Wednesday, April 22, 2026

A cure for Gerrymandering

I read this thread today and I think it makes a lot of sense, but I thought I'd write down my ideas on how to fix gerrymandering.  Here are some legitimate concerns I think we should consider when drawing districts.

Rule 1. All districts should have roughly the same population. To not do so is unconstitutional. 

Rule 2. Maps have to be in compliance with the VRA

Rule 3. There should be some consideration for preserving incumbents do to the value of seniority (even as this has become less valuable)

Rule 4. There should be some consideration for boundaries/compactness.

That being said, my idea is to use computers, randomization and a round-robin veto to end up on something that everyone likes and dislikes pretty evenly. Here's the basic idea:

A. Start with the existing map. The computer could choose a starting precinct for each district at random, but starting with the existing map is meant to address Rule 3. 

B. If a district has to be added, the computer chooses a random precinct to start with and makes that the new district for use in following steps. If one has to be eliminated, the computer eliminates one chosen at random

C. Remove or "purge" selected edge precincts from the most populous district and then add them to neighboring ones until it is no longer the most populous district. Then switch to the new largest district and repeat until all the district are in compliance with Rule 1.

D. Analyze the map to see if it is in compliance with Rule 2. If not, throw it out and start over at Step A. If it is, keep it as an option. 

E. If you have X options go on to Step F and, if not, go back to Sep A to start on another option. X would be equal to the number of state legislators plus 2. 

F. With X options in hand, start with the least senior member of the legislature and let them veto one. Then on to the next least senior, etc... until every one has vetoed a map and only two remain. The Governor then chooses from among the two last options. 

To address rule 3, the computer should give more weight in the random selection of precincts to purge to those that are least connected to the others. For example, if the existing District includes one entire county and 300 precincts in another, those 300 precincts would be made more likely to be purged. Same with precincts that are far from the geographic center of the District. 

Starting with the old maps means that past gerrymandering would - to some extent -carry over to the next one but with each decade this would become less and less true until one day it no longer is. 

Letting legislators and governor veto maps means that the most partisan-unfair ones are removed - because Democrats will remove the maps that most benefit Republicans and Republicans would remove the ones that most help Democrats. And the use of randomness brings with it fairness. 

Wednesday, April 8, 2026

How the parties should order their primaries

Parties choose their candidates for president with the primary/caucus system which seems reasonable. What is not reasonable is that it was not designed to pick a candidate who is most likely to win the election. If I ran a party, I'd ask the question " how can we best design the system to pick a general election winner" and this is my stab to change the order to do that. 

It's been well documented that the states that vote first in the primaries have an outsized voice in choosing the candidates for President for both parties. It therefore seems to make sense that the parties should take more control over the order they go in. I think they should reorder the primaries every election based on how close each state's vote was in the last one. And spread them out one (or more) a week instead of bunching them so that candidates can just roll into the next one. Along those lines I created a mock 25 week schedule (which would start in mid- January and end at the end of June). 

It makes no sense for South Carolina to go so early for Democrats since they're not going to win in South Carolina. Having a candidate popular in SC does the Democrats no good. Or the Republicans for that matter. 

Each week in my mock up has an "anchor" state (ordered by vote margin) and sometimes smaller states to fill it out so that. When a very large state is the anchor it is alone. The other weeks all have 10-11 million people. I threw the territories in where with their closest states were voting. 

In some cases I considered geography to try and keep neighbor states together. Here's what I got: 

1. Wisconsin, New Hampshire, Nevada 
2. Michigan 
3. Pennsylvania 
4. Georgia 
5. North Carolina 
6. Minnesota, New Mexico, Maine, Rhode Island 
7. Arizona, Iowa 
8. Virginia, Delaware 
9. New Jersey, Montana 
10. Illinois 
11. Colorado, Oregon 
12. Ohio
13. New York 
14. Florida, Puerto Rico, VI 
15. Texas 
16. South Carolina, Connecticut, South Dakota, North Dakota 
17. Washington, Idaho, Alaska 
18. Missouri, Mississippi, Vermont 
19. Indiana, Utah 
20. California 
21. Louisiana, Kentucky, Hawaii, Pacific territories 
22. Massachusetts, Arkansas 
23. Maryland, Oklahoma 
24. Tennessee, Kansas 
25. Alabama, Wyoming, DC, West Virginia, Nebraska 

By coincidence NH is still an early state, as is Nevada. Wisconsin would be a great place for the kick-off Primary.