Read on Twitter
Despite the (ill-informed) assurances of people who should know better, there is no way to run a secret, anonymous, secure ballot over the internet. It's a science fantasy, like faster-than-light drives or time machines. It's a thought experiment, not a plan.
1/
1/
(If you'd like an unrolled version of this thread to read or share, here's a link to it on http://pluralistic.net , my surveillance-free, ad-free, tracker-free blog:)
https://pluralistic.net/2021/04/12/tear-down-that-wall/#bmds
(or follow along below)
2/
https://pluralistic.net/2021/04/12/tear-down-that-wall/#bmds
(or follow along below)
2/
Elections are actually easy: paper ballots, hand-marked and hand counted in sight of scrutineers from opposing parties. But thanks to a highly consolidated vote-tech sector with plenty of money to spend, Americans have been convinced that this can't work for America.
3/
3/
It's a bizarre and innumerate proposition: America has more people, so it will have more ballots, so it can't count them by hand.
Uh, folks?
Canada and the UK don't consolidate all their ballots to a single counting-house where, like, eight people tally the nation's votes.
4/
Uh, folks?
Canada and the UK don't consolidate all their ballots to a single counting-house where, like, eight people tally the nation's votes.
4/
The votes are counted at the polling place. America has more polling places than Canada, but there's no reason it can't have the same ratio of polling places - and ballot counters - to voters as Canada does. To a first approximation, that's already true.
5/
5/
The American concern for electoral fraud is forever in tension with the American exceptionalist insistence on using dumpster-fire vote-tech sold by litigious grifters who sue the critics who blow the whistle on their awful security.
https://pluralistic.net/2021/01/11/seeing-things/#ess
6/
https://pluralistic.net/2021/01/11/seeing-things/#ess
6/
Thus it is that, year after year, security researchers - like the merry crew at Defcon's annual Voting Village - publish reports of jaw-dropping incompetence in vote-tech systems, each revealing how little progress has been made since the last.
https://media.defcon.org/DEF%20CON%2027/voting-village-report-defcon27.pdf
7/
https://media.defcon.org/DEF%20CON%2027/voting-village-report-defcon27.pdf
7/
To the extent that anyone serious about this stuff believes in electoral automation, the one technology they're willing to admit might someday be made secure is the "ballot marking device," a machine that fills in your ballot in a way that facilitates automatic counting.
8/
8/
But BMDs are still hugely controversial: even if you solve the problem of making sure that the machine-readable part and the human-readable part say the same thing (a Very Big Problem), there's a stubbornly intractable problem lurking right behind it: human factors.
9/
9/
In real-world situations and under laboratory conditions, people just don't carefully check their ballots, and the best interventions we have to encourage ballot-checking don't perform very well.
https://ieeexplore.ieee.org/document/9152705
10/
https://ieeexplore.ieee.org/document/9152705
10/
Now, an innovative BMD design from Professor @DrJuanGilbert of the University of Florida demonstrates a very clever technique and promising approach for increasing the likelihood that a voter will notice if their ballot is mis-marked.
11/
11/
The Transparent Voting Machine is a clear plastic box that shows you your ballot immediately after it's marked, and requires you to touch the spot directly over the place where your selection appears. Intuitively, this sounds like it should work, and experimentally, it does.
12/
12/
In his seminar presentation to @PrincetonCITP, Gilbert discusses his user studies, where he was able to substantially increase the likelihood that voters would both detect and report irregularities in their ballots.
https://mediacentral.princeton.edu/media/CITP+SeminarA+Juan+Gilbert+%E2%80%93+Can+Voters+Detect+Ballot+Manipulations+with+a+Transparent+Voting+MachineF/1_67s0hsf9
13/
https://mediacentral.princeton.edu/media/CITP+SeminarA+Juan+Gilbert+%E2%80%93+Can+Voters+Detect+Ballot+Manipulations+with+a+Transparent+Voting+MachineF/1_67s0hsf9
13/
In his writeup of the seminar, Princeton's Andrew Appel does the math on these studies and concludes that with Gilbert's technology, ballot mismarking would be detected if it occurred above the 0.5% mark: a major improvement over existing designs.
https://freedom-to-tinker.com/2021/04/12/juan-gilberts-transparent-bmd/
14/
https://freedom-to-tinker.com/2021/04/12/juan-gilberts-transparent-bmd/
14/
But Appel also follows up with some important caveats: about the physical security of a BMD, the sociology of pollworkers, and what a secretary of state should do if the winning margin is within 0.5%.
15/
15/
BMDs aren't useless. They can be a boon to accessibility, transcending barriers of disability and language fluency. But they're incredibly hard to get right, and the industry that produces, sells and services them is characterized by incompetence and bullying.
16/
16/
America loves vote-tech, but vote-tech is pretty iffy on American democracy. Gilbert's design is incredibly clever and vastly superior to the existing technologies, and if it were adopted, it would be preferable to everything in use today.
17/
17/
That is an admittedly low bar. And meanwhile, people all over the world continue to mark their paper ballots with golf pencils and put them in boxes so that the nice poll-workers can count them by hand later that day.
eof/
eof/
