[Image via MIT]

MIT’s Election Data and Science Lab (MEDSL) recently hosted an Election Audit Summit featuring presentations and discussion on the rapidly-growing practice of post-election audits nationwide. Fortunately for those of us unable to attend, MIT’s Claire De Soi has a wrap-up in the latest electionlineWeekly:

Last week, over a hundred engineers, social scientists, legal scholars, election officials, and others invested in improving U.S. elections braved Boston’s dropping temperatures to participate in theElection Audit Summit. A project of the Caltech-MIT Voting Technology Project, the summit was supported by MIT’s Election Data & Science Lab.

The two-day conference was designed to explore the intersections of the scientific, policy, and legal issues related to post-election auditing. As concerns over the integrity of the computer systems that manage elections in the United States increased following the 2016 election, the conference organizers began to develop ideas for a forum that would allow the scientific and election administration communities to collaborate and advance the ways we identify and prevent attempts to subvert the effective administration of elections.

What resulted from their efforts was not just a meeting of the minds to discuss technical innovations and methods in election audits. It was also a lively conversation of how we can, collectively, put those ideas into practice—the essential steps to setting audits up for success, and in doing so, secure both an election’s integrity and voters’ confidence in it. If you weren’t able to attend, recordings of every session are available online—but if you don’t have a day and a half to spend with them, we’ve also summarized a few of the major themes and takeaways for you here.

There were, of course, important discussions of the scientific side of things. “Elections are being run under more exacting tolerances,” said Charles Stewart III, a professor at MIT and one of the conference organizers, in the opening panel, underlining the need for election policies, equipment, and even the people involved to strive for new levels of technical precision and clarity. Presentations covered Bayesian and k-cut sampling methods, risk-limiting audit (RLA) strategies, the possibilities and challenges of end-to-end verifiability, and more. In parallel, participants discussed suggestions for how to explain complicated techniques to voters and policymakers in ways that are easy to grasp.

Foundational concepts of auditing and elections were pushed, stretched, and tested by participants over coffee in the hallways and during panels. How—and which—stakeholders should be involved at which stages of an election audit? Where and when should data be accessible, and to whom? What steps can we take to improve and normalize audits, build voter confidence in election processes, or deter malicious interference? Who might be a useful resource as we do so, beyond the usual #electiongeek suspects?

An important theme that emerged in nearly every panel was the absolute impossibility of effective innovation without connecting it directly to the mundane realities and complexities of election systems as a whole. “An audit is no better than the paper trail it uses,” offered Philip Stark of the University of California at Berkeley on the first morning; Whitney Quesenbery of the Center for Civic Design took that and ran with it, challenging attendees to consider “democracy as a design problem,” and to think more critically about the tangled implications of a decision as seemingly simple as the formatting of a single paragraph of ballot text.

Throughout the summit, we were lucky to have insights and expertise from folks around the country who have been in the weeds of audit implementation for years. A panel of Coloradan experts drew from their experiences to illustrate the necessity of a safe learning environment and layered trainings for election workers. To implement an RLA effectively, they emphasized, they had to “go slow to go fast,” planning ahead to provide enough time and space for workers to ask questions, make mistakes, and build their confidence. Stories from New Jersey, Indiana, Virginia, California, and Michigan—where a pilot RLA had wrapped up just a few days before the summit convened—also offered important lessons on successful audit research and implementation.

None of the panelists or participants pretended that audits offered a panacea. “There are no silver bullets in elections,” commented Matthew Masterson (of the Department of Homeland Security), in a closing discussion; a truth that perhaps no one understands better than those who work inside those elections. That said, we’ve gathered some of the best advice offered at the summit to offer to you here:

– Make laws vague; make rules specific. Instead of codifying (one might say calcifying) all of your audit policies and procedures up front, give them the space to evolve. Allow yourself the capacity to improve and adapt as you gain knowledge.

– Have empathy. To improve elections, academics must build relationships with election officials that are based on trust. Understand the needs of voters and stakeholders first, and provide recommendations that are actionable and digestible for election administrators.

– Complex systems fail in complex, nonlinear ways. No one likes to dwell on failure, but Ben Adida of VotingWorks put on his Silicon Valley hat and called for the election community to do just that. Success gives us one set of lessons learned; what can the failures teach us?

– Common. Data. Format. This could be an entire post on its own, but standards for a common data format came up repeatedly as an important element in ensuring that all states can conduct the most effective RLAs going forward.

– Collaboration is key. As co-organizer Jennifer Morrell of Democracy Fund put it, collaboration is essential to solving many of the complex problems facing election officials today, including audits. Ultimately, this is a hinge on which the rest of these takeaways depend.

So, where to from here? Beyond a few days of conversations, where does this Election Audit Summit put us?

Well, for one thing, we hope that it provided fertile ground for new collaborations (see what we did there?) on post-election audits. Strong auditing procedures, conducted with transparency, can play a critical role in ensuring that voters have high confidence in the election process and integrity of the results. There’s a growing community of experts in academia, public service, and grassroots circles alike that are developing more scientifically rigorous approaches to auditing elections—the summit was just a taste. We hope you’ll stay tuned to what they’re doing (see, for example, our expanding and not at all exhaustive list of suggested resources), or dip a toe into a new collaboration yourself.

This was by all reports a fantastic meeting; if you’re interested in seeing the panels for yourself you can check out the archived webcast here. Thanks to Claire for sharing this recap (and to electionline’s Mindy Moretti for giving it a place in the newsletter) – and kudos to the organizers and all the panelists for their efforts to not only give the idea of audits a wider audience but also to continue the effort to build a community around the notion that audits make our voting system stronger.

Have a great weekend – and stay tuned…