Ballot Hand Counts Lead to Inaccuracy

We hand count in Canada and don't have a notable accuracy problem, and ballot security is a rare concern. We manage that by having simple ballots which make voter intent clear while having the count being run by an independent, non-partisan national organization _and_ each table where tabulating is occurring can be watched over by representatives of the parties running.

_And_ we have mail-in ballots.

When issues do arise it tends to occur when a ballot box needs to be transported between locations; when this occurs it is taken quite seriously by Elections Canada.

It works great. Perhaps the USA should contact Elections Canada and learn a thing or two.

It's important to add some context to the "hand count" idea in the US. It was part of a larger scheme to overturn the results of the 2020 election which involved creating doubt about the validity of results in key states. The larger scheme involved various extra-legal attempts to have state legislatures invalidate the election, have the Supreme Count step in, or have the counting of Electoral Votes disrupted and the final decision on the presidential election made in the House of Representatives. Calling into doubt the machine counts and asking for hand counting of ballots is part of that scheme.

Not one official who was elected in 2020 at the state legislature level called for invalidating their own election and recounting their race.

And one county in Arizona that was considering hand counting, a county that voted overwhelmingly across the board for the party attempting to overturn the election, looked at the practicality of a hand count and decided it was too expensive and problem prone. https://www.naco.org/news/numbers-stack-against-hand-count-m...

We do hand counts here in Canada and it's pretty accurate. We do have scrutineers as part of the process and ballot security is pretty paramount. Recounts are mostly automatic and done when margins are too close.

I think America leaned too far in on the voting machines, TBH.

I think we can use technology to do better than paper ballots with hand counting, but one tweet mentioning 25% error rate is not reason enough for this title in my opinion. This smells of "big tabulator" sponsored article. :D

In Germany we do hand counting and i have participated on multiple occasions and would say

1. it is really hard to do any fraud for multiple reasons:

  \* the people counting are pseudo randomly chosen

  \* but even if you somehow managed to get "your" people to count then you've only secured one very small portion of the votes

  \* the counting areas are near to each other so some random citizens would be able to observe multiple counting areas at once

  \* there is no "winner takes all" rule in Germany and neither is there a electoral college. 

2. The counting and reporting processes are well monitored and documented so that simple inconsistencies become visible right away either in the counting group (7-9 people in general) or when they phone in the results to the central point.

3. There is obviously a cost involved in having people count BUT i can't imagine it being more that paying for the counting machines since the people involved aren't paid normal wages but instead expense allowances.

I fully support handcounting over machine counting! But the US would still have to get rid of old traditions, namely electoral college and gerrymandering practices. The popular vote should be the deciding factor most of the time!

The article is trying to combat a bad faith initiative with factual information. The truth is that adding delays and uncertainty to the process are the goals, not just some unhappy byproducts. Delays and uncertainty make room for contesting results you don’t like. That’s the whole plan.

What many people outside the US may not realize is that ballots here may include 40-60+ races. The presidential race appears on the same ballot the races for Senate, House of Representatives (both national races,) state wide positions like governor, attorney general, superintendent of public instruction, state treasurer and secretary of state, and state mine inspector. In some states superior court judges are on the ballot for "retention" as well as State Supreme Court positions. Then come the state legislature races. Most states have two houses so a state-level Senate and House. Then come county positions, Sheriff, Board of Supervisors, Recorder, Treasurer, County Attorney, Justice of the Peace and constable. Then the school board races so you may have a high school district board and an elementary school district board. Then come the referendums and initiatives that either the state legislature put on the ballot or citizens gathered enough signatures for; these typically change state law but sometimes implement or remove new taxes. In 2024 in Maricopa County, AZ we had a two page ballot because the legislature added 20+ initiatives to try and implement laws the governor had vetoed.

In any case, the idea that these ballots can be accurately hand counted is absurd. In some of the hand count examples in the states mentioned in the article only one or two races on ballots with 10's of races were recounted and even those recounts were problematic.

What's really telling is that none of the people elected in 2020 or 2022 or 2024 at the state legislature level are calling for recounts of THEIR races. Machines are fine when the right party wins.

Machines are the only way to deal with a ballot with 60 races. There are other parameters that could be put in place to help improve people's faith in the system. San Francisco makes ballot images available via a web portal. It's entirely possible that an AI model could be trained to rapidly recount any and all of the races to validate the official results. Tighter ID requirements would be OK, it's 2025 and even people in the hills and reservations should have IDs.

Interestingly, in most of Europe, paper ballots and hand counting are preferred methods. We have seen times when electronic ballots have had glitches and have had inaccurate counts. Electronic ballots have experienced errors in tabulation and also loss of data, so it's not foolproof. In addition, vendors are known to be slow to patch security vulnerabilities in their systems --this leads to loss in confidence. Also as others have observed, in order to subvert paper ballots and counting, you have to compromise lots of people.

Quick googling shows that japan manually counted 55 million votes in less than 24 hours in 2024 parliament elections.

Personally I agree.

However, I think a weakness in the arguments is that many of the tabulation machines we have are very old, poorly designed for vote integrity (eg no paper ballots or confusing scan sheet design), and closed source which can lead to accusations of both hardware/software flipping votes and the systems being impossible to audit.

To counter these concerns you can use paper ballots everywhere, open source software for the machines, and risk limiting audits to verify the count.

Voting Works builds software for both sides of these voting processes as OSS with machine scanned paper ballots.

https://www.voting.works/audits

https://www.voting.works/machines

Article without references to other countries where manual count is the only way. At least here in Chile we count votes by hand, results are ready at most 6 hours after poll closing time and Zero fraud has been reported in more than 30 years.

Hand counts is the way it is done in Sweden. It means that anyone can easily (well, as easy as possible) both understand the system and inspect it locally (all counting is open to the public).

The result is delivered in a matter of hours, and since vote counting is a parallell process it scales well enough that I doubt it would take much longer in the US. The US have massively more complicated ballots though, which I think is another issue entirely that you guys must solve.

I doubt it would be very expensive either. You don't have to pay people very much to do their civic duty every couple of years and count a couple hundred ballots.

The only secure way is privately produced black box voting machines with secret proprietary software. It's really the most efficient way to manipulate huge numbers of votes while involving as few people as possible. Manipulating vote counts with hand-counting requires large numbers of human agents distributed across many counties - a logistical nightmare.

Of all of the approaches in the article, I like the one about doing both to audit each other the most. Even if the extra time delays mentioned were unresolvable, I'd rather wait a couple of days and pay slightly more to run the election than spend more time and money with political battles about which counting approach is superior.

If you insist on 'immediate' results just do a machine count first with a mandatory hand count afterwards. The machine count gives fast results which can be confirmed by the hand count. If there are significant discrepancies redo the hand count until there are two counts which give the same results.

Handcount in Slovenia, first count done on local polling location, with a recount done by a central election comission and certification of the result. It works.

edit: First results are in within a couple of hours, depending on how many things are on the ballot.

Are we talking about intentional or unintentional inaccuracies?

Unintentional ones should be distributed more or less evenly, if everyone have 10 less votes it should not change the end result.

But intentional ones, with the objective of trying to rig the result in one particular direction, and deep enough into whoever is doing the count/election. But in the end, it goes to how big the conspiracy should be? You may need just a few to rig all the voting machines (do you have the source code? of what was actually running in production everywhere?), but with human counters to get to the right scale you may need to involve really a lot of people.

What a load of nonsense. In several EU countries, ballots are counted by hand and an accurate result is there usually by midnight.

In Germany, the Supreme Court has banned voting machines as unconstitutional.

The method is simple: Have relatively small voting districts, recruit civil servants as supervisors. Allow anyone from all parties to supervise and be present, especially during counting. Hand out the ballot sheet upon displaying an ID. Upload the counted result to a website where people can check that the reported count matches the actual count.

The EU has a lot of problems, efficient voting isn't one of them.

This field is ripe for security consulting firms, but they seem completely uninterested. Weird.

Electronic voting is the way forward, like done in Brazil.

25% error rate seems more like corruption than errors.

Are minor, random inaccuracies due to hand-counting mistakes the main danger a voting system must guard against, or is it against covert, malicious manipulation from compromised computer systems?

> Prior to 2022, no legislatures were considering this [mandating hand counts] extreme proposal.

An honest article wouldn't call the counting method used in most of Europe, Canada, Japan, India, and countless other countries, "extreme". Of course it deftly omits mention of any other country, leaving the reader uninformed. I'm sure it's just a coincidence..