Why I created VoteValue
I'll be honest about where this started: it wasn't a movement. It was a research question I couldn't shake.
In the summer before my junior year, I was doing research at George Mason University's ASSIP program on the mathematical structure of gerrymandering. The model I worked on, called SEIR-GA, treated voter engagement as a coupled system of ordinary differential equations, the same kind of math epidemiologists use to study how diseases spread through a population. The idea was to see whether gerrymandering produces measurable, predictable depression in voter turnout the same way certain conditions produce predictable epidemic curves.
The math was interesting. What surprised me more was the obvious thing the math kept pointing at: in most American congressional districts, the outcome of the general election is decided well before any ballot is cast. Not by anyone in particular. By geometry. By turnout history. By the cumulative weight of decisions made in a state capitol years ago by people whose names most voters in the district have never heard.
That's not a controversial finding. It's been part of political science for decades. What I couldn't get past was that this picture is invisible from inside the voting booth.
The gap between what we measure and what voters see
There are a lot of good civic tools out there. Brennan Center publishes serious research on redistricting. Princeton's Gerrymandering Project tracks state-level fairness scores. MIT and Tufts host Districtr, which lets researchers and advocates draw alternative maps. Most state secretaries of state have polling place lookups. The League of Women Voters has plain-language voter guides.
What I couldn't find was a single tool that did this:
- Take any U.S. address as an input.
- Look up the dozens of legislative districts that overlap that address, federal and state.
- Apply a consistent, transparent measure of how much a single vote in each of those districts actually weighs.
- Show the answer to the voter who lives there, in language that doesn't require a political science degree.
The pieces all existed. They just didn't talk to each other, and the result wasn't designed for the person whose vote was the unit of measurement.
So I started building VoteValue.
What it has to be, and what it cannot be
A tool like this has two failure modes that scared me from the beginning.
The first is partisanship. The moment a tool that measures voting power can be read as taking sides, it loses the audience it needs most. A red voter in a blue district and a blue voter in a red district have to be able to use the same site and get the same kind of honest answer. That's why VoteValue is nonpartisan by design, not by performance. There are no recommendations on the site. No "this is what your representative should do." No candidate scorecards. Just data, methodology, and the user's own judgment about what to do with it.
The second is opacity. If I tell a voter their VoteValue Index is 41 out of 100, and I can't tell them precisely what went into that number, I haven't given them information. I've given them an opinion dressed in math. The whole project is staked on showing the work. Every component of every score is documented, every data source is named, every formula is published, and where the model is uncertain, the site says so.
I'd rather have a smaller tool that voters can trust than a slicker tool they can't audit.
What the project looks like in practice
Today, VoteValue covers all 50 states and the District of Columbia, with over 163,000 precincts in its database. A voter types in an address and the site returns:
- A VoteValue Index score for every legislative district that represents that address. Federal House. State Senate. State House. U.S. Senate at the statewide level.
- The five factors that go into each VVI: competitiveness, mobilization potential, electoral leverage, district integrity, race significance. Each one is broken out with its own subscore so you can see what's driving the number.
- The names and contact information of the people who currently hold those seats.
- An interactive map showing district boundaries at every level, with color modes for partisan lean, competitiveness, and compactness.
- A community-driven Marketplace where users can publish proposed redistricting maps and gather petition signatures that auto-generate constituent letters to representatives.
- A short game, the Gerrymander Guesser, that turns the question of "which district looks more gerrymandered" into a community-derived ELO rating per district, complementing the pure geometric compactness measure.
I built the first version of this alone. The site is built on Flask, Leaflet, and a Postgres database hosted on AWS. The data comes from public sources: U.S. Census ACS, MIT Election Data and Science Lab, official precinct shapefiles, and the VEST 2020 dataset.
Where it's going
The Marketplace is the long-term bet. Most civic platforms are content-out: experts publish, users consume. I'd rather have the inverse. Anyone should be able to draw an alternative district map for their state, attach a written rationale, and put it in front of other voters and, eventually, the people who actually draw the lines. The community-derived gerrymander ranking from the Guesser feeds into the same loop: it's the public's eye for shapes, formalized.
I'm under no illusion that a website fixes a problem this old. But I do think that more honest information, distributed evenly, makes the rest of the work easier. Voters with a clearer picture of the system they're inside are harder to talk past. Organizers with a fairer measurement of where the slack is have better targets. Legislators who know their constituents have a number for the structural shape of their seat have a slightly different conversation about what to do next.
That's the thesis. The data is the test.
If you've used the site and have feedback, or you want to write a guest post on something in your state, please reach out at votevalueteam@gmail.com.
Thanks for reading.