Gerrymandering with AlphaPhoenix

You probably know that gerrymandering concerns drawing voting districts, as for state representatives, in such a way that one side gets an unfair advantage. How can you tell if it’s happening? Generally people like to point out the strange snakelike districts that are used to pack certain groups of voters together or break apart others. After all, it was a funny salamander-shaped congressional district that gave gerrymandering its name. So you might expect it to always result in spidery-lizardy districts. But look at these two maps of North Carolina.

Maps courtesy of Brian Haidet

This first one is an actual district map for North Carolina. It has some of those tell-tale salamander districts. By population, North Carolina is a red state, but not dramatically so. With 13 districts, you might expect a 7-6 split in favor of Republicans. But this map manages to squeeze out 10 red districts.

This second map was made by a physicist named Brian Haidet who took a great interest in gerrymandering. He wrote a program that can gerrymander successfully while avoiding the problem of snaky regions. This map “looks” fair, in the sense that the regions are reasonably shaped. And they all have roughly the same population. Yet he’s managed to give North Carolina Republicans 11 out of 13 representatives.

Here’s the video where he talks about the process: Algorithmic Redistricting: Elections made-to-order. It’s long, but it’s both entertaining and eye-opening. And it uses the prize-winning phrase “Markov Chain Monte Carlo Simulated Anneal.”

Here are my big takeaways from the video.

  • Cheating does not mean funny district lines. If you’re good at math and maps, you can totally cheat with nice, sane-looking districts.
  • Fairness is predicated on people’s voting preferences being stable and predictable. You must start with a baseline population and voting preference. This may be a safe assumption in these polarized times, but you still need to keep in mind that “fairness” is based on a distribution that needs to be resampled from time to time.
  • If preferences are indeed stable and predictable, then the outcome is completely known. It is purely an exercise in geometry, foreordained well before the election by an algorithm. This seems weird, but there’s no escaping it.
  • Even when you draw the districts so that they are fair, you still have a few degrees of freedom. Would you like the races to be close, or incumbent-favoring landslides?

Gerrymandering is a hot topic these days, partly because it lends itself to fun simulations that are now tractable on personal computers. So we now know what danger lurks in the redistricting process. There are plenty of good ways to limit cheating, but political interest means that whoever thinks they have the upper hand has no interest in trying to make the process fair.

In a sense, we’re lucky that the people trying to rig elections have been so ham-fisted about it. By making such obviously partisan maps, we are now at least alert to the danger. I think of this as the Photoshop Effect. Photos have been doctored since photography was first invented. But it took the widespread use of Photoshop for ordinary people to realize it. You never could trust photos, but now you KNOW you can’t trust photos. Voting districts have been doctored since Elbridge Gerry was the salamander-loving governor of Massachusetts. Thanks to the good work of people like Brian Haidet, more of us are starting to get wise to it.

Idea Gardening – from Perl to Obsidian

One of my favorite writing activities is idea gardening. I’ll think about something that might make a good blog post and make some notes. But more often than not, I’m not ready (read: I’m too lazy) to finish it. So it gets set aside, incomplete. But that leads me to another idea. I move that idea forward a little bit. But then I get distracted by Twitter.

“Idea gardening” is, of course, a charitable term. It might be less charitably called diffuse attention-deficit bullshit non-writing. I get to pretend like I’m writing, but really I’m moving distractedly from one topic to another without actually completing anything.

But years of desultory experience have taught me a few things. One is that I genuinely enjoy idea gardening, whether or not I finish anything. So, like playing the ukulele badly, I’m willing to call it a hobby and feel good about it. The other other lesson is that good tools can make a big difference. Good tools can help me cycle more quickly through my garden of ideas, pruning and weeding and watering, and occasionally harvesting. I love that moment when a topic becomes substantial enough, mature enough, that it almost seems to grow to completion on its own.

Tools and Tool Hounds

So, if tools matter, what tools should you use? I like watching what the cool kids, the tool hounds who review new tools, are playing with these days. But switching tools can be a dangerous time sink. It’s easy to convince yourself that these are the shoes that will (finally!) let you play like Michael Jordan. It’s easy to spend more time installing and polishing tools than using them. And it’s easy to mistake the honeymoon glow of mere novelty for a genuine productivity boost. After all, what Michael Jordan really does is pull on his shoes and play ball. It’s the rest of us who obsess about the shoes.

But still! Tools really can make all the difference. When it comes to idea gardening, these are the tasks that I’m asking the tool to help me with.

  • Catch ideas. Quickly create new documents
  • Survey ideas. Move quickly between documents.
  • Connect ideas. Find patterns in documents.
  • Polish ideas. Refactor and edit documents.

I could try to do all this in Microsoft Word. But what would be the fun in that?

My history of gardening tools

I’ve been at this for a long time, so some of these go back into the last century. This list is just from my personal history. Any comprehensive list would go on for pages.

  • Perl. When the web was young, I wrote tiny web-enabled perl scripts that would let me edit any text file by clicking on a link. Effectively, this was just a crude proto-wiki. Which may be why I instantly fell in love with wikis when they came along.
  • Trellix. This was a creation of Dan Bricklin, the same guy who wrote VisiCalc. It was a revelation to me that someone else valued this workflow and was willing to make a product out of it.
  • Wiki. Built to support collaboration among multiple authors, wikis also support the idea gardening of one person. I even installed a personal version of MediaWiki (the software behind the Wikipedia), but it proved too heavy for my needs.
  • Scrivener. This is a professional’s writer’s tool I learned about from journalist and tool hound James Fallows. It was fun to play with, but also more machinery than I need.
  • Simplenote. This was closer to my ideal of speed and simplicity. But maybe a little too simple. Moving between files and ideas was still a little clunky.
  • Evernote. I eventually took the big dive into Evernote. I’ve used it for a long time and I still like it a lot, but it’s on a path to excess functionality and bloat. Beyond that, I can’t think of any feature (that I care about) that Obsidian doesn’t do better and faster.
  • Notability. This is a beautiful iPad app, but it’s geared to the pen and tablet note-taking experience. It’s not so great for searching and assembling text.
  • Obsidian. For now, for me, this is the clear winner. Free, fast, text-based, extendable, it checks all the boxes.

I feel like I’ve been searching for the perfect idea-gardening tool for ages, and Obsidian is it. I’m sure it’s not the last word in idea gardening, but it’s passed a threshold that, for me, means it has fully arrived. I’m having so much fun with it. Will it actually cause me to write more? We’ll see.