4 ms·

Isn't it better to measure every device, including paper and pencil, to see what sorts of errors you get with each?

MiddleEndian 8y ago
In theory, maybe. In practice, how do we do that? There have been zero successful investigations as a result of "flaky" voting machines in America to my knowledge. And what are the incentives for using voting machines? They are harder to verify than people from multiple parties counting lines drawn on a sheet of paper. They are susceptible to conflicts of interest, general incompetence, and security risks.
- greglindahl 8y ago
  I've read a few studies in the past. What you do to experiment is send people to use the equipment, then quiz them about how they voted, and see what the equipment says they voted. Problems like the confusing "butterfly ballot" in the famous Florida election pop right out. You can also look at issues like undervotes and overvotes from the actual election. And ask people if they have any complaints about the process. This is all somewhat studied already. We should study it a lot more. This is in addition to doing systematic post-election recounts of a small % of ballots to verify that the counting machines are working. Finally, everything is a voting "machine", including optical scan cards. Even if hand-counted. You should measure how well they work.
  - MiddleEndian 8y ago
    Even if we somehow verify that reliable machines exist, how do we verify that the machines haven't been tampered with? So far we've seen all sorts of problems with the machines in practice, usually in favor of the party in power. What's our incentive to switch to machines? Speed isn't a concern here.
    - greglindahl 8y ago
      Here's one audit scheme to verify that the counting machines are reliable: https://www.verifiedvoting.org/resources/post-election-audits/ https://www.verifiedvoting.org/resources/post-election-audit... One good way to implement this is with paper ballots and optical scan machines. You get speed/less labor with accuracy. But you still need to study that the paper ballots are reliably usable by all types of voters, in addition to verifying that the machines count accurately.
- sturgill 8y ago
  Paper ballots aren’t immune to issues. Like the Al Franken election and ballots in the trunk; hanging chads in Florida; etc. I don’t know what the answer is here. I do not trust electronic voting machines. Paper isn’t immune to issues. Personally I’d love to see a dual system: votes are recorded electronically with a paper receipt.
  - greglindahl 8y ago
    Paper ballots with optical scan machines are popular in states that care about these issues. Compared to your suggestion there are fewer machines, easier recounts and audits, and it's very easy for the public to understand.
    - reaperducer 8y ago
      Paper ballots with optical scan machines are the reason we have electronic voting now. It was paper ballots and optical scan machines that led to the whole "hanging chad" situation. And filling in a bubble with a pen or pencil isn't going to help. Anyone who read BBS G-philes in the 80's knows how to defeat a Scan-Tron.
      - greglindahl 8y ago
        ... I wasn't recommending using the worst possible paper ballot. The modern version, that tests well, is where you draw a line to complete an arrow. And this modern version is the one that's becoming popular.
  - MiddleEndian 8y ago
    Re: hanging chads This is a problem with overtechnical solutions already. Random mechanical points of failure. Pencil drawing a line is unambiguous. Re: ballots in a trunk This was an auditing/ procedural issue. Re: receipts Why not just turn in the receipts and count them. Why have a black box do the counting? If a machine says candidate A had a thousand forty votes and B had a thousand thirty, what are you going to do? How do you know if your receipt lied? If people keep their receipts, you no longer have anonymous voting. And what are you going to do, track down every voter? If those running the election keep the receipts, they should just count them manually.
wavefunction 8y ago
Yes but given limited resources it's more feasible to obtain a random sample of the population you're looking to gain insight with and derive your observations from the sample. Assuming it's a properly representative sample.
- greglindahl 8y ago
  ... that's what is usually done, yes? If anything is done at all, which is the typical worst case today.
atoav 8y ago
It is not only about bugs. It is also about the question whether everybody can wrap their head around it. This is a problem with every electronic/digital voting process – even if you are a engineer figuring out if that machine is safe would be not that straightforward. If speed is not an issue then the voting mechanism that everybody (even the technically unskilled) can understand is the best. And beeing slow is not a bug, it is a feature. A feature that prevents all kinds of things from happening that would undermine trust in the system as a whole.
- greglindahl 8y ago
  When I said "errors" I didn't only mean bugs. I mean failing to have a counted vote that's the same as what the voter intended. The method I like most is paper ballots with optical scanners, which the general public can understand. Random audits for confirmation, yes, the general public can understand. Recounts, and hand recounts, yes, the general public can understand.
  - jacques_chester 8y ago
    It just doesn't take that long to count by hand. You wait an extra day or week or two to get the final count. So what? This has consequences for years, so take your damn time and count every single ballot. A machine cannot be watched in the same way a human can. No scrutineer can prove that a ballot was counted correctly without ... seeing it counted.
- btilly 8y ago
  And beeing slow is not a bug, it is a feature. A feature that prevents all kinds of things from happening that would undermine trust in the system as a whole. 2004, Bush vs Kerry in Ohio. The voting commissioner was Republican and had promised to deliver Ohio. Without Ohio, Bush would not have won the election. Heavily Republican precincts were set up so that voting was easy all day. Heavily Democratic precincts in Cincinnati got so backlogged that they closed the door and people had to be in line for hours to vote. Bush won the state by 2.1%. He carried Harris County, which includes Cincinnati. No Republican has carried Harris County since. Now tell me. Does the feature of voting "beeing slow" improve your faith in the fairness of that election?

In theory, maybe. In practice, how do we do that? There have been zero successful investigations as a result of "flaky" voting machines in America to my knowledge. And what are the incentives for using voting machines? They are harder to verify than people from multiple parties counting lines drawn on a sheet of paper. They are susceptible to conflicts of interest, general incompetence, and security risks.

greglindahl 8y ago
I've read a few studies in the past. What you do to experiment is send people to use the equipment, then quiz them about how they voted, and see what the equipment says they voted. Problems like the confusing "butterfly ballot" in the famous Florida election pop right out. You can also look at issues like undervotes and overvotes from the actual election. And ask people if they have any complaints about the process. This is all somewhat studied already. We should study it a lot more. This is in addition to doing systematic post-election recounts of a small % of ballots to verify that the counting machines are working. Finally, everything is a voting "machine", including optical scan cards. Even if hand-counted. You should measure how well they work.
- MiddleEndian 8y ago
  Even if we somehow verify that reliable machines exist, how do we verify that the machines haven't been tampered with? So far we've seen all sorts of problems with the machines in practice, usually in favor of the party in power. What's our incentive to switch to machines? Speed isn't a concern here.
  - greglindahl 8y ago
    Here's one audit scheme to verify that the counting machines are reliable: https://www.verifiedvoting.org/resources/post-election-audits/ https://www.verifiedvoting.org/resources/post-election-audit... One good way to implement this is with paper ballots and optical scan machines. You get speed/less labor with accuracy. But you still need to study that the paper ballots are reliably usable by all types of voters, in addition to verifying that the machines count accurately.
sturgill 8y ago
Paper ballots aren’t immune to issues. Like the Al Franken election and ballots in the trunk; hanging chads in Florida; etc. I don’t know what the answer is here. I do not trust electronic voting machines. Paper isn’t immune to issues. Personally I’d love to see a dual system: votes are recorded electronically with a paper receipt.
- greglindahl 8y ago
  Paper ballots with optical scan machines are popular in states that care about these issues. Compared to your suggestion there are fewer machines, easier recounts and audits, and it's very easy for the public to understand.
  - reaperducer 8y ago
    Paper ballots with optical scan machines are the reason we have electronic voting now. It was paper ballots and optical scan machines that led to the whole "hanging chad" situation. And filling in a bubble with a pen or pencil isn't going to help. Anyone who read BBS G-philes in the 80's knows how to defeat a Scan-Tron.
    - greglindahl 8y ago
      ... I wasn't recommending using the worst possible paper ballot. The modern version, that tests well, is where you draw a line to complete an arrow. And this modern version is the one that's becoming popular.
- MiddleEndian 8y ago
  Re: hanging chads This is a problem with overtechnical solutions already. Random mechanical points of failure. Pencil drawing a line is unambiguous. Re: ballots in a trunk This was an auditing/ procedural issue. Re: receipts Why not just turn in the receipts and count them. Why have a black box do the counting? If a machine says candidate A had a thousand forty votes and B had a thousand thirty, what are you going to do? How do you know if your receipt lied? If people keep their receipts, you no longer have anonymous voting. And what are you going to do, track down every voter? If those running the election keep the receipts, they should just count them manually.

wavefunction 8y ago

Yes but given limited resources it's more feasible to obtain a random sample of the population you're looking to gain insight with and derive your observations from the sample. Assuming it's a properly representative sample.

greglindahl 8y ago
... that's what is usually done, yes? If anything is done at all, which is the typical worst case today.

atoav 8y ago

It is not only about bugs. It is also about the question whether everybody can wrap their head around it. This is a problem with every electronic/digital voting process – even if you are a engineer figuring out if that machine is safe would be not that straightforward. If speed is not an issue then the voting mechanism that everybody (even the technically unskilled) can understand is the best. And beeing slow is not a bug, it is a feature. A feature that prevents all kinds of things from happening that would undermine trust in the system as a whole.

greglindahl 8y ago
When I said "errors" I didn't only mean bugs. I mean failing to have a counted vote that's the same as what the voter intended. The method I like most is paper ballots with optical scanners, which the general public can understand. Random audits for confirmation, yes, the general public can understand. Recounts, and hand recounts, yes, the general public can understand.
- jacques_chester 8y ago
  It just doesn't take that long to count by hand. You wait an extra day or week or two to get the final count. So what? This has consequences for years, so take your damn time and count every single ballot. A machine cannot be watched in the same way a human can. No scrutineer can prove that a ballot was counted correctly without ... seeing it counted.
btilly 8y ago
And beeing slow is not a bug, it is a feature. A feature that prevents all kinds of things from happening that would undermine trust in the system as a whole. 2004, Bush vs Kerry in Ohio. The voting commissioner was Republican and had promised to deliver Ohio. Without Ohio, Bush would not have won the election. Heavily Republican precincts were set up so that voting was easy all day. Heavily Democratic precincts in Cincinnati got so backlogged that they closed the door and people had to be in line for hours to vote. Bush won the state by 2.1%. He carried Harris County, which includes Cincinnati. No Republican has carried Harris County since. Now tell me. Does the feature of voting "beeing slow" improve your faith in the fairness of that election?