4 ms·

> All your system tells a voter is that there's a mapping somewhere of their key to their vote. It doesn't say that vote is included in the total. It's a datab

> All your system tells a voter is that there's a mapping somewhere of their key to their vote. It doesn't say that vote is included in the total. It's a database of votes and tokens used to cast them. You could also query what tokens were used to vote X and check that they make up the total. You could see the token you used on the list of Y you voted. > Implementing and maintaining this system, as well as training administrators how to use it, would cost millions of dollars, and lead to all kinds of controversies because people don't remember their votes all that well [1]. It's one thing to remember what you voted a few days after the election, and another to remember it months later when polled, which is what [1] is about. This system is meant to increase public confidence in the election result. Wouldn't you think it worth a few millions, considering the aftermath of 2020 elections? > It's also worth saying whether or not your vote was counted is already public record, resulting in even less relative benefit for a system like this. You are assured that it is counted, but with this scheme you could easily check that there's a vote cast with token you used. Btw, this system sort of exists already in countries where you write the candidate's name or number on the ballot, and ballots are publicly available for recounting; the token is your handwriting. It just isn't anonymous as someone else could also recognise it.

camgunz 5y ago
> It's a database of votes and tokens used to cast them. You could also query what tokens were used to vote X and check that they make up the total. You could see the token you used on the list of Y you voted. Not everyone will use a token or remember it. If even 50% of voters do this 100% accurately, it still isn't statistically helpful. > It's one thing to remember what you voted a few days after the election, and another to remember it months later when polled, which is what [1] is about. Maybe, but to me it's about the frailty of memory, and the US has a lot of elections. They won't remember who they voted for for State House/Senate, Judge, Sheriff, Lieutenant Governor, Attorney General, etc. Ballots are also complicated; maybe your vote for US Representative counted, but you goofed on Mayor and it didn't. You might look up your vote and think "shenanigans!", and there won't be any way of looking at your original ballot to find out what happened. And even if there were, you still wouldn't have any way of verifying that your vote was included in the count. It's also worth saying that if there's any problem with this system, it calls the entire election into question, even if the election itself was fine or statistically far from error. > Btw, this system sort of exists already in countries where you write the candidate's name or number on the ballot, and ballots are publicly available for recounting; the token is your handwriting. It just isn't anonymous as someone else could also recognise it. Sure and I can see your system as an anonymized upgrade of it. But again, because you can't audit the count and there's a wide gulf between the votes you can audit and the count, it doesn't move us closer to the goal of verifying an election. Continuing down the path of existing election systems, there is of course the recount system. So for most elections where it might make a difference, recounts go through all the ballots meticulously with party representatives, government officials, and lawyers present. Barring things like SCOTUS interference, these processes are really good, oftentimes finding more votes, curing lots and lots of ballots, and contributing to the corpus of wacky ballot analysis (counties get to design ballots and, wow, they could use some assistance). Which is to say, we already have a system that maps voters to votes, audited by elected officials, legal officials, and partisan representatives. Whenever there's any question, we activate that system that has maximum information in order to verify an election result. Cryptography only muddles this system--perhaps hopelessly. > This system is meant to increase public confidence in the election result. Wouldn't you think it worth a few millions, considering the aftermath of 2020 elections? This is a great topic change! We (especially in tech) focus myopically on election fraud in the US, as though that's the thing that would undermine the legitimacy of an election. But the incidence of that is basically nil, and doesn't merit any effort to improve. The current hullabaloo (which is an inflated case of the going hullabaloo) comes from Republican leaders gaslighting Americans in order to cling to power. I am 100% sure that even if we implemented this system, they would still work just as hard to undermine election legitimacy. It's not a tech problem. If we want to make our elections more legitimate, we need to make them simpler, cheaper, and more accessible. Policies that do this include: early voting, same day registration, mail voting, compulsory voting, election day holidays, mandated time off to vote, adequate polling place capacity, non-partisan districting boards, and enforced laws about voter intimidation and misinformation. Or, even something as simple as common ballot design would help a lot of older voters and voters with low levels of literacy. But to wrap up: I think we'd both like to increase the legitimacy of elections, but spending resources (whether that's time, money, or something more abstract like system complexity or fragility) on a cryptographic vote verification system is misguided because the US doesn't have fraudulent or incorrect election results, and there are no elections where such a system would have done better than the existing recount system. On the other hand, decades of political science study has discovered and reaffirmed time and time again that some simple policies would go a long way towards fixing: - bonkers disenfranchisement (via gerrymandering, intimidation, class-based barriers like long lines, etc.) rendering our representation completely non-descriptive (in terms of race, sex, gender, politics, immigrant status, ethnicity, sexual preference, class, you name it) - the majority of our representative bodies being elected with wildly meager portions of the vote, or highly partisan sections of the vote (in the case of very partisan districting), or both These are the things that truly threaten the legitimacy of our elections.
- TheMonarchist 5y ago
  > Not everyone will use a token or remember it. If even 50% of voters do this 100% accurately, it still isn't statistically helpful. I think this is its main weakness. Those who don't care enough to use a token would leave the token of zero effort (nothing) open for manipulations. Unless those are few, this system won't make the election more secure. It would however give those who care a way to ensure that their vote was counted, increasing public confidence in the results. For I assume that those who care enough to question the results also care enough to use a token and check it. > It's also worth saying that if there's any problem with this system, it calls the entire election into question, even if the election itself was fine or statistically far from error. > You might look up your vote and think "shenanigans!", and there won't be any way of looking at your original ballot to find out what happened. The token is only needed to identify a vote, so it could be dead simple, like a bitmap that could be a grid to be painted on paper ballot. It can be just added to existing systems. > And even if there were, you still wouldn't have any way of verifying that your vote was included in the count. How is querying the tokens used to vote what you voted, seeing your token in the list, and checking that the tokens make up the reported total not verifying it? > We (especially in tech) focus myopically on election fraud in the US, as though that's the thing that would undermine the legitimacy of an election. I think it's just because it seems like something that could be solved. > I am 100% sure that even if we implemented this system, they would still work just as hard to undermine election legitimacy. It would be a little harder when those who care can just check that their vote was counted. > I think we'd both like to increase the legitimacy of elections -- It seems to me that we have somewhat different aims. You have zeal for democracy and are willing to compromise ballot secrecy to verify the election result. I accept that verifying it is impossible while maintaing the secrecy, unless everyone is willing to check that his vote was counted. In practice I would be content if everyone just had that possibility, as I don't have unrealistic expectations of its popularity.
  - camgunz 5y ago
    >> And even if there were, you still wouldn't have any way of verifying that your vote was included in the count. > How is querying the tokens used to vote what you voted, seeing your token in the list, and checking that the tokens make up the reported total not verifying it? There can still be two separate systems: the system that has your actual vote and the system we use to determine who won the election. One of the (many) issues I have with cryptographic vote verification systems is that they don't prove there isn't a "second set of books" that all checks out, meanwhile we use the crooked books to determine who won. Fundamentally, without a way to know all the possible voters and how they voted, you can't defend against that. >> We (especially in tech) focus myopically on election fraud in the US, as though that's the thing that would undermine the legitimacy of an election. > I think it's just because it seems like something that could be solved. Yeah but it's typical, and sorry for the bluntness here, tech arrogance--or maybe tech optimism is nicer. But either way, there's a consistent dynamic of tech people trying to solve things with tech that they really don't know a lot about. I've worked as a Democratic operative specifically on elections on the past few cycles and I'm pretty close to an expert. I'm not making an appeal to authority or anything, and I'm more than happy to talk about ways we can use tech to improve US elections, but it's just that I know these schemes are way, way more trouble than they're worth and either hinder a lot of the techniques we currently use to verify elections, or make political firestorms that make the election process even more fragile than it already is way more likely. It's also worth saying experts get this wrong too! The Iowa caucuses in 2020 had some tech hiccups, and that's, in my opinion, because no one can make sense of a caucus. They're beyond broken, and the more you measure them the more you'll find they're 100% unreliable and indicative of nothing. We certainly shouldn't be showering the person who wins by .5% with millions of dollars of donor money and earned media. But more broadly, it's also that elections are fundamentally messy, and the more you measure the more you find problems. Either we accept that and build solutions that accept that, or we don't and we live with constant allegations that our elections are fraudulent, courtesy of the perspective that we should be able to have 100% certainty. > It seems to me that we have somewhat different aims. You have zeal for democracy and are willing to compromise ballot secrecy to verify the election result. Honestly, I'm not sure what my ideal system is. I think anonymous voting is really important, but I also strongly support compulsory voting, and those two things are in direct tension. I think it's just hard to accept that elections are fuzzy measurements of sentiment and that the role of math in their verification is limited to some basic statistical analysis while the rest is process and politics. And I think it's really easy to be nerd sniped by the verification part, when tech actually could do so much good in the process and politics part. For example, a lot of States have "wet ink" statutes that mean voter registration forms have to be signed by a literal pen. So guess what Democrats did: they built a robot that holds a pen and signs a voter registration form remotely when someone fills out a voter reg form online. Hooray for tech! But also, that cost a lot of money and time. Wouldn't it have been easier if State legislatures just weren't assholes? Anyway, I would say if we're gonna implement a system that costs whatever the cheapest cryptographic verification system costs, we should just fund early voting programs and law enforcement initiatives against voter intimidation and misinformation. I know even that is impossible, which is depressing, but c'est la vie.

camgunz 5y ago

> It's a database of votes and tokens used to cast them. You could also query what tokens were used to vote X and check that they make up the total. You could see the token you used on the list of Y you voted. Not everyone will use a token or remember it. If even 50% of voters do this 100% accurately, it still isn't statistically helpful. > It's one thing to remember what you voted a few days after the election, and another to remember it months later when polled, which is what [1] is about. Maybe, but to me it's about the frailty of memory, and the US has a lot of elections. They won't remember who they voted for for State House/Senate, Judge, Sheriff, Lieutenant Governor, Attorney General, etc. Ballots are also complicated; maybe your vote for US Representative counted, but you goofed on Mayor and it didn't. You might look up your vote and think "shenanigans!", and there won't be any way of looking at your original ballot to find out what happened. And even if there were, you still wouldn't have any way of verifying that your vote was included in the count. It's also worth saying that if there's any problem with this system, it calls the entire election into question, even if the election itself was fine or statistically far from error. > Btw, this system sort of exists already in countries where you write the candidate's name or number on the ballot, and ballots are publicly available for recounting; the token is your handwriting. It just isn't anonymous as someone else could also recognise it. Sure and I can see your system as an anonymized upgrade of it. But again, because you can't audit the count and there's a wide gulf between the votes you can audit and the count, it doesn't move us closer to the goal of verifying an election. Continuing down the path of existing election systems, there is of course the recount system. So for most elections where it might make a difference, recounts go through all the ballots meticulously with party representatives, government officials, and lawyers present. Barring things like SCOTUS interference, these processes are really good, oftentimes finding more votes, curing lots and lots of ballots, and contributing to the corpus of wacky ballot analysis (counties get to design ballots and, wow, they could use some assistance). Which is to say, we already have a system that maps voters to votes, audited by elected officials, legal officials, and partisan representatives. Whenever there's any question, we activate that system that has maximum information in order to verify an election result. Cryptography only muddles this system--perhaps hopelessly. > This system is meant to increase public confidence in the election result. Wouldn't you think it worth a few millions, considering the aftermath of 2020 elections? This is a great topic change! We (especially in tech) focus myopically on election fraud in the US, as though that's the thing that would undermine the legitimacy of an election. But the incidence of that is basically nil, and doesn't merit any effort to improve. The current hullabaloo (which is an inflated case of the going hullabaloo) comes from Republican leaders gaslighting Americans in order to cling to power. I am 100% sure that even if we implemented this system, they would still work just as hard to undermine election legitimacy. It's not a tech problem. If we want to make our elections more legitimate, we need to make them simpler, cheaper, and more accessible. Policies that do this include: early voting, same day registration, mail voting, compulsory voting, election day holidays, mandated time off to vote, adequate polling place capacity, non-partisan districting boards, and enforced laws about voter intimidation and misinformation. Or, even something as simple as common ballot design would help a lot of older voters and voters with low levels of literacy. But to wrap up: I think we'd both like to increase the legitimacy of elections, but spending resources (whether that's time, money, or something more abstract like system complexity or fragility) on a cryptographic vote verification system is misguided because the US doesn't have fraudulent or incorrect election results, and there are no elections where such a system would have done better than the existing recount system. On the other hand, decades of political science study has discovered and reaffirmed time and time again that some simple policies would go a long way towards fixing: - bonkers disenfranchisement (via gerrymandering, intimidation, class-based barriers like long lines, etc.) rendering our representation completely non-descriptive (in terms of race, sex, gender, politics, immigrant status, ethnicity, sexual preference, class, you name it) - the majority of our representative bodies being elected with wildly meager portions of the vote, or highly partisan sections of the vote (in the case of very partisan districting), or both These are the things that truly threaten the legitimacy of our elections.

TheMonarchist 5y ago
> Not everyone will use a token or remember it. If even 50% of voters do this 100% accurately, it still isn't statistically helpful. I think this is its main weakness. Those who don't care enough to use a token would leave the token of zero effort (nothing) open for manipulations. Unless those are few, this system won't make the election more secure. It would however give those who care a way to ensure that their vote was counted, increasing public confidence in the results. For I assume that those who care enough to question the results also care enough to use a token and check it. > It's also worth saying that if there's any problem with this system, it calls the entire election into question, even if the election itself was fine or statistically far from error. > You might look up your vote and think "shenanigans!", and there won't be any way of looking at your original ballot to find out what happened. The token is only needed to identify a vote, so it could be dead simple, like a bitmap that could be a grid to be painted on paper ballot. It can be just added to existing systems. > And even if there were, you still wouldn't have any way of verifying that your vote was included in the count. How is querying the tokens used to vote what you voted, seeing your token in the list, and checking that the tokens make up the reported total not verifying it? > We (especially in tech) focus myopically on election fraud in the US, as though that's the thing that would undermine the legitimacy of an election. I think it's just because it seems like something that could be solved. > I am 100% sure that even if we implemented this system, they would still work just as hard to undermine election legitimacy. It would be a little harder when those who care can just check that their vote was counted. > I think we'd both like to increase the legitimacy of elections -- It seems to me that we have somewhat different aims. You have zeal for democracy and are willing to compromise ballot secrecy to verify the election result. I accept that verifying it is impossible while maintaing the secrecy, unless everyone is willing to check that his vote was counted. In practice I would be content if everyone just had that possibility, as I don't have unrealistic expectations of its popularity.
- camgunz 5y ago
  >> And even if there were, you still wouldn't have any way of verifying that your vote was included in the count. > How is querying the tokens used to vote what you voted, seeing your token in the list, and checking that the tokens make up the reported total not verifying it? There can still be two separate systems: the system that has your actual vote and the system we use to determine who won the election. One of the (many) issues I have with cryptographic vote verification systems is that they don't prove there isn't a "second set of books" that all checks out, meanwhile we use the crooked books to determine who won. Fundamentally, without a way to know all the possible voters and how they voted, you can't defend against that. >> We (especially in tech) focus myopically on election fraud in the US, as though that's the thing that would undermine the legitimacy of an election. > I think it's just because it seems like something that could be solved. Yeah but it's typical, and sorry for the bluntness here, tech arrogance--or maybe tech optimism is nicer. But either way, there's a consistent dynamic of tech people trying to solve things with tech that they really don't know a lot about. I've worked as a Democratic operative specifically on elections on the past few cycles and I'm pretty close to an expert. I'm not making an appeal to authority or anything, and I'm more than happy to talk about ways we can use tech to improve US elections, but it's just that I know these schemes are way, way more trouble than they're worth and either hinder a lot of the techniques we currently use to verify elections, or make political firestorms that make the election process even more fragile than it already is way more likely. It's also worth saying experts get this wrong too! The Iowa caucuses in 2020 had some tech hiccups, and that's, in my opinion, because no one can make sense of a caucus. They're beyond broken, and the more you measure them the more you'll find they're 100% unreliable and indicative of nothing. We certainly shouldn't be showering the person who wins by .5% with millions of dollars of donor money and earned media. But more broadly, it's also that elections are fundamentally messy, and the more you measure the more you find problems. Either we accept that and build solutions that accept that, or we don't and we live with constant allegations that our elections are fraudulent, courtesy of the perspective that we should be able to have 100% certainty. > It seems to me that we have somewhat different aims. You have zeal for democracy and are willing to compromise ballot secrecy to verify the election result. Honestly, I'm not sure what my ideal system is. I think anonymous voting is really important, but I also strongly support compulsory voting, and those two things are in direct tension. I think it's just hard to accept that elections are fuzzy measurements of sentiment and that the role of math in their verification is limited to some basic statistical analysis while the rest is process and politics. And I think it's really easy to be nerd sniped by the verification part, when tech actually could do so much good in the process and politics part. For example, a lot of States have "wet ink" statutes that mean voter registration forms have to be signed by a literal pen. So guess what Democrats did: they built a robot that holds a pen and signs a voter registration form remotely when someone fills out a voter reg form online. Hooray for tech! But also, that cost a lot of money and time. Wouldn't it have been easier if State legislatures just weren't assholes? Anyway, I would say if we're gonna implement a system that costs whatever the cheapest cryptographic verification system costs, we should just fund early voting programs and law enforcement initiatives against voter intimidation and misinformation. I know even that is impossible, which is depressing, but c'est la vie.