Who's behind Tessera
Hey, I'm Ryan.
I'm a Data Science Fellow at VoteHub, where I make racecalls and work on our election models. Generally, I'm just a political data nerd: I built our Voter Power Index, and on my own I used Google DeepMind's AlphaEarth satellite embeddings to estimate how a precinct votes by looking at it from orbit. I care a lot about onion futures. There's more on my website.
I built Tessera because it seemed like a fun thing to make: redistricting is a puzzle I care about, the existing tools were slow, and I wanted to see how fast and how good a Mac app for it could get. I live in Evanston, Illinois. If you want to reach me, email me.
Under the hood
Tessera in brief
Draw and explore
A native Mac app with live district maps, 3D analysis, gerrymander experiments, reports and posters. Public maps cover every state, with 2024 presidential results where available, checked against certified totals.
Compare with neutral maps
An exact sequential Monte Carlo sampler in Rust, public as obk-redist, written by Claude with me directing and reviewing the code. Its Metal GPU port runs about 100 times faster than redist in our benchmarks.
Find a better plan
Mosaic's annealed ReCom search, rewritten in Rust with the same scoring formulas. The optimizer runs 2.4 to 2.7 times faster in our tests on an M5 MacBook Air.
Credits
Tessera builds on Mosaic by Matt Mohn (MIT license), ReCom from the MGGG Redistricting Lab, and sequential Monte Carlo redistricting from Harvard's ALARM Project.
Map data comes from the U.S. Census Bureau (2020 Census, TIGER/Line), the Voting and Election Science Team (CC BY 4.0), the ALARM Project (CC BY-SA 4.0), and Dave's Redistricting (CC BY-SA 4.0). Where used, 2024 results from the Redistricting Data Hub and the New York Times retain their non-commercial terms.