Hacker News Reader

Comment by shaki-dora | Hacker News Reader

Excellent! It used to be that the tech community was the most vocal against any scheme to increase the use of tech in voting. That consensus seems to have weakened.

So to all the blockchain-proponents etc: remember that it’s not enough for elections to just be safe. They also need to be safe in a way that is obvious to non-cryptographers.

Many countries run elections on paper ballots and with public ally-observeable counting of votes. They do so in rural communities as well as cities with millions of voters. They rarely see voters waiting in line for more than a few minutes. And the costs are neglible in the grand scheme of things.

That is the gold standard for elections, and it’s actually damn easy to achieve.

View on HN

> Excellent! It used to be that the tech community was the most vocal against any scheme to increase the use of tech in voting. That consensus seems to have weakened.

How? Here's the main conclusion from the report

> The monitoring group came into the conclusion that online voting should not be introduced in general elections, because the risks involved are greater than the benefits.

I was commenting on the impression I get from reading comments here, on Twitter, blog posts etc. When the topic used to come up, for example in response to the Florida fiasco in the election of 2000, the response was almost universally negative. That seems to have changed somewhat, with many more people willing to advocate complicated technical solutions.

This report, and others by professional organizations like the EFF still follow the old consensus, possibly because they are run by people socialized in that environment. But I’m worried what such reports will say in 2030 if the arguments I made above are not universally understood.

> Excellent! It used to be that the tech community was the most vocal against any scheme to increase the use of tech in voting. That consensus seems to have weakened.

What is excellent about this? Paper ballots, with multi-party observers are incredibly secure, auditable, and transparently fair ways of running an election.

Yes, you can have shitty ballot designs, and insane voting machines (Florida, 2000), and your entire electoral process may be infiltrated by Reptilians, who take the ballot boxes out into the woods, to use for kindling, but the same can be said for electronic devices. The UX, and security procedures for the Canadian elctions I have voted in was very simple.

Shade in the circle by the name of the candidate you want to vote for. Fold your ballot. Drop it in a box. At the end of the night, the box is opened, and the ballots get counted. There are eyes on the ballot box, for the entire duration of the process.

Adding a closed-source, opaque electronic voting machine to this process solves a non-problem that nobody has.

An alternative: Mark ballot, feed to scanner, which either successfully scans a ballot, or returns it to the voter to be re-marked (modulo write-in votes, see link). The scan is the fast & easy result, and the paper ballot can be the separate source of a reproducible result.

https://www.verifiedvoting.org/resources/voting-equipment/se...

Also a note for international HN readers - elections in the USA often have 20+ races on the same ballot - we really like elective offices - and that makes hand counting more difficult.

That's a complimentary tool, not an alternative. Most election integrity activists / experts are fine with precinct-based opscans used to tabulate paper ballots the moment the polls close.

The nice thing about scanning either by the voter or while the voter is present is that spoiled ballots - by stray, faint, or partial marks, or overvoting for too many multiple candidates - are reduced to near zero, and are replaced by fresh voter-marked ballots, which simplifies all later tabulation, and reduces confusion and challenges about mis-marked ballots, thus increasing voter confidence in the whole election process.

Anyone who is interested in election processes should volunteer to work an election, either helping operate a polling place, or as a poll watcher, or similar. This is one of those areas that looks simple, but has a lot of difficult edge and corner cases. I've been a poll worker and a registrar of voters, and it is not as simple as most people would think.

+1 Emphatic agreement that everyone should work the polls a few times. A lot the silly rhetoric would be quickly mooted if the belligerents had direct experience.

Sorry to pendant, because I know you already know this: some jurisdictions have (selectively) disabled the second chance feature for spoiled (or unreadable) ballots, effectively disenfranchising voters.

Until recently, we opponents have been the vocal minority. Shouted down, arrested, ignored.

One newspaper called me "a sweaty paranoid kook" for having the audacity to explain, citing the procedures and rules, how central count works.

Even today, you'll still see some yahoos pitching Chaum or RSA or Pret-a-Vote (?) based whackadoodle solutions. Completely divorced from how real world election administration works. But someday Tinkerbell will fly, if we all just clap hard enough, so stop interrupting with physics, logic, experience, thank you very much.

The crew that played the long game, and moved the needle, are the various members of the Election Verification Network. Unlike me, who burned out early.

We all owe the EVN a huge debt of gratitude.

Observation of vote counts is only a 1 to 1 relationship. You can't know if the whole election at scale is safe based on watching one person count votes. Perhaps it might feel safe to some, but that doesn't mean it is.

Moreover, physical observation doesn't scale. Only a handful of people can observe due to the physical limitations of meatspace. A simple attack could be to "DDoS" the count observations by crowding them out with your agents.

But that's not how vote counting works.

Typically there's a member of each party represented in teh election present for the counting, and often they're open to observers from the outside.

This means that you'd have to collude with your political enemies to steal their votes....

Collusion between supposed political enemies has happened frequently throughout history

There are currently 20 political parties in parliament representing vast ideological differences. They couldn't even agree on what pizza to order if their life depended on it lol.

At this point you obviously know you're mistaken. Why keep pushing it?

What am I mistaken about?

You don't even know how votes are counted yet you're trying to argue about the subject. Guess what, you can't DDoS polling stations with your "agents" and competing parties do not collude. Collude to do what? Lose the election? This isn't a joke.

>competing parties do not collude

That is literally the definition of collusion.

https://en.m.wikipedia.org/wiki/Collusion#Examples

https://www.cambridge.org/core/journals/world-politics/artic...

the observers at polling places are not highly ranked party members making deals in smoky rooms. they're almost certainly residents of the tiny area being polled, or the polity conducting the count.

certainly shady deals between a couple of folks at the tops of parties is a possibility, but collusion between the thousands or tens of thousands of local volunteers of both parties?

Never forget the Manhattan project. Mass collusion being difficult isn't necessarily a 100% certainty that it's not occurring.

It'd take quite a bit of money, though, this.

In my country, there are 50,000 polling stations [1] and the ruling party has 100,000 members [2].

The polling stations are open 7am-10pm, so 15 hours, before being driven to a central location for counting.

Thus, many polling stations don't have observers from the party at all - let alone the 2-4 observers needed to keep a constant watch from the first vote to the count.

[1] http://www.bbc.co.uk/news/election-2015-32229745 [2] https://en.wikipedia.org/wiki/Conservative_Party_(UK)#Member...

Polling stations aren't where the count happens, the ballot boxes are pooled at one site per constituency and counted there.

This is particularly noticeable at the spread-out constituencies like Na h-Eileanan an Iar, where the count has to wait for ballot boxes to arrive by boat or, if the budget is available, helicopter.

So then how can we be sure the ballot boxes weren't switched or stuffed inbetween the polling station and the count site?

Security seals. https://www.mfsecurityseals.uk/securing-general-election/

The ballot papers themselves are traceable, so if you're going to stuff the box you need to add the relevant number of ballots to the record and fake their issue to particular voters.

They're also in the custody of ballot officials; in the UK this is one area where I genuinely trust the ability of those people making up the system to operate genuinely according to the rules. We've even managed to hold fair elections in occupied Belfast.

Switching to e-voting would turn the whole thing over to known insecure consumer PCs and known suspect contractors like Diebold and Capita, all of whom I trust about as far as I can throw them; I trust the small army of ordinary people who make the paper count happen far more.

Maybe next election I'll volunteer as an observer.

Or you could help work on an open source technical solution so that this ridiculous corruptible process is fixed.

The paper solution is not ridiculous, it's perfectly workable.

The electronic solution is the one where malicious actors can substitute key parts of the process and things may be vulnerable to attacks that nobody has thought of (Meltdown/Spectre passim).

You would presumably have to compromise a lot of ballot boxes in order to influence an election. This would require a coordinated conspiracy potentially involving thousands of people.

If you can pull that off you might as well say fuck it and take over the country with force.

Each polling place only has about a thousand votes, making it a process that I can easily audit completely. The results for each polling place are reported online, so I can check a part of the process from beginning to end. If you have just ten people or so, each auditing one random polling station, you can easily get to a statistically meaningful assurance of the election’s integrity.

> it’s not enough for elections to just be safe. They also need to be safe in a way that is obvious to non-cryptographers

Not as much so in nations where people generally trust experts and government officials, and where conspiracy theories are mostly on the fringe (i.e. not the U.S.). Although I agree that it's preferable if non-techies can understand the mechanisms, and the downside of paper ballots plus mail-in is already pretty small.

Although I trust experts and, for the most part, the government, I see that sort of trust breaking in the society I live in (Germany), as well as others I follow (the US, Britain). The power of taking a conspiracy-minded neighbor to witness a transparent process to count the votes is breathtaking, and even a description of the process is a powerful argument.

It’s a really easy call in this debate, because the downsides of the process are neglible: the paper and pen process in my country produces results just as fast as what I see in the US.

Among the only substantive arguments I can think of is that US elections have a lot more items up for vote (ballot initiatives and the proverbial dog catchers), which makes hand-counting more cumbersome.

> Not as much so in nations where people generally trust experts and government officials

The election is where that trust is anchored at. Whether people trust the current government can change. If trust in the election result and thus trust in the next government's legitimacy depends on trust in the current government, you have no safe way out of a situation where trust in the current government is gone--which ultimately is the reason we have elections in the first place: As long as everyone is content with the current government, you don't really need elections at all. The whole point of having elections is that you can change a government that you don't trust, should the need arise.

   and it’s actually damn easy to achieve

And it's very very expensive

How expensive? In my country, where we still use paper ballots dropped into sealed boxes, the whole budget for the election operations is less than 1€/citizen.

Wow really ? Sounds reasonable.

(Should also take into account people transportation to the location, and the opportunity cost of their time)

My only personal experience is in cities, but my home was never more than 20m on foot from my voting place. I also know that even villages with less than 300 residents have one.

We also have many thousands of voting booths, so waiting times are low. One hour is rare.

UK election: £142m for 42m voters, or about £3 each.

http://www.bbc.co.uk/news/election-2017-39882972

The Irish e-voting debacle cost €55m for an electorate of roughly 3 million people: https://www.independent.ie/irish-news/54m-voting-machines-sc...

I'm not advocating for e-voting because they don't seem to offer (yet) the same good qualities as regular voting.

But you need to take into account people transportation and opportunity cost (if I lose 1 hour going to vote, this time could have been directed at working, thus it cost me ten or twenty the price you just spoke about).

I'm inclined to believe many of these reports are biased and still pushing agendas, even if it is out of peoples on lack of understanding of how a blockchain solution would actually work. Largely, their inability to think creatively about technical solutions.

Current e-voting systems don't seem to offer privacy, and security.. (are they even open source ?)

But yes a distributed e-voting system based on blockchain or whatever could work and maybe much cheaper as it would only take a few minutes online to vote. The result could be that more referendums would be held more often.

Home PCs are too easily compromised for this to be really trustworthy, even if you issue all the voters with hardware tokens. It might be achievable with iPhones, but that's rather exclusive.

In other words, it's not the backend that's the worry so much as the frontend. (Leaving aside all the blockchain-specific issues)

There is no electronic voting system, real or imagined, which can guarantee both the secret ballot and the public vote count. Because there is no algorithmic equivalent to the secure one-way hash for dropping physical ballots into a ballot box.

Block chains record votes cast in order. Which correlate with the order those same ballots were issued (or received, as with postal ballots). Voila, no more secret ballot.

The proposal to fudge the time stamps (chunking, roundoff or jitter) does not work for real world USA elections. Because the smallest administrative unit is the precinct (from 0 to 1000 voters) and ballot casting is bursty (vs spread evenly over time).

Reply on news.ycombinator.com