3 ms·

There is a major difference between trusting regular infrastructure and trusting elections infrastructure. The threat model is completely different. For one, vo

There is a major difference between trusting regular infrastructure and trusting elections infrastructure. The threat model is completely different. For one, vote secrecy makes it impossible to prove that your vote was mis-registered. For another, in normal situations the cyber-adversaries I need to worry about are regular criminals, for which basic computer security practices are good enough. But votes in an election are improtant enough that nation-state actors are a real threat, and I am absolutely certain that my phone or laptop are not secure against hacking attempts by a nation, or even potentially my own government! Not to mention, elections are the only situation where trust in my government is not fully possible: the current government has too much incentive to steal votes secretly. So, unlike the electrical grid, roads, financial infra and so on, I can't rely on implicit trust in the government to trust elections. > If someone was able to record a signature that their voting gateway signed vote for X as vote for X, but then later that same service claimed they voted for Y, they are caught red-handed cryptographically signing with the private key two conflicting votes. If. What if instead my voting app is showing me that my vote was correctly registered, and that key verification succeeded while the polls are still open, but once the polls close, it shows me that in reality the server registered the opposite vote? How do I prove to anyone else that I voted for X and the app was showing me I voted for X, but now the system shows I voted for Y, and that is what is recorded in the official counted results? This doesn't even require anyone breaking the encryption: the app can just show me a lie, or some malware can intercept the display info and show a different result, etc. To not leave any trace, the malware even deletes itself from the system as soon as the election timeout expires. Or maybe I am just lying and nothing wrong happened: I voted for Y, the system recorded I voted for Y, and now I'm just trying to cast doubt. Same as a paper based election, anyone can claim anything, and it's exactly as impossible to prove one way or the other. Plus, again, only a very very very small group of people can actually confirm for themselves that all of these complicated crypto algorithms actually do what they promise to do. Especially when looking at the end to end system. I for one am certain I couldn't verify for myself that all steps of such a system is secure. I would bet you can't either. If, say, Ron Rivest (of RSA fame) came out and said the cryptography used in the election is broken, while Adi Shamir (same) said it isn't, I would have no way to be certain which is right, and even if I tried to verify the math myself, I wouldn't trust myself as much as either of them.

EGreg 2y ago
You’re still misunderstanding how it works. First of all, blockchains are a third party ledger, which is maintained by many independent nodes are the large ones are infeasible to corrupt by a nation-state. The attack could only happen at the point where you sign transactions for the smart contracts. Voters are required to use at least 2 devices, such as scanning a QR code on their laptop (which runs Chrome) with their phone (which runs Safari). The QR code contains (or points to) a vote that is cryptographically signed by one gateway. The website or app on the phone checks this QR code and displays the same result back to you, and you confirm it on eg your phone’s screen. On both gateways (call them Services A and B) you indicated your preference, and digitally signed it, not just with your own key, but there is attestation by the device’s own private key, which is derived from the vendor’s key, meaning the vendor stands behind what their device or app does. Let’s assume absolutely every signature service cannot be trusted, including all your crypto wallets, incouding Apple’s secure enclave, everything is designed to be sleeper agent to mislead you on the day of the election. They just really want to change everyone’s vote. You can still prove which services were corrupted! Let’s say that you got signature Service A and Service B to sign two different candidates during the same chain of QR code confirmations. The proof is there that at least one of the services was corrupt. Even if it happened only once, with one voter. The indelible proof is on a blockchain and replicated so nation states can’t hide it. So no Service would agree to volunteer such a blatant proof of its own corruption, given the cost to its vendor. It would only happen if the Service would be hacked by an employee of the vendor, and that would only hurt the vendor, not the election. The vendor would try to eliminate this possibility as much as possible. However, if service A one lied to you, and you found out after scanning the QR code with Service B, then you wouldn’t want to submit your faulty vote with service B when it revealed that to you. But the service B would already would have provable dirt on service A. Not conclusive, since the voter could after all be someone who would rather complain about a non-faulty system than vote. I won’t speculate on chances of many registered voters not wanting to vote but simply make up fake complaints about the system, but I don’t think regular users should face penalties for lying, so I’ll just accept this as a serious possibility. All I will say is, these people are similar to those who stay out and don’t vote now. It’s an issue of “turnout”. But even in this scenario (of a malicious voter rather than malicious service), Service B would then be required to do the reverse — process your other vote, and sign the transaction, then anonymously submit it to Service A to be signed. Service A would have to either refuse to cooperate with Service B, or sign it. After that, you’d be given a QR code presented by Service B, and verify it with Service A. Of course there could be far more than just a of Services A and B. There could be 100 services (eg web-based) and voters could be required to go through a chain of 3 of them, as determined from a random oracle (ie they don’t get to pick who to collude with). You’d get the list of 3, and an honest service would simply redirect you to the next one as you bounce between two devices via QR codes. Service A on Laptop Service B on Phone Service C on Laptop - done Everything that’s signed goes into a third-party gossipped / replicated log (doesn’t have to be a blockchain, there doesn’t have to be a total order). This log / heap is what contains the indelible proofs that can be found out anytime after the fact, which is why every service must be careful to mess up even once. You see, there is a huge difference between actors/nodes simply voting between some arbitrary choices A and B, and nodes voting while also following cryptographic constraints amd creating a trail where cheating at any step can be caught and proven later. The latter is much harder to pull off and, given costly enough consequences, creates chilling effects and strong incentives to be honest. This is what many BFT algorithms get wrong and why they fail in the presence of over 33% malicious nodes. https://youtu.be/BYRTvoZ3Rho?si=AGbuwZlJ85G3KXPg https://youtu.be/BYRTvoZ3Rho?si=AGbuwZlJ85G3KXPg

EGreg 2y ago

You’re still misunderstanding how it works. First of all, blockchains are a third party ledger, which is maintained by many independent nodes are the large ones are infeasible to corrupt by a nation-state. The attack could only happen at the point where you sign transactions for the smart contracts. Voters are required to use at least 2 devices, such as scanning a QR code on their laptop (which runs Chrome) with their phone (which runs Safari). The QR code contains (or points to) a vote that is cryptographically signed by one gateway. The website or app on the phone checks this QR code and displays the same result back to you, and you confirm it on eg your phone’s screen. On both gateways (call them Services A and B) you indicated your preference, and digitally signed it, not just with your own key, but there is attestation by the device’s own private key, which is derived from the vendor’s key, meaning the vendor stands behind what their device or app does. Let’s assume absolutely every signature service cannot be trusted, including all your crypto wallets, incouding Apple’s secure enclave, everything is designed to be sleeper agent to mislead you on the day of the election. They just really want to change everyone’s vote. You can still prove which services were corrupted! Let’s say that you got signature Service A and Service B to sign two different candidates during the same chain of QR code confirmations. The proof is there that at least one of the services was corrupt. Even if it happened only once, with one voter. The indelible proof is on a blockchain and replicated so nation states can’t hide it. So no Service would agree to volunteer such a blatant proof of its own corruption, given the cost to its vendor. It would only happen if the Service would be hacked by an employee of the vendor, and that would only hurt the vendor, not the election. The vendor would try to eliminate this possibility as much as possible. However, if service A one lied to you, and you found out after scanning the QR code with Service B, then you wouldn’t want to submit your faulty vote with service B when it revealed that to you. But the service B would already would have provable dirt on service A. Not conclusive, since the voter could after all be someone who would rather complain about a non-faulty system than vote. I won’t speculate on chances of many registered voters not wanting to vote but simply make up fake complaints about the system, but I don’t think regular users should face penalties for lying, so I’ll just accept this as a serious possibility. All I will say is, these people are similar to those who stay out and don’t vote now. It’s an issue of “turnout”. But even in this scenario (of a malicious voter rather than malicious service), Service B would then be required to do the reverse — process your other vote, and sign the transaction, then anonymously submit it to Service A to be signed. Service A would have to either refuse to cooperate with Service B, or sign it. After that, you’d be given a QR code presented by Service B, and verify it with Service A. Of course there could be far more than just a of Services A and B. There could be 100 services (eg web-based) and voters could be required to go through a chain of 3 of them, as determined from a random oracle (ie they don’t get to pick who to collude with). You’d get the list of 3, and an honest service would simply redirect you to the next one as you bounce between two devices via QR codes. Service A on Laptop Service B on Phone Service C on Laptop - done Everything that’s signed goes into a third-party gossipped / replicated log (doesn’t have to be a blockchain, there doesn’t have to be a total order). This log / heap is what contains the indelible proofs that can be found out anytime after the fact, which is why every service must be careful to mess up even once. You see, there is a huge difference between actors/nodes simply voting between some arbitrary choices A and B, and nodes voting while also following cryptographic constraints amd creating a trail where cheating at any step can be caught and proven later. The latter is much harder to pull off and, given costly enough consequences, creates chilling effects and strong incentives to be honest. This is what many BFT algorithms get wrong and why they fail in the presence of over 33% malicious nodes. https://youtu.be/BYRTvoZ3Rho?si=AGbuwZlJ85G3KXPg https://youtu.be/BYRTvoZ3Rho?si=AGbuwZlJ85G3KXPg