Sierra Leone government denies the role of blockchain in its recent election

There's also people providing their public address as proof of a vote in exchange for something, but the risk for the buyer is that someone just hands over someone else's transaction and votes as they please.

A real potential solution would be to make the votes private and "untraceable" (A la Monero ring signatures or something of the sort), but that kind of comes with it's own pros and cons and could always be broken later. I still think it's worth investigating if this is more secure than plain old paper ballots.

Yes they are. That's not what 'immutable' means. Immutability refers to an inability to change the past. Not an inability to fork the future. A blockchain is a persistent, immutable data structure. This is just a simple fact of computer science.

That being said, this particular project is not a good example of any of that. But blockchain for elections is actually one of the few truly technically legitimate uses of the technology.

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.

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.

Don't get me wrong - there's a ton of hype around blockchains that is completely unjustified. But there are a few applications where it may have some utility, and I think this is one of them. So it irritates me when I see people on HN jump on the "blockchain therefore hype" bandwagon without spending the time to actually think about when this sort of technology is useful and when it isn't.

A decentralized blockchain is far more trustworthy than a centralized database in this scenario. Say Russia controls the centralized voting database, they can just edit whatever value they want to say "Putin" and call it a day. It's not public. No one can verify their vote. It's under the control of one party. The entire point of decentralization is to ensure there is public consensus among multiple parties.

There are other discussions about how to allow or incentivize citizens to run nodes, mitigate lost or stolen IDs/private keys, prevent against spam attacks, etc...but it sounds like you aren't even open to discussing those because you equate the word "blockchain" with "snake oil".