4 ms·

I'd say a significant amount of our problems are not transaction log failures - they are authentication/identity failures. The catch is that blockchain technolo

by spydum 9y ago

I'd say a significant amount of our problems are not transaction log failures - they are authentication/identity failures. The catch is that blockchain technologies force you to solve both (you cant take part in a transaction without an identity, can you?). Plenty of people who've lost wallets will concede this is still hard (what happens when you lose your government issued private-key, for example?).

darawk 9y ago
I agree that blockchains do not solve all of the problems with elections. Here's what blockchain can offer, assuming you have a cryptographic identity mechanism (which has its own problems, of course): 1. Any person can validate that their vote was counted. 2. Any person can validate that their voted counted towards their intended candidate. 3. The collective can validate that no person (i.e. no crypto-id) voted twice. 4. The collective can validate the exact true number of tallied votes (and, e.g. make sure it doesn't exceed the total number of eligible voters). 5. The collective can validate the tallying process. I.e. they can count all the votes on the chain, and themselves reproduce the sums. That does not address the following: 1. Ensuring that a crypto-id was used by the appropriate person (Although if someone stole your ID, this would likely be discovered by the person it was stolen from, due to previous property #2). 2. Ensuring that a person was not coerced to use their crypto-id against their will. Which are real problems that I would love to see solutions to. But that doesn't mean we shouldn't try to solve those first 4 as well.
- Nursie 9y ago
  >> 2. Any person can validate that their voted counted towards their intended candidate. This is a bad idea even in concept. If anyone can validate their vote went for the candidate of their choice, someone else can force them to perform the same validation for money, under threat of violence, familial ostracism, or whatever really. I get that you understand there is a problem with coercion here, but it's not just an implementation problem, it's a problem with the concept.
  - darawk 9y ago
    > This is a bad idea even in concept. If anyone can validate their vote went for the candidate of their choice, someone else can force them to perform the same validation for money, under threat of violence, familial ostracism, or whatever really. I responded to this issue in the previous thread about this. But short answer is: create a verifiable but also deniable scheme. https://news.ycombinator.com/item?id=16598591 https://news.ycombinator.com/item?id=16598591
    - acdha 9y ago
      That’s a proposal, not a proven solution, and it assumes that the chosen algorithm, implementation, and operation are all perfect or it fails open. That’s why you need to look at this as a system rather than a chance to use the current fad: this class of proposal replaces something which is cheap, robust, easy to understand, and fails closes with something complex and unproven which most people would cannot make accurate threat assessments about. There’d have to be a huge win to make up for that and “it costs a lot more and my startup could cash in” doesn’t count.
- jacques_chester 9y ago
  Goals 1 through 5 are addressable by using election scrutineers[0], without introducing problems 1 and 2. I like crypto as much as the next person. But the bigger picture means that simpler, dumber systems that rely on lots and lots of humans are more suited to the problem space. [0] http://www.aec.gov.au/Voting/scrutineers.htm http://www.aec.gov.au/Voting/scrutineers.htm
  - darawk 9y ago
    Problems 1 and 2 exist just as much with scrutineers as they do with a blockchain solution.
beat 9y ago
Why is authentication/identity failure an actual problem in practice, that can't be solved with simple good process? Despite misleading propaganda by a certain American political party, it is simple and straightforward to reduce voter identity fraud to irrelevance. Let's start with imagining a statewide election with a million votes. To move the results by 1% would require 10,000 fraudulent votes. Now, how do we get 10,000 fraudulent votes? First, we could stuff the boxes with fake ballots. Second, we could have false voter registrations and send "professional voters" around to vote at multiple polling places. Third, we could alter the results at the count, by giving a false count. Now, let's introduce a process step: all activities that involve touching ballots, whether marked or unmarked, must be done in the presence of representatives of at least two political parties. In other words, if Team A is going to try to physically manipulate ballots or counts, they have to do it where Team B can watch them do it. How does that affect things? And there's the process of running a polling place. At the beginning of the day, you get X number of blank ballots. They were delivered by two parties, and counted by two parties, and are presumably under some sort of seal. We know how many ballots are in the building. And there's a paper voter registration roll, with the names of every registered voter in the precinct. When a voter comes in, they look up the name on the roll, and the voter signs to say they voted. A tally is kept of how many voters sign. Each voter then gets a ballot, marks the ballot, and puts it in the box. At the end of the day, count the number of ballots. It should match the number of voters on the tally, and can be hand-checked against the number of signatures on the roll in a recount. Spot-checks of this at random precincts prevent widespread alteration without triggering a recount. The number of votes plus the number of remaining blank ballots equals the number of blank ballots at the beginning. So inserting fake ballots is hard. Inserting them in a way that can't be detected at count time or at recount time is harder. So, how about false registrations and professional voters going around making multiple votes in different precincts? Remember, we need 10,000 votes to move the tally 1% in our theoretical election. Let's say one professional voter can make it to ten polling places over the course of a day. So you would need a thousand of them, trained, with targets. They would need to be either fanatical, or paid. And you need to do this with no leaks, no turncoats, no one who thinks "I could sell my story to the news for tens of thousands", no hidden cameras... yeah. (As an aside, this is why insistence on voter ID is misplaced. The difficulty of the problem prevents it far more than an ID step would.) So how about altering the count? Well, if it involves altering the ballots themselves, it's a hard problem. (Unless you use those idiotically dangerous touchscreen machines, which ought to be illegal.) You need to remove valid ballots and replace them with altered ones, again with multiple parties observing. Or you can alter the final result count, but then it mismatches the paper ballots, and would be detected as fraudulent at recount time. Spot check hand-counts detect such issues. Technical implementation example here: In Minnesota, we use Scantron machines. If there are multiple Scantrons at a polling place, we will hand-count one of them and compare to its electronic tally. The electronic tally from the machine is an unofficial result only, until certified. Now, let's look at some easy things to do to manipulate elections! First, gerrymandering - draw district lines that make it nearly impossible for one party to win. I live in such a district. We're more likely to get visited by aliens than elect a Republican in my district. Second, voter suppression. Make it increasingly more difficult to vote - limit polling station resources in certain districts to force multi-hour lines, require ID, etc. These are popular and common. See how this works? Blockchain doesn't fix ANY of this. Either it's already fixed, or could be fixed by improving the process based on models that are demonstrably good in other states. Heck, they do fair elections in poor countries with high illiteracy rates, too. Just stamp each voter's hand when they vote. No one can vote twice because stamps are checked. Ballots can have symbols for parties rather than words. Election validation is easy.

by spydum 9y ago

darawk 9y ago

I agree that blockchains do not solve all of the problems with elections. Here's what blockchain can offer, assuming you have a cryptographic identity mechanism (which has its own problems, of course): 1. Any person can validate that their vote was counted. 2. Any person can validate that their voted counted towards their intended candidate. 3. The collective can validate that no person (i.e. no crypto-id) voted twice. 4. The collective can validate the exact true number of tallied votes (and, e.g. make sure it doesn't exceed the total number of eligible voters). 5. The collective can validate the tallying process. I.e. they can count all the votes on the chain, and themselves reproduce the sums. That does not address the following: 1. Ensuring that a crypto-id was used by the appropriate person (Although if someone stole your ID, this would likely be discovered by the person it was stolen from, due to previous property #2). 2. Ensuring that a person was not coerced to use their crypto-id against their will. Which are real problems that I would love to see solutions to. But that doesn't mean we shouldn't try to solve those first 4 as well.

Nursie 9y ago
>> 2. Any person can validate that their voted counted towards their intended candidate. This is a bad idea even in concept. If anyone can validate their vote went for the candidate of their choice, someone else can force them to perform the same validation for money, under threat of violence, familial ostracism, or whatever really. I get that you understand there is a problem with coercion here, but it's not just an implementation problem, it's a problem with the concept.
- darawk 9y ago
  > This is a bad idea even in concept. If anyone can validate their vote went for the candidate of their choice, someone else can force them to perform the same validation for money, under threat of violence, familial ostracism, or whatever really. I responded to this issue in the previous thread about this. But short answer is: create a verifiable but also deniable scheme. https://news.ycombinator.com/item?id=16598591 https://news.ycombinator.com/item?id=16598591
  - acdha 9y ago
    That’s a proposal, not a proven solution, and it assumes that the chosen algorithm, implementation, and operation are all perfect or it fails open. That’s why you need to look at this as a system rather than a chance to use the current fad: this class of proposal replaces something which is cheap, robust, easy to understand, and fails closes with something complex and unproven which most people would cannot make accurate threat assessments about. There’d have to be a huge win to make up for that and “it costs a lot more and my startup could cash in” doesn’t count.
jacques_chester 9y ago
Goals 1 through 5 are addressable by using election scrutineers[0], without introducing problems 1 and 2. I like crypto as much as the next person. But the bigger picture means that simpler, dumber systems that rely on lots and lots of humans are more suited to the problem space. [0] http://www.aec.gov.au/Voting/scrutineers.htm http://www.aec.gov.au/Voting/scrutineers.htm
- darawk 9y ago
  Problems 1 and 2 exist just as much with scrutineers as they do with a blockchain solution.

beat 9y ago

Why is authentication/identity failure an actual problem in practice, that can't be solved with simple good process? Despite misleading propaganda by a certain American political party, it is simple and straightforward to reduce voter identity fraud to irrelevance. Let's start with imagining a statewide election with a million votes. To move the results by 1% would require 10,000 fraudulent votes. Now, how do we get 10,000 fraudulent votes? First, we could stuff the boxes with fake ballots. Second, we could have false voter registrations and send "professional voters" around to vote at multiple polling places. Third, we could alter the results at the count, by giving a false count. Now, let's introduce a process step: all activities that involve touching ballots, whether marked or unmarked, must be done in the presence of representatives of at least two political parties. In other words, if Team A is going to try to physically manipulate ballots or counts, they have to do it where Team B can watch them do it. How does that affect things? And there's the process of running a polling place. At the beginning of the day, you get X number of blank ballots. They were delivered by two parties, and counted by two parties, and are presumably under some sort of seal. We know how many ballots are in the building. And there's a paper voter registration roll, with the names of every registered voter in the precinct. When a voter comes in, they look up the name on the roll, and the voter signs to say they voted. A tally is kept of how many voters sign. Each voter then gets a ballot, marks the ballot, and puts it in the box. At the end of the day, count the number of ballots. It should match the number of voters on the tally, and can be hand-checked against the number of signatures on the roll in a recount. Spot-checks of this at random precincts prevent widespread alteration without triggering a recount. The number of votes plus the number of remaining blank ballots equals the number of blank ballots at the beginning. So inserting fake ballots is hard. Inserting them in a way that can't be detected at count time or at recount time is harder. So, how about false registrations and professional voters going around making multiple votes in different precincts? Remember, we need 10,000 votes to move the tally 1% in our theoretical election. Let's say one professional voter can make it to ten polling places over the course of a day. So you would need a thousand of them, trained, with targets. They would need to be either fanatical, or paid. And you need to do this with no leaks, no turncoats, no one who thinks "I could sell my story to the news for tens of thousands", no hidden cameras... yeah. (As an aside, this is why insistence on voter ID is misplaced. The difficulty of the problem prevents it far more than an ID step would.) So how about altering the count? Well, if it involves altering the ballots themselves, it's a hard problem. (Unless you use those idiotically dangerous touchscreen machines, which ought to be illegal.) You need to remove valid ballots and replace them with altered ones, again with multiple parties observing. Or you can alter the final result count, but then it mismatches the paper ballots, and would be detected as fraudulent at recount time. Spot check hand-counts detect such issues. Technical implementation example here: In Minnesota, we use Scantron machines. If there are multiple Scantrons at a polling place, we will hand-count one of them and compare to its electronic tally. The electronic tally from the machine is an unofficial result only, until certified. Now, let's look at some easy things to do to manipulate elections! First, gerrymandering - draw district lines that make it nearly impossible for one party to win. I live in such a district. We're more likely to get visited by aliens than elect a Republican in my district. Second, voter suppression. Make it increasingly more difficult to vote - limit polling station resources in certain districts to force multi-hour lines, require ID, etc. These are popular and common. See how this works? Blockchain doesn't fix ANY of this. Either it's already fixed, or could be fixed by improving the process based on models that are demonstrably good in other states. Heck, they do fair elections in poor countries with high illiteracy rates, too. Just stamp each voter's hand when they vote. No one can vote twice because stamps are checked. Ballots can have symbols for parties rather than words. Election validation is easy.