Hacker News Reader

Comment by eru | Hacker News Reader

The article says:

> “There’s a strong desire to see how ballot counting machines are actually counting the ballots,” New Hampshire’s Secretary of State, David Scanlan, told Click Here in an interview. “And open-source software really is the only way that you can do that effectively.”

In any case, I don't think voting machines are a good idea.

I know that you can use enough cryptography and enough open sourcing to make them secure, but what might convince me wouldn't convince the 'village idiot'.

Paper voting and traditional election observers work in a nice 'analog' way. It's tangible, and even a 'village idiot' can understand how it works.

Most voting systems aren't complicated enough to need a computer to evaluate them.

(And those that are should arguable be changed.)

View on HN

IMO secure isn't enough. You also need to make it auditable. As far as I can see, it's pretty much impossible to do this satisfactorily with a digital solution. It also seems entirely unnecessary. Manual counting systems work well, and allow people to feel like they are participating in the democratic process (because they are).

Ballot boxes that scan ballots are auditable because they keep the paper ballots.

In my state, election workers have to ensure that the number of ballots is the same as the number recorded by the ballot box on election night - although we don’t recount who voted for who. The ballots are preserved.

Most states use paper ballots that are then run through electronic tallying machines. This process is completely auditable. This is how people lying about the 2020 election were able to do all of their sham audits in Arizona and Wisconsin.

> This process is completely auditable

Presumably the auditing process would involve tallying the votes manually... in case why bother with the electronic process? Unless you only run the audit if there's a complaint. Which I guess works. But personally for something as important as elections I would much more comfortable with a system where the transparent, hard-to-corrupt process is used by default.

You are missing the part that somebody could add additional ballots which is a claim that is being made. Supposedly there was a water leak and then people brought in some extra ballots.

Electronic ballot-tallying machines are necessary to running accurate and speedy elections. As we saw in the Arizona fiasco with the "Cyber Ninjas," hand-counting ballots takes weeks, if not months. Additionally, hand-counting has an error margin of 1-2%. In a era where many races are within a point or two, hand-counting ballots will lead to the wrong person winning office.

We also have tons of checks in place to make sure the machines work. There are calibration runs where known stacks of votes are sent through the machines and confirmed at the end. The vote totals on the machines are checked against the number of people who voted. Post-election, election officials randomly choose machines, open them up in front of the public at preannounced times, and confirm that the ballots that have been locked inside the machine match the tally the machine produced.

In the UK all ballots are hand counted, and the counting is done within a day.

I would expect that the amount of time it takes to tally a paper election is related to how many people you have counting.

Same in Germany.

How many races are on your ballot? Next Tuesday, I'll have 11 races to vote for, plus 4 ballot referendums. That's probably a median election ballot in the US.

In the UK? Typically just one race: you vote to elect your local MP.

You might also vote for someone to represent you in the devolved Welsh/Scottish/Northern Irish governments if you live in those countries, but that's only 15% of the population and those elections aren't necessarily on the same day as the general election.

In the UK you usually have 1-3, fewer since we left the EU. However, the number of ballots is relatively immaterial for two reasons:

Spend: In the UK, price of counting ballots is far lower than price of paying poll booth staff. It costs approximately £0.10 per person (note, not per ballot) in the UK to count ballots in a nationwide election (as opposed to about £0.35 on keeping polling booths open). It’s likely that if the number of ballots went up by a factor of 10, further efficiencies would be found, but simple ‘ballots counted per hour’ metrics imply that the marginal cost should be relatively affordable.

Difficulty: While running a fair and free election is undeniably challenging, it naively would appear to be much less work than E.g. the USPS does each day to deliver the daily 300M pieces of mail.

While the USPS is a government department, it is allowed to charge its users and does so.

The people running elections don't charge voters for voting.

So I'm not sure it's useful to compare the two?

(That the Americans feel like they need government mail when the supposedly more socialist Europeans mostly managed to privatise their former monopolists probably tells us.. something?)

> The vote totals on the machines are checked against the number of people who voted.

Isn't that just checking that "x == x" in software if everything is done by machine?

Nope. The people running the election sign every voter in uaing a system independent of the voting machine (often paper). At the end of the day, they know that 216 people voted at that precinct. If the counting machine doesn't say 216 votes on it, something is awry and must be investigated.

Couldn't the person just signing me in record my vote on paper & then add it to a tally?

1. That's not a secret ballot.

2. The typical US ballot has a dozen or so different votes that need to be done.

I disagree. While I do agree that hand counting paper ballots is extremely time consuming and error prone, it's not the only alternative to an electronic voting machine. The other alternative is a mechanical voting machine. No electronics and no software involved. I grew up in south Louisiana and this is the kind of voting machine that was used when my parents went to vote. The downside is that the machines were big and heavy. The voter steps inside the curtains and pulls a big lever that closes the curtains behind them and makes the machine ready for voting. The voter physically moves a lever for each choice. When they're done voting, they pull the big lever in opposite direction and it tallies their votes, returns the levers to their original positions, and opens the curtains. I don't know the exact procedure of reporting the polling station results, but I'm sure it required multiple individuals (and maybe even a supervisor) to read off the results from each voting machine. Even if you have a polling station with say 20 voting machines, that's just 20 additions per voting choice that must be summed. This is much easier and faster than hand counting paper ballots. Additionally, I think it would be nearly impossible to alter the voting machines in an attempt to steal votes. I would like to see this everywhere. Voting integrity and quick results.

What I had in mind when thinking about mechanical voting machines would have been punched cards and 19th century style machines for sorting and counting.

I’m curious, do you distinguish between ballot counting machines (what the NH SoS is talking about) and voting machines (your term)?

I understand “voting machine” to be any mechanism that acts as a substitute for a paper ballot, or that acts as an intermediary between you and a paper ballot. And I would agree that the former are a terrible idea and the latter probably don’t provide much net benefit.

But ballot counting machines are a different matter. Here’s the thing: there’s no voting process without a ballot counting machine. It’s just a matter of whether that machine will be a carbon-based one subject to bias, fatigue, boredom and hunger, or an electronic one that excels at stupid-simple repetitive tasks and is subject to none of those things.

There's a third option: a mechanical ballot counter.

I know that human beings are fallible. That's why I wrote:

> I know that you can use enough cryptography and enough open sourcing to make [electronic machines] secure, but what might convince me wouldn't convince the 'village idiot'.

People are used to the failure modes of other humans. And they trust other humans in a way they don't trust machines.

This is similar to how juries still put a lot of stock in eye witnesses.

How is “mechanical ballot counter” a third, distinct option from “ballot counting machine”? Are you saying that if it had no electronics and operated solely by a human turning a crank on a bunch of gears, springs and levers, people would see it as more reliable and trustworthy?

> Are you saying that if it had no electronics and operated solely by a human turning a crank on a bunch of gears, springs and levers, people would see it as more reliable and trustworthy?

Potentially, yes. Especially if you used off-the-shelf machines that IBM built in 1920.

In any case, that's just a distraction. Just use humans.

I live in New Hampshire, and though I've never voted on a computer where I'm touching a screen, I still feed my paper ballot into a scantron machine that seems to do all the counting. I’ve voted in maybe 20 elections (between local, state, and federal) in my lifetime, and it's always been done this way around here. Maybe we all forget that that's a computer, too.

All that said, I live in a town of less than 3,000 people. I don't know that the solutions that work for us here would work everywhere.

Ranked choice voting systems (e.g. single transferable vote) have a lot of "game theory" advantages over e.g. first-past-the-post, but are hard to implement with manual counting.

Edit: I suppose that doesn't mean voting machines are required though, just that the results on each ballot paper need to be inputted to a computer to determine the outcome. A lot of data inputting required, but still auditable.

First-past-the-post is pretty horrible. Alas, experience from Australia shows that single transferable vote isn't any better in practice, either. It's more complicated with no benefit in practice. (See https://rangevoting.org/rangeVirv.html for more details on Australia.)

If traditional voting is required, I have higher hopes for approval voting or the slightly more complicated range voting. These are quite easy to evaluate completely by hand. Especially the former.

If you are happy to go a bit off the beaten track, using sortition amongst volunteers might be a good way to fill up a parliament. (Or any other body big enough for the law of large numbers to kick in.)

As a slightly more complicated system, I would have every voter indicate their favourite candidate on a ballot sheet, and then draw enough votes at random to fill up parliament.

(Use your favourite mechanism to handle the same candidate being voted for multiple times in your sample. Perhaps give them more weight, or re-draw, or have people write down an ordered list of preferred candidates, and admit the highest ranked one who isn't already in parliament, etc.)

Random sampling is surprisingly powerful. It also completely bypasses Arrow's impossibility theorem.

Does it ?! Wow, that is tempting...

Well, Arrow's theorem only applies to deterministic procedures.

Yes, but there's literally no way to prove that a random number that you "generated" is actually random. The best idea would be to write people's names on balls, put them in a clear funnel thing, and blow them around or something like that on live TV, but for sortition, I'm not sure how that would scale. In the movie Contagion they did that for each day of the year someone could be born on, and that's how they distributed the vaccines. Not sure how that would scale to millions of balls and how you would audit that each person's name is on one ball and only one ball.

Yes, building a purely mechanical trustworthy source of randomness is hard.

However, you can look at how the lottery does it, perhaps?

For example, they don't write the players names on the balls. They use more indirection. (And lots of other cleverness.)

Also keep in mind that I suggest to use this mechanism for something like filling up a parliament. The German Bundestag has about 600 members. The UK has 650 MPs in their House of Commons.

I posit that getting into parliament is perhaps comparable to winning the lottery, ie something people might trust lottery equipment and procedures to handle.

Becoming the president of the US is a bigger deal than winning the lottery, so we can't naively expect lottery equipment and procedures to be above suspicion.

Reply on news.ycombinator.com