On the Weaponization of Voting Channels

The speaker, a computer scientist who has observed elections as an international delegate, starts from a worry: when a country offers several ways to vote, the results from each channel can differ, and those gaps can be exploited to undermine confidence in an outcome. He points to a 2020 Pennsylvania results screen whose lead appeared to shift as mail-in ballots were counted, and argues that public trust matters as much as accurate counting.

The main case study is Estonia, which lets voters cast ballots online or on paper. In the 2023 count, preliminary results appeared stable after the internet votes were checked against paper records, then changed sharply when internet ballots were uploaded. A party that had told supporters to avoid internet voting lost about half of its apparent support and challenged the result in the Supreme Court, which dismissed the case on procedural grounds. A survey with a political scientist colleague found a significant drop in trust in internet voting, with the political extremes losing confidence most.

He closes with untested ideas: withholding channel-level breakdowns, publishing no preliminary results, moving to a single uniform channel, and creating independent bodies that run channel-specific risk-limiting audits. He admits these are rough and may not be practical. The takeaway for practitioners is that trust in each voting channel needs protecting, and election officials should plan for it before it is exploited.