An expert on voting machines explains how they work
scientificamerican.com
scientificamerican.com
Germany uses (mostly) all paper ballots which are counted in each voting location by the local election committee, which is usually chaired by a government employee, and filled up with randomly chosen citizens. First results are often available 3-4h after closing of the voting location. With final results available during the next 1-2 days. All ballots will be collected on the next org level, to be available for audits/recounts.
Just the staff that processes absentee ballots (on normal years, even more so this year) is a large group of people who operate 24/7 in a large warehouse. We're fortunate that the county has the resources to dedicate a large building to the election bureau, in other counties they have to borrow space in community centers or courthouses for counting and it's more of a logistical challenge to get equipment in/out. Yes, many of these same challenges exist and are managed for the physical polling locations, but hand-counting would be piling a lot more workforce and space to manage (I would guess about double) on a county that is already struggling to afford the operation. The use of precinct tabulators keeps the polling-place staff fairly small while meaning that the ballots are already tabulated when they are delivered to the election bureau.
I have been involved in hand-counting ballots and it is remarkable 1) how tedious and time consuming it is, and 2) how error-prone the process is. With our typical ballots (18" long paper with three columns of questions on both sides, typical Dominion design), it takes a while for each ballot and it's a two-person process for accuracy. Even then the election administration feels that the error rate is unacceptable and so takes every measure possible to avoid ballots that must be hand-tallied. Much of this is a result of election code requirements for accuracy that involve typically three-person (and two-party) sign-off on every tally operation. Security restrictions on access to the computer totaling system adds further complexity and has lead the county to abandon "distributed" hand-tally operations, with all handling now being done centrally in the warehouse.
My gut feeling is that countries which hand-count ballots probably have a lower emphasis on absolute accuracy than US election administrators typically do. I have no idea but would also be interested in knowing how common it is for election workers to be paid in Europe vs in the US. My gut feeling, though, is that hand-tallying is not practical in the US without a significant expansion of the budget for election administrators. I'm also unconvinced that hand-tallying has any meaningful advantages over precinct tabulation, since both allow for post-tally audit and in this jurisdiction and many others, hand-tally audit of a sample of machine tabulation is a standard part of post-election canvassing.
As an addendum on something somewhat poorly understood, in most US jurisdictions the long delay between "unofficial" and "official" election results is a result of the canvassing process, which is a combination of hand-processing "difficult" ballots like provisional and mismarked ballots, and a standardized post-election audit which may include hand-tallying a sample of the machine-tabulated ballots (I would like this to be a mandatory practice nationally). Much of this wait is not because the machine tabulation is slow, but because the machine tabulation is not completely trusted and there is an auditing process that takes some time, plus there being ballots which are inherently not suited for machine tabulation. In my area the unofficial tabulation is usually completed overnight after election day, but the canvassing process takes substantially longer because, among other things, every provisional ballot must be manually researched by a more highly trained records team, which is a measure to avoid denying anyone the right to vote because of procedural problems with looking them up in the pollbook.
Finally, as discussed, not all states allow absentee/mail ballots to be processed before polls close. There's a tradeoff here and I don't think this rule is entirely unreasonable, but it does obviously add to the delay.
And the counting is done in polling stations after the voting process: just more hands and eyes that way.
And because it's that distributed, it's pretty much parallelized as much as you could. Another quality is you'd have to corrupt a fair amount of polling stations to make a substantial difference if you wanted to cheat.
Results are then sent to the next level (usually city hall which aggregates that and then sends that to the next level).
If we are talking about the allocation of money, then this is a political problem, not a monetary one, as money is clearly available.
[1] https://www.opensecrets.org/elections-overview/cost-of-elect...
They're completely different thing.
[0] https://en.wikipedia.org/wiki/Brooks_Brothers_riot
[1] https://www.msn.com/en-us/news/politics/what-is-the-brooks-b...
Actually I believe it's a feature, as it would (politically) be a lot harder to argue for the poll workers to be one-sided if your own party members are part of it.
You can observe the vote happening, count on election night or recount as long as you follow the instructions and don't interfere with it in any way, this includes adjacent rooms.
I don't understand how the US manages the election like a 3rd world country.
To add, at the local level, voter suppression strongly benefits incumbent candidates, so there's not really an incentive to make the election process easy and smooth.
It's not nuclear physics.
In Switzerland every adult citizen get‘s a ballot sent home for every election and referendum. I don‘t know of there‘s a way to unsubscribe. You can send it by mail or take it to the polls on election day.
I can‘t imagine why you would restrict voting more than that (except for political games).
Most of the cases of extremely disrupted voting (such as 12 hour lines, mysteriously 'broken' voting machines in urban areas, or the 300,000 mail-in ballots that the postal system refused to deliver) are, however, intentional policies intended to reduce the turnout of Democratic voters. The US has a very long history of making it as hard as possible for non-conservatives to vote[0]; these irregularities are just the modern descendants of poll taxes and the Jim Crow laws.
Politically, the US has much more in common with Central/South America, and contemporary Poland under Law and Justice, than it does with places like Spain or Germany.
[0] https://en.wikipedia.org/wiki/Voter_suppression_in_the_Unite...
That we elect authoritarian national socialists and that division of power is a myth here due to our half-assed constitution is another matter. We are just pretty backwards as a society. PiS and their ideas are very popular here sadly.
Realize that in some states, the offices which produce identity documents for people who don't drive might only be open twelve days a year and are located in areas that are not accessible by public transit. For people at the bottom of the American income inequality pyramid, getting documentation might be something they can't afford, in terms of the time off work, taxi fares, and the costs of possible legal assistance.
Seems to me the good faith solution is just to make sure everyone can get identification easily. However, the foundational issue is that voter ID requirements, combined with hard-to-get documentation, is a bad faith policy intended to keep poorer people from voting. Fix the 'bad faith' problem and everything else is trivial to put into place.
> there's no documented problem with voter fraud that needs to be fixed through ID requirements
Maybe not, but that's not an argument against voter ID, it's an argument against one of the arguments for it. Slightly different.
> ID requirements are often imposed with the explicit intention of reducing voter turnout among the poor
Implicit, maybe, but do we have anyone on record saying they want voter ID to raise the median income of the average voter? Or that they don't want poor people to vote so we're going to make you bring in a driver's license?
I buy that this is a side effect/consequence of voter ID, but not that it's explicit.
> there is allegedly a large swath of Americans who don't have valid identification and can't get it easily
Maybe, but like someone else said in many European jurisdictions you are required to show ID, it costs you money to get it, and you have a process to get it that takes time.
> the offices which produce identity documents for people who don't drive might only be open twelve days a year and are located in areas that are not accessible by public transit
I'm definitely calling BS on only being able to get a non-driver state ID one day every month, and most of the US is not accessible by public transit. We are not as densely populated as Europe and expecting a nationwide public transit network that reaches 100% (or even 50%, honestly) of our citizens is unrealistic.
> For people at the bottom of the American income inequality pyramid, getting documentation might be something they can't afford
This I actually agree with, and this[0; PDF] is actually a great breakdown of costs associated with getting an ID. It's absolutely a biased source the number are accurate and you can ignore the politicizing narrative.
I agree that the good faith solution is making identification free for anyone who can actually prove their identity and needs the financial assistance. It's hard to argue against VID if anyone who can prove their identity gets an ID if they want it.
[0] http://sharedprosperityphila.org/documents/Revised-ID-Waiver...
The fact that everybody is cheering Philadelphia for only taking 4 days to count a couple hundred thousand ballots is just silly.
[0] https://files7.philadelphiavotes.com/maps/ward-maps/Wards_St...
The problem is mail in ballots, which were significantly more this year. When do you start processing them? In many states, the answer is: before election day.
In other states, notably PA, the law says "day of election". The legislature there deliberately did not take any action to change this. Why?
We're literally seeing a race right now between the vote counters trying to finish the count and the leader of the Republican party suing to stop the count and invalidate as many mail-in votes as he can (because he knows they are likely to be overwhelmingly for his opponent).
Why not do the pre-canvass prior to election day? What's wrong with verifying eligibility, via whatever legally mandated process, then holding the ballots. Or pre-tabulating them?
The latter is what San Francisco does. The mail in ballots are verified, and put into a tabulating machine specifically for the precinct. It holds the cards securely in a locked bin, and on election night, preliminary results can be pulled off the tabulation machine and combined with the in-person votes.
I am an elections clerk in San Francisco, and familiar with the processes. I say this is secure, and easy to verify and re-verify and saves a ton of effort on election day.
Where is your evidence that what you suggest is more secure, or important for... some reason?
I don't think you realize that during election day, the ballot cards are 'counted' (tabulated) during the day, all day. The scanning machine on site is tabulating the cards, then at the end of the day it's a simple matter of pulling the final tally after the last vote, both via paper print out and electronic media (CF cards). There is no "end of day counting" - just closing procedures, and clean up.
In a state you might have two state legislative seats up, plus some county offices.
https://chalog.files.wordpress.com/2011/03/stimmzettel.jpg
Every voter had 93 votes, one for each seat in the city council. It's a rather complicated system, where you can freely assign up to 3 votes of your 93 votes to each candidate, can strike candidates from the list or simply vote for party lists.
Federal elections are way simpler, with just 2 votes. But often several elections are combined. But there are less elections overall in Germany, as we usually only vote for: city council, county council, state and federal parliaments, and finally EU parliament. 5 in total, most of them every 4 years, so on average 1.2 votings per year.
But if you put aside the security or the bug free status of the whole thing, a true democratic vote has at least 2 properties:
1. it's anonymous
2. it can be audited by a layman
Because of the complexity or bugs or whatever, #1 cannot really be guaranteed. Voting machine vendors will say they have no way to track voters until someone finds a flow or figure it out. Or machines could be altered or hacked or...
But really the biggest problem is #2: because if there's a problem, you need a paper trail. Because in the event of a recount, you want your average joe or jane to be able to recount ballots.
You could bring all the blockchain stuff (because I often hear that) or what not into the argument, you still wouldn't be able to do both #1 and #2.
So to me, voting must remain a paper thing.
If all counting machines are the same, how do you know they're not all wrong? Or not all tampered with?
And even if they all give you the same results, how do you know it's actually correct or how do you decide to recount? You have no real way to: you have to count. So at that point you might as well have counted in the first place.
Ideally you should have multiple vendors and make sure they all give the same results. But who's going to do that?
You can't do both #1 and #2 now. An anonymous ballot can't be audited, because it has no provenance -- if somebody slips an extra ballot in, how would you tell?
In Denmark you don't know what the ballots look like before election day, so you can't produce them in advance, and they aren't printed on printerpaper but is larger and the paper has a distinct texture/feeling.
The ballot box is made of transparent plastic (something like [1]), and you don't slide the envelope in it yourself. The slot is blocked until an official pulls a lever to let it go through.
So you put the envelope by the slot, if you keep holding it you will be asked you to let it go (so you can't cleverly hold multiple of them while it looks like one). They'll check again who you are and if you can actually vote and so on. If all is good, an official will pull the lever.
[1] https://www.jpg.fr/content/images/product/09812-00J_1_xnl.jp...
And then you have:
- one person who opens the envelope containing the ballot and unfolds it
- one who reads out loud the ballot
- at least two who write down what the person above said
And all that is supervised by the same people who were manning the polling station (at least 4 them).
So if you have multiple ballots in the same envelope, they will call it. The only way it'd work is if they all worked together. And you'd have to do that for a couple counting stations in many given polling stations... That sounds like an inside job you couldn't keep secret for long.
Because one thing I forgot to mention: counting is done at the polling station, it's not centralized, only results are.
If you stake everything you have on the original process being incorruptible, then your only response to an audit that shows a discrepancy must be "the audit is wrong". That's unlikely to satisfy many people; if that is your only strategy for dealing with audits, you are running an inauditable process.
Volkswagen proves this to be an unsafe assumption - their technology intentionally behaved differently during an audit.
I'm pretty sure it would be straightforward for the devil to know when they were being audited or not.
Auditing means things like paper trails (get an automatic electronic tally, but produce a verifiable paper ballot that can be counted in the case of any apparent inconsistencies). This discourages cheating on the electronic tallies because you never know when an audit may take place.
A fundamental flaw of direct-recording voting machines—that is, ones where you pull the lever on an old mechanical machine or you touch the touch screen on a modern one—all of those machines end up being completely impossible to audit. There’s no way to know whether the machine was honest or not, short of taking it apart and actually being able to inspect the mechanism. We have no good way of doing that with software. The complete lack of any auditable record of the count, so that you had to completely trust programmers, was a real problem. [But] the vast majority of votes in the U.S. today are being recorded on paper ballots that are filled in by hand. That makes me feel reasonably good. And furthermore, a growing number of states have some kind of an audit law.
A fundamental flaw of direct-recording voting machines—that is, ones where you pull the lever on an old mechanical machine or you touch the touch screen on a modern one—all of those machines end up being completely impossible to audit. There’s no way to know whether the machine was honest or not, short of taking it apart and actually being able to inspect the mechanism. We have no good way of doing that with software. The complete lack of any auditable record of the count, so that you had to completely trust programmers, was a real problem. [But] the vast majority of votes in the U.S. today are being recorded on paper ballots that are filled in by hand. That makes me feel reasonably good. And furthermore, a growing number of states have some kind of an audit law.
Broken Ballots book website: http://timbergroves.com/bb/
Douglas W. Jones: http://homepage.divms.uiowa.edu/~jones/
Best of both worlds.
If it is every time, it means at the end of the day you don't really trust the machine. You could announce results early but they wouldn't be official so what's the point?
And if the count only gets verified every once in a while, what triggers the process? Something abnormal? A random check? Something else?
What tells you the count is valid or not? How do you ensure the machine is working as it should, wasn't altered in any way and so on?
To me these problems can't really be fixed: you need the paper trail. And you can't "count" (haha) on a machine you can't really audit.
Many states conduct a mandatory sample audit of machine tabulation, either by hand or using an independent machine. I would urge those concerned about election security to lobby for this requirement in states that don't have it, as it is a practical and readily achievable improvement in election integrity.
In many states which use machine tabulation, machines are tested using a reference ballot set as part of the commissioning process, immediately before being sealed for delivery to the polling place. Once again, a requirement that should exist in those states which do not have it.
The only way to make sure it's partially secure is to audit the code and even with that you could miss stuff. Code for these things must be open.
Plus a quick google search shows (facepalm) the following (there are many others like that): https://www.nbcnews.com/politics/elections/online-vulnerable...
The three largest voting manufacturing companies — Election Systems &Software, Dominion Voting Systems and Hart InterCivic — have acknowledged they all put modems in some of their tabulators and scanners. The reason? So that unofficial election results can more quickly be relayed to the public. Those modems connect to cell phone networks, which, in turn, are connected to the internet.
The largest manufacturer of voting machines, ES&S, told NBC News their systems are protected by firewalls and are not on the “public internet.” But both Skoglund and Andrew Appel, a Princeton computer science professor and expert on elections, said such firewalls can and have been breached.
And you don't count alone, there are people with you who count with you and who double check what you do.
It works the following way. You legally must have: - one person who opens the envelope containing the ballot and unfolds it - one who reads out loud the ballot - at least two who write down what the person above said
And all that is supervised by the same people who were manning the polling station (at least 4 them).
So it becomes a lot harder not to trust humans when you have at least 4 persons who saw the ballot and a couple others observing if everything is being done as expected.
At the end of the night, the electronic tabulation is removed from the machine (CF cards), then given in a security sealed bag to a deputy officer who takes it to city hall to be summed up for the results. The paper results are organized, and put in security sealed bags, and given to a sheriff (person with gun), and locked into a van to be transported back to city hall.
So now we have the best of both worlds: fast results, and paper audit trail. Any precinct can be retabulated - just run all the cards thru a machine again. The mail-in ballots are processed separately, and once the envelopes have been verified, the ballots are put thru a tabulation machine as well. My understanding is city hall has a ballot tabulator for every precinct mail-in, and as they process mail they drop them in the appropriate machine. If a mistake was made - it's easily correctable because each ballot has the precinct number printed on it.
It's a very easy to understand system, that requires only basic elements of computer security. Cross checking a ballot tabulator for accuracy is easy. Recounting is easy.
And finally, for accessible voting, people use a tablet to choose their selections, then their ballot is printed out, which contains the summary of the vote, and a QR code the ballot machine scans.
Oh yes, and the ballot machine scans and keeps image files of every ballot. I'm not sure what they do with that, but there we go.
It's fascinating, and my experiences in the SF elections department gives me high confidence that the SF election, and any run like it, are very secure, and accurate.
With fill-in-the-bubble ballots, a typical error would be printing the ballot with the candidates in one order but programming the scanner with the candidates in a different order. There was a famous incident of this happening [in] Pottawattamie County, [Iowa], in 2006 in its June primary. The net result was that the numbers that came up were pretty nonsensical and bore no relationship with anyone’s expectation of the outcome of the race. The election officials noticed. They did a hand count, and that completely resolved the problem. And they could because they had paper ballots.
I think like most things there are tradeoffs and the qualities most needed in the system run counter to one another--speed can cause errors to be missed, security reduces access, etc.
And it's probably easy in the firmware to check if you're being tested (ask Volkswagen). And even if it's not, you could write a firmware you could trigger with a specific ballot.
The bottom line is firmware for these things should and must be open so they can be audited by experts. And hackers must be given access to them so they can try to do their worst.
As long as you treat it as a black box, I'm afraid to say you aren't trusting the device, you are just having faith in it.
Generally, from a network security perspective, electronic pollbook systems are a much greater concern as they are connected over the internet as a matter of standard practice, and it's difficult to avoid this while preserving functionality. There are also more, smaller vendors for EPBs and they receive significantly less scrutiny or security review than tabulators.
Ohio, the only other state I've voted in (many years ago) had electronic machines that produced a paper audit receipt. The problem I see with that is that now you have two sources of truth, and no easy way to count one of them to reconcile them.
DIY Election Fraud Analysis Using Benford’s Law
https://www.kdnuggets.com/2020/09/diy-election-fraud-analysi...
python code
Raw Data
https://www.kaggle.com/unanimad/us-election-2020?select=pres...