,
6 tweets,
1 min read
Read on Twitter
This is potentially quite important; truly secure hardware would help with a number of difficult election problems. Architecture matters, of course. It will be interesting to see if software independent systems can be built from this.
What's important here is that the design not attempt to *substitute* hardware and software security (however strong and "provable" it may be) for election audibility and software independence. They appear to be taking the these requirements seriously, which is encouraging.
Put another way: risk limiting audits allow us to have trustworthy elections even when the computing systems used to conduct them are untrustworthy. But that doesn't mean we shouldn't be working to make these systems more trustworthy - current systems are truly awful.
One other note here: vote casting/counting are not the only aspects of elections that use computers or that might benefit from secure hardware. Voter registration, precinct poll books, ballot creating, tally reporting, etc, are all critical functions that require strong security.
And if you think current voting machine security is bad, wait till you see how these backend systems are set up. They are arguably the most vulnerable part of our election infrastructure.
But they get much less attention than voting machines, because they aren't as visible, and the problems are much messier to solve.
