4 ms·

I'm as much of a technophile as any other HN reader, but I am continuously flummoxed that anyone makes use of the sort of electronic voting machines described i

I'm as much of a technophile as any other HN reader, but I am continuously flummoxed that anyone makes use of the sort of electronic voting machines described in this article. This is mainly from the perspective that a lot of the redundancy and auditability of the system seemed to rely on a bunch of printers working perfectly. Paper ballots and a pencil/pen are still hugely more reliable and trustworthy, IMO.

mhh__ 6y ago
When all you have is a hammer, everything looks like a nail. In the UK, everything is done on paper - even if there was potential for fraud, scaling it across the whole nation would take years and huge amounts of money (and even more hush money)
- smt88 6y ago
  We had a presidential election in 2000 that was stolen in public by the GOP, despite being on paper. They did it with a technicality. They're doing the same with signature matching (your signature on your ballot has to match a different signature that could be many years older). Paper is safe from people outside the system, but not from politicized courts that are able to ignore paper ballots for minor mistakes.
  - specialist 6y ago
    I liked how an election advocate buddy said it one time: "GWB doesn't lose his right to speech when he makes another gaff. Under no circumstances should a voter lose their right to vote for a mistake." It just slays me that we're still arguing about this.
  - mhh__ 6y ago
    Keep in mind the whole hanging chad situation is also impossible because we just use pencils. But the legal aspect is troubling too
carapace 6y ago
I'm with you. To me electronic voting machines means you've given your democracy to the best hackers. "Hackerocracy"?
scollet 6y ago
IME the physical issues I've seen are entirely operator/technician error. I've really not seen any issues inside my wheelhouse. I feel for the devs having to codify abberant voting procedures.
- specialist 6y ago
  I'm not entirely certain that you're defending the gear. Or blaming the users. In case you are, I'll share two of my personal experiences. I hot fixed the VVPAT for my poll site's touchscreen. The cheap bent metal got warped, preventing correct operation. Imagine you had to change the thermal paper for a cash register made by lowest bid contractor designed by grifters who've never seen a cash register. These units were worse. I attended a handful of the "logic and accuracy" tests. One time, both the printer and memory card reader failed, of the the unit randomly chosen for testing. So they swamped parts and cards until they got it to work. Deemed a success. In truth, the only thing proven was that one of their printers could make marks on paper, given enough fiddling.
  - scollet 6y ago
    I will defer to your knowledge because it appears you have a deeper understanding of the internals. At a user level for me in this last general, the only issues I've observed have been independent of the machine(s). We are perhaps looking at different machines on different timelines and locale, but I don't doubt that there is a long way to go in terms of reliability and efficacy.
    - specialist 6y ago
      Thank you for followup. Because everyone does everything their own way, it's very hard to generalize about elections in the USA. Which is a problem all on its own. In reflecting on your direct observations (experiences), I'm chewing on why we have different assumptions. I'd say most of the time, most of the gear and procedures work pretty good. Prior to a botched local election in 2004, I had a very high opinion of my local election administration. And it was only after really digging into the problem space that I started to see when the reality didn't quite match the narrative. One thing noob me didn't appreciate was that with FPTP form of elections, our system is unnecessarily brittle. My large county has at least one mandatory manual recount every general election. Triggered when the margin of victory is smaller than the error rate. That's when the real drama starts. Honestly, I feel ridiculous obsessing over jitter in the process when the correct fix is to switch to a more robust form of elections. If we switched to a combo of ranked choice voting and proportional representation, the stakes would be so much lower, and trust in the process would be so much higher. Again, thank you for replying. I'm grateful that you and many others have leaned into this issue. More eyeballs (attention) is better.
sfteus 6y ago
IMO, the benefits of having electronic voting with an auditable paper trail and backup outweighs the downsides, if implemented properly. The process described in the article seems a little different than the ones we used in Texas (Dallas County). We have our registration verified and check in, and then are given a paper ballot. You put that into the voting machine, cast your ballot which is recorded electronically on the machine, then printed on the paper ballot and returned to you. You're told to inspect it for accuracy, then deposit it in a _separate_ tabulating machine, which scans it and displays what it recorded when reading your vote, tabulates your vote electronically, and stores the paper backup in a locked deposit box inside the machine. From my understanding, if anything goes wonky during this process and you notice it, you can have the ballot invalidated and receive another to cast. At the end of the day, the counts are printed and stored along with the memory cards on every machine, and those and the secure ballot boxes are all stored with tamper evident seals. So you have the electronic count on the voting machine from the memory card, the printed count from the voting machine generated at the end of the day, the electronic count on the tabulating machine from its memory card, the printed count from the tabulating machine generated each day, as well as the paper ballots as a backup, which from my understanding are re-tabulated en-masse on larger and faster machines after the election to get a second count as part of the certification process. All data points have to match. None of the machines are connected to the internet, so you'd have to gain physical access to them or the memory cards, and anything done would have to match the counts on all the other associated machines and receipts, plus the paper ballot backups. The Secretary of State also generates a list of voters who cast ballots each day by county and voting location based on when you first check in to vote, so that's another publicly auditable list that would have to match that is not related to the voting machines themselves. Paper ballots are definitely more simple and probably just as secure, if not more, if handled properly. However, having multiple points of redundancy is invaluable to help quickly identify issues and ensure everything matches over multiple steps of the process.

mhh__ 6y ago

When all you have is a hammer, everything looks like a nail. In the UK, everything is done on paper - even if there was potential for fraud, scaling it across the whole nation would take years and huge amounts of money (and even more hush money)

smt88 6y ago
We had a presidential election in 2000 that was stolen in public by the GOP, despite being on paper. They did it with a technicality. They're doing the same with signature matching (your signature on your ballot has to match a different signature that could be many years older). Paper is safe from people outside the system, but not from politicized courts that are able to ignore paper ballots for minor mistakes.
- specialist 6y ago
  I liked how an election advocate buddy said it one time: "GWB doesn't lose his right to speech when he makes another gaff. Under no circumstances should a voter lose their right to vote for a mistake." It just slays me that we're still arguing about this.
- mhh__ 6y ago
  Keep in mind the whole hanging chad situation is also impossible because we just use pencils. But the legal aspect is troubling too

carapace 6y ago

I'm with you. To me electronic voting machines means you've given your democracy to the best hackers. "Hackerocracy"?

scollet 6y ago

IME the physical issues I've seen are entirely operator/technician error. I've really not seen any issues inside my wheelhouse. I feel for the devs having to codify abberant voting procedures.

specialist 6y ago
I'm not entirely certain that you're defending the gear. Or blaming the users. In case you are, I'll share two of my personal experiences. I hot fixed the VVPAT for my poll site's touchscreen. The cheap bent metal got warped, preventing correct operation. Imagine you had to change the thermal paper for a cash register made by lowest bid contractor designed by grifters who've never seen a cash register. These units were worse. I attended a handful of the "logic and accuracy" tests. One time, both the printer and memory card reader failed, of the the unit randomly chosen for testing. So they swamped parts and cards until they got it to work. Deemed a success. In truth, the only thing proven was that one of their printers could make marks on paper, given enough fiddling.
- scollet 6y ago
  I will defer to your knowledge because it appears you have a deeper understanding of the internals. At a user level for me in this last general, the only issues I've observed have been independent of the machine(s). We are perhaps looking at different machines on different timelines and locale, but I don't doubt that there is a long way to go in terms of reliability and efficacy.
  - specialist 6y ago
    Thank you for followup. Because everyone does everything their own way, it's very hard to generalize about elections in the USA. Which is a problem all on its own. In reflecting on your direct observations (experiences), I'm chewing on why we have different assumptions. I'd say most of the time, most of the gear and procedures work pretty good. Prior to a botched local election in 2004, I had a very high opinion of my local election administration. And it was only after really digging into the problem space that I started to see when the reality didn't quite match the narrative. One thing noob me didn't appreciate was that with FPTP form of elections, our system is unnecessarily brittle. My large county has at least one mandatory manual recount every general election. Triggered when the margin of victory is smaller than the error rate. That's when the real drama starts. Honestly, I feel ridiculous obsessing over jitter in the process when the correct fix is to switch to a more robust form of elections. If we switched to a combo of ranked choice voting and proportional representation, the stakes would be so much lower, and trust in the process would be so much higher. Again, thank you for replying. I'm grateful that you and many others have leaned into this issue. More eyeballs (attention) is better.

sfteus 6y ago

IMO, the benefits of having electronic voting with an auditable paper trail and backup outweighs the downsides, if implemented properly. The process described in the article seems a little different than the ones we used in Texas (Dallas County). We have our registration verified and check in, and then are given a paper ballot. You put that into the voting machine, cast your ballot which is recorded electronically on the machine, then printed on the paper ballot and returned to you. You're told to inspect it for accuracy, then deposit it in a _separate_ tabulating machine, which scans it and displays what it recorded when reading your vote, tabulates your vote electronically, and stores the paper backup in a locked deposit box inside the machine. From my understanding, if anything goes wonky during this process and you notice it, you can have the ballot invalidated and receive another to cast. At the end of the day, the counts are printed and stored along with the memory cards on every machine, and those and the secure ballot boxes are all stored with tamper evident seals. So you have the electronic count on the voting machine from the memory card, the printed count from the voting machine generated at the end of the day, the electronic count on the tabulating machine from its memory card, the printed count from the tabulating machine generated each day, as well as the paper ballots as a backup, which from my understanding are re-tabulated en-masse on larger and faster machines after the election to get a second count as part of the certification process. All data points have to match. None of the machines are connected to the internet, so you'd have to gain physical access to them or the memory cards, and anything done would have to match the counts on all the other associated machines and receipts, plus the paper ballot backups. The Secretary of State also generates a list of voters who cast ballots each day by county and voting location based on when you first check in to vote, so that's another publicly auditable list that would have to match that is not related to the voting machines themselves. Paper ballots are definitely more simple and probably just as secure, if not more, if handled properly. However, having multiple points of redundancy is invaluable to help quickly identify issues and ensure everything matches over multiple steps of the process.