Public trust demands open-source voting systems

That pretty much undermines the entire concept of unit and integration testing.

If you're saying we should be writing voting machine code in ML and keeping the firmware in Fort Knox, I'm going to make the argument that it's a lot cheaper to do sampled hand-counts to check against machine error or tampering... Which we already do.

Did you look at the link at all? That's what this company sells. They make ballot marking devices that print your vote on a ballot paper, then a separate ballot box that counts the votes by scanning the ballot papers.

I'm watching him talk about the two key ingredients of an election (anonymity and trust, for those not watching the video) and thinking "We don't have those in U.S. elections".

I live in California, where the voting method is vote-by-mail and you sign your ballot. That breaks anonymity right there, plus there's a barcode that matches address and ballot for traceability, so in theory anyone involved in the election process could look at my ballot, cross-reference against address, and figure out how I voted. In practice I've never heard of anyone being pressured or confronted based on how they voted, so my default assumption is this doesn't happen much or at all.

But even broader, in the U.S. your party registration is public information. That's why whenever there's a political shooting, the media always says "He was a registered Republican" or "registered Democrat" or "was not registered to vote". And this mechanism is actively and publicly being exploited to alter elections. Since the U.S. is a two-party system and party membership is public, you have a fairly good idea how each precinct is going to vote before they vote, and can gerrymander maps to get the outcomes you want.

Plenty of trust issues in physical ballot transfer as well. California is vote-by-mail, but that assumes the postal service is a reliable carrier, while there was just a recent news story [1] about ballots being stolen. Before I lived in California, I was in Massachusetts, where we voted on 1930s-era lever voting machines where you hit a lever down and it marks a paper ballot without you ever seeing the real ballot. Between elections, these were stored backstage at the local middle school, so a mechanically-inclined middle schooler with knowledge of how an upcoming election's ballots would be formatted (and we did mock elections in middle school) could have rigged the machines to deliver the local precinct to their preferred candidate.

The useful points in the video were basically that decentralization and redundancy are what make physical elections hard to rig: you have to hack multiple locations to influence the overall election, and at each point you have multiple eyes watching you. He sets up the contrast with software voting, where you have the same software running on each machine, and even if the software is open-source, you can't be sure that the rest of the stack it's running on is secure (an oblique reference to the Ken Thompson Hack [2]).

But decentralization and redundancy are properties that you can introduce into software systems just as easily as real-wold systems. The KTH can be countered through Diverse Double-Compiling, for example [3]. zkStarks and digital signatures give you ability to prove that you authored something without revealing what that something is or who you are. The importance of client diversity for the security of the network as a whole has been well-known in the filesharing and crypto worlds. And anyone who has worked in Big Tech, aviation, or telecom could tell you that having multiple paths to success that are developed by independent teams is important for any computer system that is in a safety- or reliability-critical area.

[1] https://www.almanacnews.com/election/2025/10/14/ballots-stol...

[2] https://aeb.win.tue.nl/linux/hh/thompson/trust.html

[3] https://dwheeler.com/trusting-trust/

No. Democracy is not about reducing cruelty, or any other vaguely activists points of views. It is about having people choose where they want to go. It might be that these choices unveil that humanity, statistically speaking, is actually a cruel bunch. And, what you think is cruel, might be just fair to someone else. Democracy is about surfacing the human nature.

That's an inquisitive 7 year old. Definitely reward them. Let's explain. A good voting system needs to guarantee

- Secrecy of who voted for whom

- Transparency of everything else. The names of everybody in the process, the process itself and all the statistics should be verifiably public.

Being an observer to your polling station must be a guaranteed voter right. Similarly all participating parties must have the right to send representatives to observe the entire process.

Before opening the polling station all ballots are counted by multiple observers from all sides. This is recorded into files / documentation of each observer. So the number of possible ballot papers that can be voted on is documented.

Then each ballot paper needs to be stamped with a official local seal. This is also observed by every observer. The number of stamped ballots is also counted and documented. The number has to match the original ones.

The number of people who can vote in that voting station is determined by a population survey. In bigger cities each region must have roughly the same number of constituents.

The number of ballots that are stamped must match the number of eligible voters in the polling station. A voter can request to change a damaged ballot paper. The replacement should be done in front of all observers and the voter. The replaced ballot is destroyed in front of everyone.

After putting their ballot into the box, the voter has to sign their name in multiple printouts of the list of eligible voters of that polling station. These printouts of the lists are held by observers from multiple sides. The number of signatures has to match the number of ballots in the box.

Everybody can observe the count. All the numbers are checked against each other.

If you think that this is infeasible, I come from a country of 80 million people and live in a similarly sized one. Both of them use the same system. It works. It scales since it is an almost trivially parallelizable problem. We get the election results in the same day of voting.