Hacker News Reader

Comment by captn3m0 | Hacker News Reader

I'm never trusting software for two things:

1. Voting

2. Irreversible transactions

Don't be naive - the transaction is definitely reversible. It's just that enough Important people need to agree it is reversible. If it was an Important person who lost money, or the loss was big enough to be bad PR, they would hard fork the chain and pretend like the transaction never happened.

No it's irreversible. Ethereum had one hard fork in its history (the DAO hack of 2016) and it was contentious and caused a split in the community. It can't be done Willy nilly.

Wallets should do more to prevent these types of sends from occuring, providing warnings. But you should also never ever send $50K without a test transaction of some token amount first.

(2) the transactions are irreversible, and the software is clearly doing that well.

The problem is people don't actually want irreversible transactions, they've just been told that they do.

Yes, they do. That is a feature, not a bug. It gives digital transactions the same properties as cash. When you hand over $50K in cash to someone, you better make sure you trust that someone. When you send $50K in crypto to an address, you better make sure you trust that address. People have to learn new habits and precautions with crypto. With great power comes great responsibility.

Maybe the UI for wallets should do more to defend against erroneous sends, or warn the user to make a test transaction before sending large amounts? A lot more can be done with UX.

No, people want reversible transactions.

They want transactions where they pay someone, and that person doesn't send the goods they can reverse the charge.

They want to be able to correct erroneous or fraudulent transfers.

These are all things that the normal financial system does well, and crypto systems absolutely cannot.

You cannot force someone to return a crypto currency transfer. No matter what the case: error, fraud, or theft. Even the latter is feasible within the regular financial system.

When you hand someone cash you can get the cash back. You know the person you can chase them down.

No, you can't necessarily get it back. It's like sending cash internationally, but you sent it to the wrong address and now you don't know who the fuck has it.

If you mis-address cash sent internationally it does get held for reversal. That's why international wires require so many details.

A wire is not "cash." Cash, in this conversation, refers to paper and metal transfered physically.

I suppose I'll be downvoted for this but my unpopular opinion is that I feel like the voting problem is solvable with some combination of decentralized and cryptographic tech. It's just that the people who can solve it don't want to solve it.

The thing about voting is that for the public to trust voting, it needs to be something a general member of the public can understand.

That is why paper balloting is such a durable voting mechanism. Pretty well anyone can understand how you would take a stack of paper ballots and decide who got the most votes. From there you just need to maintain a chain of custody where all ballots are monitored by any interested parties from the time they are cast to the time they are counted.

The more technical you get with your voting solutions, the less overall trust the public will have in the final vote counts.

I understand the argument here, but to play devil's advocate...

In the 2020 presidential election, tens of millions of people had trouble understanding how paper ballots worked. I'm not sure the argument about public trust is that durable or cogent at this point.

Most of the arguments I saw were about the chain of custody being broken and allegations that ballots were tampered with at that point. That is different from not understanding how a paper ballot works.

Most people who claim that there was fraud in the election focus on the voting machines — that they were programmed to "flip" votes, there were servers in Germany, or they had some linearly "weighted" mode which was allegedly turned on. They claim that "the algorithm" caused the election to be stolen from Trump.

If we created a cryptographic voting system with homomorphic encryption to preserve privacy, they would be claiming the exact same thing — except in that case, you wouldn't be able to recount the ballots by hand.

I think paper + blockchain. Paper ballot with anti-counterfeit tech (maybe fingerprint requirement as well for uniqueness check), scanned with an image recorded on blockchain (or hash of image stored on chain), then I think ballot images should be publicly viewable (with personal info redacted) so anyone can audit.

Its an active research area, there are proposals like helios https://en.m.wikipedia.org/wiki/Helios_Voting which check some of the boxes but aren't perfect . I'm sure cryptographers would love to come up with a fully scalable, trustless, verifiable, anonomous, coercion resistant system. But that's a hard set of properties to satisfy especially if nothing is in the physical world.

If you allow physical voting places, then just make a machine that gives a paper recipt and do risk limiting audits. Problem solved.

No, scan the full paper ballots with personal info redacted and make them publicly accessible so anyone can count the result.

So you don't trust the machines to scan the ballots to count them, but you do trust them enough to scan the ballots so that the public can count them? If the machine is compromised the scanning step will be compromised.

Personally i think risk limiting audits of the original physical documents is a much more secure system.

I mean, it's been already solved with a decentralized, low-cost, secure technology. It's called paper, and it has this wonderful security property that the amount of work needed to subvert the result is proportional to the number of votes you need to change, whatever you do.

Many people have tried to solve it in the past and all of the attempts ended in failure. It appears that it is non-trivial to have both secret voting and at the same time make it possible for the voters to prove that the election was not rigged.

I'm not sure it is solvable. Voting has to satisfy two requirements:

1) Votes must be private and anonymous (i.e. not linked to your identity)

2) Voting must be secure

It's easy to satisfy either one of these requirements, but currently impossible to satisfy both.

Pieces of paper in big boxes work pretty well, especially when they are enclosed in a sealed envelope with return address information for verification in case of fraud accusations.

Having the return address information on voting envelopes is a simple way to destroy deniability. Now you can prove who voted X if you have access to the envelopes.

I would argue that the paper voting doesn't satisfy both conditions either. And while putting a paper in a box is vividly understandable to the average voter, whatever happens to the vote after that point isn't, as evident by the number of people believing that a Texas lawsuit may overturn Michigan election results.

That's because there's video evidence of Fulton County Georgia poll workers clearing out all poll watchers on the basis of a bullshit pipe leak story and then as soon as everyone left they brought out suitcases of ballots and started scanning them in. There's 1000 affidavits of egregious election fraud and abuse in this election. The Supreme Court's refusal to even hear Texas's case is outrageous.

Define 'secure'?

Seems fairly obvious and self explanatory to me.

A secure vote is one that has a result that cannot be altered once it has been cast.

Reply on news.ycombinator.com