Discussion
Election security is hard, and US elections have some unique problems and constraints (we have, overwhelming, the most complex ballots in the world, for example). There are non-intuitive tradeoffs everwhere.
An excellent resource for understanding US election security, especially as it relates to technology, is this 2018 National Academies consensus study, "Securing the Vote". It was published eight years ago, but is still current.
@mattblaze There should be no announcing of results until all the ballots are under lock and key. Knowing how many votes you need to fabricate makes cheating a lot easier.
This stuff is mostly about restoring TRUST in the vote. Nobody trusts an election where votes just keep coming in for two weeks and the outcome changes.
Yes we keep hearing about the 2020 election, but remember the other side kept questioning the 2016 election. Nobody trusts the current system. That's a problem.
@mike805 I'm not aware of any US jurisdiction that announces results before the polls are closed.
@mattblaze In California we get "day of" counts and then most of the time mail-in votes keep coming in for weeks and several races usually flip.
Not saying that's cheating but it creates a loss of confidence.
When Abraham Lincoln ran in 1860, he spent the day at the telegraph office, where the staff read him the returns as they came in. By the end of the day he was pretty sure he had won, and congratulations were coming in.
We ought to be able to at least equal that performance today.
@mike805 California law requires a postmark by election day for ballots received up to (I think) five days after election day. Are you suggesting that there is some scalable way to inject fraudulently postmarked ballots into the US mail? That would be very surprising. I've certainly never heard of that.
Again, in no US jurisdiction are results announced while it is still possible to cast a vote.
@mattblaze Mark sense voting does seem to be the best approach. They can be counted at electronic speed, and verified manually if required. Stuffing the ballot box is a large-scale physical problem.
Touch screen voting with no physical ballot was a bad idea and never should have existed.
State ID should be required for in person voting. There is no reason except cheating to not require it.
Mail in should be a small exception not the norm.
Purple fingers are also an option.
"we have, overwhelming, the most complex ballots in the world, for example"
@resuna I’m not sure what your point is. That looks like an Australian ballot , with a single contest. It’s large because there are a large number of ranked candidates.
US ballots typically have fewer candidates, but many, many contests, often with multiple voting methods represented.
I’m not really interested in having an argument over basic, well understood factual information.
In Australia you would have one form like that for each contest you're voting in, instead of a single form for all the contests, so that's equivalent to one single race.
@resuna Again, US ballots are far more complex. We can have 50-100 different contests, often with multiple different voting methods used, depending on the state.
I’m not sure what point you’re trying to make. You appear to have a lot of opinions about how to improve US elections.
@mattblaze Many thanks for this thread, I've been keeping my eye open for an authoritative source on US election security for a gullible friend (it seems she got the wrong sort of autism... 😐 ), didn't want to pester you but hadn't got round to a proper search....
Greatly appreciated, thanks again
@mattblaze "the most complex ballots in the world, for example"
I'd like to know more about this.
Do you mean having votes fir lots of things at once?
Have you seen an Australian Senate ballot paper? They elect 6 or 12 members and have the option of numbering every box "below the line" which can be over 100 boxes, using multi-member single transferable vote.
@mattblaze > Full ballots dissociated from individual voters might be posted in the interest of ensuring transparency and/or to facilitate auditing, but it may be possible to tie particular ballots to individual voters.
Couldn't this be accomplished with a hash provided to the voter that would be semi-anonymous. It seems like having access to the full ledger of all the votes tallied in a public place where you can see your record would allow trust through transparency.
@mattblaze Excellent information! As an IL election judge, I have seen over the last 8-10 years quite a few improvements in electronic voting technology that makes voting easier to manage.
That give us more time to argue with voters who claim "the sharpies bleed through the paper, and that's how Biden got elected in 2020!" or explain to people why they must remove their "electioneering" hats, tshirts, and other paraphrenalia. 🤦♂️
@jimthewhyguy Thank you for your work! It's an important service.
@mattblaze It's actually what keeps me hopeful about the future of our democracy. There's nothing like registering a new voter to counteract feelings of futility against the state of our country's politics today!
@jimthewhyguy It's so great. I've been brought near tears multiple times doing this work myself.
@mattblaze
Do you have a list of all the things Americans vote on?
Here in Alberta, Canada, we have:
1) federal elections: candidates for Member of Parliament of Canada
2) provincial elections: candidates for Member of the Legislative Assembly or Parliament of a province
3) municipal elections: city councils (not sure what the rural equivalent is), mayors (usually the weak kind that are functionally counsellors at large), & school board trustees
4) occasional referenda
@virtuous_sloth @mattblaze for me, a Californian, I regularly vote for school boards at all levels of education, judges, local and state officials (usually legal or financial responsibilities), regional boards for water or transit, sheriff, prosecutors. On top of that, it's usually 10-20 direct ballot measures like approving bonds or changing the state constitution on every ballot.
@evan @mattblaze
I wonder if the sheer number of decisions ends up hiding what functionally ends up being less actually agency, mostly in terms of a lack of diversity in the choice options.
@virtuous_sloth @evan California is an extreme case, but the size and complexity of US ballots is driven by multiple layers of government (federal, state, county, local), each with its own elected offices, our propensity to have specialized representatives (school boards, etc), requirements for direct approval in some states for things like bond issues, and the ability in some states to put voter-initiated referendums on the ballot.
Everything is there for a "good reason", but It all adds up.
@mattblaze I want to underline what you said about complexity.
My family are dual citizens, Canadian and US.
My daughter turned 18 and was entitled to vote in both countries.
Her first ballot in Canada had 1 choice: member of parliament.
Her first ballot in California had 81 choices: dozens of local, state and federal elected positions as well as a wall of local and state ballot measures.
@evan @mattblaze does Canada not elect local and province officials, or run those elections separately from federal offices? It seems convenient to the voters to get it all done with at once - the multiple special elections in California to replace Swalwell are somewhere between nuisance and botheration (especially since I'm voting by fax from the EU, now).
@oddhack @mattblaze Here in Quebec, they are separate elections. The municipal one has had the most choices I've had to do so far -- four. City mayor, city council, borough mayor, borough council.
@evan @mattblaze in California we also elect judges, school board administrators, and such, which bumps up the obscurity. Personally I just look for the most socialist Democratic candidate possible, if I don't have a more nuanced take.
@evan @mattblaze Not to mention the party-specific elections?! Which were totally unexpected news to me.
That said, there has been significant progress in securing elections since the NASEM study was published in 2018. Paperless "DRE" voting machines have become largely obsolete, greatly reducing the attack surface of elections in most states. And statistically rigorous post-election audits are starting to be performed as a routine part of elections in more and more states.
There's still quite a bit of work to do, but US elections have been getting steadily more secure and robust.
@mattblaze this is a question of genuine curiosity: have you ever compared the US electoral management and counting to the Australian system? Since we have to count the numbers on the ballots but also have a “washminster” system, I’m curious.
@piecritic They are just not comparable. US ballots have a huge number of contests, often with multiple different voting methods.
Austrialian ballots often involve a huge number of candidates (and ranking), but a small number of contests.
Parliamentary systems have advantages on election day.
@mattblaze It's been too long since I last engaged with voting systems, or even worked at a precinct on voting day.
Last time I worked an election - just before Covid - I was surprised how many voters chose to fill out the bubbles or draw the lines on paper ballots rather than use the ballot marking machines. It was perhaps a 10:1 ratio.
(I personally prefer the ballot marking devices because I have trouble making accurate marks with a hand-held pen.)
When I did my first company I discovered the wonder and value of double entry ledgers. And when I began to get involved with voting I was impressed how far my state (California) and the counties had created and deployed layers of cross checks that make it hard for voters (or voting workers) to make uncaught mistakes or for outside actors to interfere.
I would like to give flowers to whoever thought of the idea of provisional ballots.
A central problem in US election security is the size and complexity of ballots in most jurisdictions make automation essential in practice for accurately tallying votes. But automation (especially involving computers and software) is inherently untrustworthy (software has bugs, and can be quietly replaced with malware).
This seemed like an impossible, unresolvable tradeoff until two recent breakthroughs: the notion of "software independence" and methods for practical "risk-limiting audits"
@mattblaze There is, of course, this alternative to the complicated elections
@PeterLudemann worst form of government, except for all the others, as they say.
"Software independence", posed by Ron Rivest in 2007, formalizes a design goal for a voting system: A software error (even malicious software) anywhere in the system must not be able to cause an undetectable, unrecoverable change to the correct outcome of an election.
Risk-limiting audits, invented a few years later by Philip Stark, provide a practical way to achieve software independence with a type of existing voting technology (optical-scan paper ballots).
@mattblaze Can you recommend any particular reading on Software Independence? (I’m asking not because I’m unable to search the web myself, but because you obviously have domain expertise).
@mattblaze Were RLAs invented that recently? I recall the idea of manual recounts being used to check machine-tabulated results earlier than that. The idea was identical: if the machines are misbehaving, their results won't match the manual count. You decide the acceptable risk of not detecting a mismatch, then calculate the size of sample needed to keep the probability under that limit.
@mattblaze Read about how Brazilian elections are conducted. The system is 100% electronic with robust custom hardware (no touch screens, actual buttons), auditable code and frequent bug hunts.
The details of risk-limiting audits are too involved to go into here, but the basic idea is that a random sample of ballots are compared, by hand, with the machine interpretation of those ballots. If the machine behaved correctly, after a surprisingly small number of samples, you gain a high (and measurable) degree of confidence in the outcome.
Install bonfire.cafe
Get the full app experience
