This of course throws a wrench in works for candidates and the organizations that support them. It’s hard to figure out if and how you are going to run for office if you don’t know where you are running for office!
Deck has had to adapt to redistricting as well. Deck is here to help candidates understand their new districts. But to do so, we need to solve a data problem: how do you figure out which district every person in the US is in? We want to do this quickly so candidates can get up to date information on their district before the voter file updates. At the same time, we also want to make sure new districts are accurate and have all the correct targets.
In this blog post I’m going to talk a bit about how Deck deals with this problem especially because so much of the process of redistricting is so opaque. I’ll talk about the issues we’ve run into and what that tells us about what redistricting means for politics this cycle.
Redistricting is complicated, but I want you to understand how we make sure each district in the United States has updated targets. When I look at the problem of assigning each person to a district, it becomes three distinct problems: where is everyone after the last census, where are the districts, and how do we figure that out across all 50 states?
First, where are people after the last census? Once every 10 years we take a census of everyone in the United States (or at least attempt to) and assign each person to what is called a census block. A census block is the smallest area of geography the census uses – it’s usually a few streets. In the United States, there are 6.9 million census blocks, 2 million of which have zero population.
But we encounter a problem here. Until now Deck has used the 2010 census blocks to build our districts, models, visualizations, and more. Because states redistrict one at a time, we need to store spatial data about both 2010 and 2020 blocks because 2022 districts are built from 2020 census blocks, not 2010 ones. Now we have to figure out what became of the 2010 census blocks and how they relate to our new census blocks. We relate the new districts to census blocks from both 2010 and 2020 so we can easily join to datasets using either set of blocks. This is an important difference in redistricting data: it’s a physical space so we have to use different tools to match and clean the data.
Now that we know who is in each census block in 2010, 2020, and how those two groups relate, it’s time to start drawing those districts! To draw a district we load files called shapefiles that tell us where in space these districts are as well as block crosswalk files that tell us which 2020 blocks make up which new districts. Our database server, called BigQuery, allows Deck to store geometry data, or a collection of points that outline an area, in space in a single row. This way we can e match these shape descriptions to the census blocks they include and the people who live in each census block. For example, a block crosswalk file will tell us a district contains census blocks 1 – 14. We know who lives in each census block and where those blocks are. After loading a district file we then know who lives in each district. This also now means that we can create voter targets and give campaigns lots of helpful information about their districts.
But to create these districts, we first need to get these files for all 50 states as they redistrict. Deck supports campaigns nationally, and each state redistricts at different times using different processes and those maps end up in different places. To get a hold of these districts we contact people at state parties, comb through secretaries of state’s websites, but most frequently find them publicly available at independent individually run sources like All About Redistricting or Dave’s Redistricting.
When I first talked to our data engineer Amanda about this, I was surprised that Deck found so much of its data from sources outside of state governments. But when I went looking through the Michigan redistricting commission website for the finished district maps, I understood why. Although I eventually found the maps, keeping track of who has redistricted, where they are in the process, and then scouring the web to find that data across all 50 states is an arduous task. That’s why we at Deck lean heavily on projects like Dave’s Redistricting, 538, and All About Redistricting to tell us when states have redistricted and to compile their new maps. As a national political data company, we get data from lots of public and private sources and we work to identify the best data for the job. But sometimes the best source is independent resources. In 2021, we used a public spreadsheet made by an individual to get the most up-to-date information on who was running for local races in New York City. That’s because keeping track of where districts are and who is running in them is a monumental task.
Having these individually run resources for redistricting is a huge benefit to people who do political work and ordinary citizens alike. Redistricting’s various legal and logistical loopholes makes it hard to follow as a regular person and these resources make that information available. But this process has huge implications for campaigns and people’s lives and although it’s exciting that people provide these resources, there is no guarantee they will stick around for the next cycle.
Working in political data means Deck has to navigate politics of all 50 states to navigate data. It is awesome that people continue to provide these centralized resources especially as court challenges to redistricting continue to make the existing maps volatile. But in an environment where data is so valuable these public interest services become scarce.
Deck has always worked to make data visible and understandable to campaigns and organizations. But like everyone else we rely on these sometimes circuitous ways to get a hold of data and we have someone whose job it is to find it! That’s why we work so hard to provide resources to all sorts of campaigns from local to federal. Redistricting has very real implications for campaigns and people’s lives, so we make everyone’s new district available as soon as the district data is public with some basic information like who your supporters are and a breakdown by demographics. We also take into account old districts in making your new district’s predictions. And you can sign up now for $50 if you are running in a primary or free for a general election to see what your new district looks like and explore your voters.