Voter suppression stories tend to be seasonal here at DKos. While we all get understandably exercised about the topic during election season itself, other issues typically command most of our attention for the other three quarters of each election cycle.
Eternal Vigilance
So why is the progressive data science organization I lead, Insightus, rolling out our most ambitious anti-voter-suppression project ever, in an off-year? Because here in the purple battleground state of North Carolina the Republican vote-suppressors who control the legislature never sleep.
A few quick stats will illustrate why their nights — and thus ours — are so restless:
- Black voters comprise almost a quarter of the state’s electorate, and almost half of the state’s Democrats
- Black Democrats outnumber black Republicans by 30-to-1
- Registered Democrats (of all races) outnumber Republicans by 650,000
In short: if NC’s Republicans wish to stay in power, their efforts to suppress the vote — and voters of color in particular — must continue 24 / 7 / 365. And thus, so too must our resistance. Fighting for voting rights isn’t just an election-year battle.
Woodhouse rules: In election administration, process can determine outcome
Unfortunately for NC Republicans, the U.S. Fourth Circuit Court of Appeals is among the nation’s more voter-friendly courts, routinely slapping down the state’s unconstitutional voter suppression laws almost as fast as legislators can pass them into law. But bad laws aren’t the only way to keep disfavored voters away from the polls. Voter suppression can be — and too often is — baked into election administration itself, managed by largely faceless bureaucrats, where its subtle and complex moving parts can be all but invisible to the prying eyes of judges, reporters, and the public.
Consider, for instance, the 2016 call by NC Republican Party executive director, Dallas Woodhouse, for Republican members of county Boards of Elections to “make party line changes to early voting” (i.e., to keep Early Voting polling places away from concentrations of core Democratic constituencies such as voters of color, 70% of whom voted early in 2012). Perhaps not coincidentally, in 2016 black voters eked out a mere 91% of their 2012 Early Voting turnout, while white Early Voting turnout was an impressive 118% of its 2012 rate — numbers that Woodhouse now wears like a badge of honor at Republican convocations.
EVE 1.0: The past is prologue
While Insightus has championed the need for fair election administration since our founding, it’s fair to say that so far we’ve been fighting the last war over and over again, shining light on outrages that have already happened rather than proactively resisting the next transgression. That ends now, with this week’s roll-out of our most ambitious project ever: EVE (North Carolina’s Early Voting Evaluator).
EVE is a sophisticated browser-based app (don’t even try to access it on a mobile device - it’s too graphics- and memory-intensive) that provides in-depth statistical analyses and powerful but intuitive data visualization of NC’s early voting plans at the click of a mouse. We developed it in reaction to the wave of requests we received last September from civil rights attorneys seeking our help to analyze the state’s voting plans in support of their court challenges. We were glad to help, but our volunteers were overwhelmed by the rushed nature of the required work. We promised ourselves back then, during numerous all-nighters, that after the election we’d develop a system that would enable voting rights activists, civil rights groups, reporters and other stakeholders to roll their own analyses quickly and easily, without our help.
Version 1.0 of EVE (which went live last week on the web) provides a retrospective analysis of 2016’s voting plan, and compares that with the ‘statutory minimum plan’ that will befall the state’s voters if the Board of Elections deadlocks in its Early Voting planning for 2018, as Republicans are currently working to ensure. Version 2.0, which will appear later this month, will extend that data set to include the state’s voting plans for 2010, 2012, and 2014 as well, providing insight into the state’s longer-term historical Early Voting trends.
what if?
But the real payoff will come with EVE 3.0 (arriving sometime later this year). Version 3.0 will include functions enabling users to perform their own ‘what-if’ analyses of proposed Early Voting plans before they’re adopted (“what if they move these five polling places from here to there, close these two, and extend the hours of these other eight sites? Would that be more equitable than 2016’s plan, or would it make things even worse?”). Today, the answer to that sort of question is unattainable unless you have weeks of time and an able statistician or two to devote to working out the answer. And with 100 counties each debating two or three different plans, you’d have to perform that kind of analysis hundreds of times over to understand North Carolina’s situation overall.
Our goal for EVE 3.0 is to enable the well-meaning members of the state’s 100 county boards of elections to make better data-driven decisions regarding the proposed Early Voting plans they’re about to vote for or against. In other words, we're building EVE to proactively fight for better voting plans in future by arming decision-makers with better insights. Of course, EVE can equally well be used by the bad guys to design even worse plans. But by making the power of better data visualization available to all, EVE ensures that those bad plans can no longer hide in mind-numbing lists of polling place addresses and their hours of operation. With EVE, we’re betting that transparency and insight are, on balance, net benefits for the good guys.
EVE has been designed since day one with that 3.0 goal in mind. And that (the goal of supporting what-if analyses) made EVE’s development an interesting technical challenge. Real-time what-if analysis precludes the typical simplifying solution of serving up pre-computed results (because we don’t know in advance what plan details you’ll ask it to analyze). But we needed to avoid the obvious alternative (real-time computation on the server side) because that’s both expensive — at least for an organization whose checking account runs on fumes — and it can be a support nightmare for an all-volunteer organization. So EVE is designed to perform all of its computations on raw data in real time in the user’s browser (it is entirely written in Javascript). And it needs to be able to do things like calculate distance-to-poll for all seven million NC voters in real time. So not only was it a coding challenge, it was an algorithm development challenge as well. But our volunteer data scientists found ways to make it all work. When you visit EVE there’s a lot of sophisticated computation going on under the hood, right in your browser. But hopefully you’ll never notice any of that.
Feel free to take EVE for a test drive. You’ll find EVE 1.0 here on the Insightus web site, where you’ll also find links to an occasional series of ‘Explainer’ articles — basic tutorials on the nuts and bolts of how Early Voting works (or doesn’t work) in NC. Again, remember: don’t even try EVE on a mobile device (nothing bad will happen, but it sure won’t be pretty). A mobile-friendly version may be in the cards for sometime in the future.
Judging from our server logs, the State Board of Elections and officials of both major state parties have already taken notice of EVE. That’s great.
We’re very interested in any and all user feedback. Post your glowing praise below, or KosMail me your scathing criticisms and bug reports ;-) If the latter, please mention the brand and version number of your browser, since many technical hiccups tend to be browser-specific.
Forward together, y’all. Fight like heaven.
Follow @InsightusOrg on Twitter for EVE and other Insightus project news and updates.