some county split rules so its just not all baconmanders and you actually have to figure out how to do it. Baconmandering would just be guessing the final margin of the state and drawing a sink or 2 and bacon mandering the rest

Opposite of 2016 is a fair idea - I could definitely back that if a majority doesn't want to just maximize Biden districts.

I think baconmandering is OK - my idea was to use DRA ratings as tiebreakers, so that would mitigate that factor.

[OBD]

Upon consulting with lfromnj, I've created some guidelines for this competition:

Rules
1. Districts will be drawn using 2010 Census numbers. This means that the number of districts drawn will be the current amount (i.e. the 2010 numbers - so 36-district Texas)
2. Districts will have a maximum population deviation of 1500 each. (some leeway may be given for urban districts with large precinct sizes)
3. All districts must be contiguous (this rule might be edited/given some leeway if we choose a state with messy precinct boundaries - like Ohio). Additionally, all districts must be contiguous by land or bridge (no abusing water precincts in coastal states). Road contiguity is encouraged but not required.
4. The DRA compactness score for the map must be over 30.
5. The DRA county splitting score for the map must be over 30.
6. The winner will be whoever draws the most Biden/Trump districts for their candidate. In the event of a tie, the DRA compactness and county splitting scores of each map will be averaged and used to determine the winner. If this is still tied, whichever map has more favorable partisan metrics for the candidate of choice will win.
7. All entries must be submitted by the state's poll closing time on November 3. Additionally, an image must be posted (from an image hosting site like Imgur or imgbb) to ensure that no cheating occurs. Once the state's poll closing time passes, the post containing the link to the map and the image cannot be edited (obviously, you can't edit the map on DRA either).
8. Once 2020 data is released on DRA, the winner(s) shall be determined.

EDITED to add for bridge contiguity.

Does this sound good to everyone?

[OBD]

Idk about baconmandering but I do think we should have some limits to make this more interesting. I support VRA requirements but you set the boundaries. For example, if we did Georgia, you would choose whether 4 or 5 AA VRA seats are required and we all work around that.

lfromnj and I agree right now that enforcing the VRA is highly subjective - plus, being effectively forced to make packs makes drawing districts way less fun. In Georgia, what you propose is probably feasible, but in Texas, it would be a nightmare to do. So, right now I think the best move is to restrict compactness and county splitting metrics only - which can be analyzed fairly unbiased by the DRA program.

[OBD]

Also would propose California(get as many expected Trump districts),

Texas as said has a large range

Florida would be the last one, one might have to play a bit safe regarding Cubans.

California is really too big. Even Texas pushes the limit - remember this needs to be done by next Tuesday.

That said, it is possible that we have multiple states, with some being optional - though we should have one consensus state that we all draw for.

[OBD]

Also would propose California(get as many expected Trump districts),

Texas as said has a large range

Florida would be the last one, one might have to play a bit safe regarding Cubans.

California is really too big. Even Texas pushes the limit - remember this needs to be done by next Tuesday.

That said, it is possible that we have multiple states, with some being optional - though we should have one consensus state that we all draw for.

Georgia and NC are good consensus picks.

Agreed. Judging by the current ballots, I might pick NC and TX as the two consensus picks, making everything else an optional bonus. I'm gonna wait till tomorrow or Friday to make that call, though.

[OBD]

I'm down.

Any prizes?

Great Honers

The adulation and respect of the ~15 people that have read this thread :/

[OBD]

When does this start?

I'll make a more official announcement regarding the states involved tomorrow, but for now feel free to start working on Texas and North Carolina maps that (you believe) maximize Biden districts. (seems to be the consensus rn)

[OBD]

Hang on, clarification needed, are using the 2010 apportionment, or the forecasted apportionment after 2020?

2010 appointment.

[OBD]

Okay, so it's now 4 days before the election so probably best to get this show on the road. Based on the poll results, we'll be maximizing Biden districts in North Carolina and Texas! All competitors with the willpower to do so are highly encouraged to make maps of these states. You can also submit additional maps, which will be considered for 'prizes' if more than one person submits maps for that state. I will be the referee - I will submit maps, but they will not be considered for prizes. DM me on Atlas with a link to your map by 4:00 PM PST on Tuesday, and after Election Night feel free to post your map here with #analysis and an image.

Anyway, here are the rules:

Rules
1. Districts will be drawn using 2010 Census numbers. This means that the number of districts drawn will be the current amount (i.e. the 2010 numbers - so 36-district Texas)
2. Districts will have a maximum population deviation of 3500 each. (some leeway may be given for urban districts with large precinct sizes)
3. All districts must be contiguous (this rule might be edited/given some leeway if we choose a state with messy precinct boundaries - like Ohio). Additionally, all districts must be contiguous by land or bridge (no abusing water precincts in coastal states). Road contiguity is encouraged but not required.
4. The DRA compactness score for the map must be over 30.
5. The DRA county splitting score for the map must be over 30.
6. The winner will be whoever draws the most Biden/Trump districts for their candidate. In the event of a tie, the DRA compactness and county splitting scores of each map will be averaged and used to determine the winner. If this is still tied, whichever map has more favorable partisan metrics for the candidate of choice will win.
7. All entries must be submitted to me by the state's poll closing time on November 3. Additionally, an image and description of your map should be posted on this thread, but after Election Night to avoid plagiarism.
8. Once 2020 data is released on DRA, the winner(s) shall be determined.

And these are the 'prizes':

Tom Delay Cup: Awarded to whoever draws the most Biden districts in Texas, while adhering to the above rules.

Bill Barr Cup: Awarded to whoever draws the most Biden districts in North Carolina, while adhering to the above rules.

Elbridge Gerry Cup: Special Contest! Whoever draws the most 2020 Trumpy district in Massachusetts (only rule is contiguity by land or bridge - no compactness, VRA, or splitting requirements, remainder of map required though) with 2010 populations wins!

Crack n'Pack Cup: Whoever wins the most contests in other states (FYI, maxxing Biden districts).

Does this sound good to everyone?

[OBD]

Announcing an edit: population deviation of up to 3,500 is now allowed.

[OBD]

Also, if y'all could title the subject of your PMs accordingly, I'd appreciate it a ton! I'm gonna have to wade through a lot on Monday and Tuesday. Btw, I'm probably going to start copying the links I have on Tuesday around 9-11 AM PST, so if you've sent me a link make your edits before then.

Do make as many maps as you can, though. No need to formally enter on this thread or anything.

[OBD]

Reminder that submissions are due late Tuesday in my inbox! Please send to me before then though if possible, as I'm going to start uploading maps early Tuesday.

[OBD]

Here are the submissions I have received and copied thus far:

Texas
- S019
- ProgressiveModerate
- Forumlurker

NC
- Forumlurker
- Blairite
- lfromnj
- tack50
- S019

Massachusetts
- TimTurner

As we only have one MA submission, I'd request any further entrants to submit an NC or TX map. I'm not able to copy maps tonight, but I'll accept any maps that land in my inbox by 4:30 PM PST (North Carolina poll closing time). It's on your honor not to edit it after that time - I won't catch you if you do so, but the all-seeing god OC will.

I haven't reviewed the maps thus far, but if I see a minor contiguity issue or population equality issue I'll probably give the offending person one chance to fix it without altering the map in any other way.

[OBD]

Well this will be interesting.

I'll have results for this when accurate 2020 numbers are up on DRA.

[OBD]

I...actually feel optimistic about NC. I went for 11-3 and I think it'll hold, although what I did with the Triad might make things close.

Hmm - well with the pro-Biden swing, you should be fine.

I won't consider my map for the contest, but I went for an 11-2 (think you meant 10-3) and I think it could hold. I think lfromnj did that as well, so we'll see what happens.

[OBD]

Hmm, I'm hearing reports that precinct-level results will be messed up by VBM, so it's very possible that this competition will not have a solid conclusion (in addition to DRA being permanently clucked for the next 4 years). Yikes.

[OBD]

Do y'all remember this?

As NC 2020 results are now out, I'll try to compute winners for that part of the competition.

[OBD]

Alright, folks, here's the results for North Carolina!

District Results
Forumlurker: Biden 9, Trump 4
Blairite: Biden 9, Trump 4
Tack50: Biden 9, Trump 4
Lfromnj: Biden 8, Trump 5
S019: Biden 8, Trump 5

This is pretty clear cut - there's a three-way tie with 9 Biden districts, Lfromnj and S019 are eliminated. Additionally, both of their maps had contiguity issues on the coast (I did say districts had to be contiguous by land or bridge), so they would have been eliminated anyway.

Contiguity, Compactness, and County Splitting
Forumlurker: Contiguous, 30 (Compactness 30, County Splitting 30)
Blairite: Contiguous, 29.5 (Compactness 29, County Splitting 30)
Tack50: Contiguous, 29.5 (Compactness 31, County Splitting 28)

A note on this section - Blairite and Tack50 technically broke the Compactness/County Splitting rule, but under further investigation they actually satisfied the requirements on the 2010's precincts. Definitely not fair to disqualify them for that here.

So, I went back to the original maps (2010 precincts) to average the scores, and these were the results. Congrats, Tack!

Tack50 - 31.5
Forumlurker - 31
Blairite - 31

If this was tied as well, the competition would have gone to 'partisan metrics' - which in retrospect was a pretty ill-defined term. If this was the case, I likely would have gone off which map had the closest Republican win (and thus was closest to having 10 Biden seats). This map is Forumlurker's - his 9th District was Trump+0.8, narrowly edging Blairite's Trump+0.9 8th District. Tack's 1st District, unfortunately, was Trump+2.1, which would have put him in third here.

[OBD]

Can you post all of our maps?

Your names are linked to your maps, I believe.