What’s wrong with using a Hand Count?
Many people advocate for the use of hand counting methods for tallying votes from paper ballots. That’s fine, but it may not be all it’s cracked up to be. Aside from the poor performance and relatively large resource cost of using hand counts, here we seek to zero in on the manner in which votes are counted by focusing on the employment of a teller committee. This committee is the “Achilles heel” of the hand count.
Keep in mind that there are many circumstances in which using a hand count will work fine for most people, particularly in small races that are won by wide margins. Even if the count is off, the question becomes whether the mistakes made are severe enough to affect the outcome of the election. If they are not, courts have held for years that it’s a moot point because there is no remedy necessary. Nevertheless, it’s important to point out the weaknesses in any process used so that a remedy can be sought if at all possible. Or perhaps hand counting isn’t as much of a “gold standard” as it is often made out to be by its advocates. You decide.
The Importance of Consensus in a Teller Committee
With the hand count of paper ballots, a teller committee is formed in order to count the ballots. The process begins with a set of secret ballots that have been anonymized so that the identity of the voter is separated from the vote. Since
there is no audit mechanism to track the completed ballot content,
instead the same ballots are counted multiple times from people who hold opposing perspectives. So for example, a set of Democrats might watch Republicans count votes, and then flip roles so that the Republicans watch the Democrats count the votes. More likely would be a team that is a mix of Democrats and Republicans in equal quantities. This concept is transitive in the sense that it works in primary races as well. Here you might have conservative / patriot / grass root Republicans mixed with establishment / RINO Republicans. Or Communists and Socialists. The point is that the teller committee members should be seeded with members that do not agree on policy, but are FORCED to agree on election results!
The implied contention among committee members is intentional. It is by design and it is a necessary ingredient in any fair election counting process involving people. Without this contention, election results are met with skepticism - and they should be! Even though the voter is forced to trust them, the committee is attempting to practice a form of consensus. You can’t expect a fair outcome when all sides aren’t properly represented.
Of course the idea here is they should come up with the same totals. We hope that they do, but can these numbers be trusted?
Recall that in elections, we would rather things be based upon consensus rather than trust. Although a well formed teller committee practices a form of consensus, it really does boil down to trust because the voter has to trust the teller committee in 3 areas:
- That the committee is formed around consensus,
- that the committee won’t make mistakes, and that
- the members cannot be compromised
Unfortunately for voters, it's not uncommon for teller committees to fail in one - or even all three of these areas.
Zero Trust
The basic idea behind consensus is to form an environment where agreement can be reached on a purely objective basis, especially among those who might otherwise disagree. That’s why teller committees are supposed to have people from multiple parties counting the same ballots. The need for trust is minimized if there is a self reinforcing policy based on consensus in place. The problem is while it can be minimized, it can never be eliminated.
What if you can’t find enough people from opposing perspectives to count the ballots in a given region? For example, are you likely to find enough conservative Republican counters and election judges in downtown Atlanta? How about finding left-leaning voters like Democrats in remote towns in Alaska or Louisiana? Unless you are willing to pay to have the appropriate observers located on premise, this is a difficult hurdle to overcome. And then, who’s to say that somebody who is registered in any given party actually votes along with that party? Even if these shortcomings could be met theoretically, in practice it’s very difficult and highly unlikely. So much for our first concern - there’s no such thing as a “zero trust” teller committee. Strike one.
People aren’t Perfect
When it comes to counting votes, people make mistakes all the time - particularly when they are under a time constraint. Even when assisted with calculators, hand counted vote tabulations are regularly off - and it gets worse the larger the ballot size. All you have to do is perform an audit of any hand count for a large election to demonstrate this - you are certain to find errors in the count. The remedy here is to simply perform a recount - in fact, using teller committees, that is the ONLY form of remedy. But this is oftentimes impractical and expensive. Worse, the entire point of employing a teller committee is to count the votes - accurately! If a teller committee is unable to consistently perform the most basic function they are supposed to perform accurately, then why are they being used? Strike two.
Who you gonna call?
And finally, we get down to the most difficult requirement to enforce and that is the trustworthiness of the teller committee participants themselves. Is there any such thing as a human being that cannot be compromised? Betrayal can be incentivized in multiple ways. It take the form of financial bribery, or any other form of extortion - for example, the threat to reveal a persons otherwise private activities and proclivities. What about ruining them financially (cancel culture)? Or what if their family is threatened? Is counting ballots worth the cost of somebody’s life? Committee member coercion is perhaps the hardest obstacle to overcome. You can never know for sure that a vote counter cannot be compromised. Strike three.
The Crux of the Issue
Now that we’ve examined the teller committee process in detail, it seems like it is a lot more problematic than it seemed at first glance. What is the root of all these problems? It turns out there is one primary cause and that is the attempt to separate the voters identity from their vote. When that requirement is unnecessary, then people can just verify their own vote and count all the votes themselves. You don’t need a teller committee at all, and you shouldn’t have one primarily because of all the issues we have uncovered here.
Introducing Fraud
While the lack of a BAM on traditional paper ballots provides a magnificent mechanism accomplishing the goal of voter anonymity, it does so at an even greater cost. The big problem here is that it introduces a VERY wide opportunity to perform vote fraud. Not only is there no way for the voter to trace their ballot, neither can anybody else. This means that it is really easy to create new completed ballots out of thin air. This practice is known as ballot stuffing.
One approach that has been taken in the past is to provide specially encoded ballots using techniques like watermarking. While this appears to be a reasonable idea, voters have been reluctant to embrace placing anything on a ballot that they cannot read, decipher or understand. The good news is that the desired capability can be automatically provided using a BAM.
Let's enable a REAL Audit Capability
If there is a requirement for identity separation, the voter is going to need some kind of audit mechanism. The problem with a teller committee is that it simply provides no audit trail at all. Therefore, no form of accuracy can be achieved without starting completely over and then risking running into the same issues all over again. In fact, it’s impossible to guarantee accuracy when using a teller committee. Simply put,
Hand counting is an unverifiable and error prone process - by definition!
But suppose the hand count were 100% accurate ...
Even if the hand count is 100% accurate, an individual still has no way of knowing that their vote was counted correctly
We've clearly established that hand counts are by nature unverifiable without a BAM, but the same goes for machine counts. Neither one should be used in any election without a BAM issued in the manner prescribed by EIP. Conversely, either one can be used when issuing a BAM!
So what’s the better alternative?
It’s easy. Use paper or machines to gather votes, but don’t rely solely on a group of people to count the votes. In other words, don’t trust teller committees. Instead, make sure you use marked ballots by placing a BAM on each ballot and count the votes yourself or use automatic calculators like machines. Provide anyone and everyone the ability to count votes and simply compare the results. As long as there’s a unique number on each ballot so that the voter knows which ballot is theirs and nobody else does, this all works like a charm.
But if you really love using teller committees, go ahead. Just make sure you place a BAM on each anonymously issued ballot, make sure each voter gets a copy of their completed ballot and audit the election by scanning the ballots in preparation for electronic ballot reconstruction.
You’ll be better off and it's a lot more accurate!