New bill could finally get rid of paperless voting machines

This seems like a step in the right direction, but I’m concerned with some of the language:

The Lankford bill would enshrine this thinking into federal law. “Funds received under a grant under this section may not be used for any voting system that records each vote in electronic storage unless the system is an optical scanner that reads paper ballots,” the bill says.

This would seem to discourage an electronic system that generates a paper ballot that can be verified by the voter and later used for recounts, etc. I’m certainly not versed in the specifics of these systems, but I don’t see an immediate problem with keeping the UI electronic while also allowing for an audit trail.


You could probably do a hybrid system, where you select (but not electronically record) the ballot on the computer, print it, and then, once the paper vote is verified, deposit it into a machine that scans it back in with something similar to a Scantron machine.

The paper ballot would be the official tally, but it would be generated through the computer interface, printed, and then scanned into electronic storage.

This also would allow people to fill in the paper ballot manually if they don’t trust the computer system.


Agreed, but I don’t understand how a system like that is more secure than one which records the vote at point of entry and then generates a paper ballot that can be used for verification/recount/etc. In either scenario, the only feedback the voter receives about their voting record is the paper. There’s no way for an individual voter to verify that the electronic record is accurate. That’s presumably what the audits would help with.


I could make a few arguments.

First, a simpler, tailor-made program is always (assuming a competent programmer) going to have fewer bugs and will be easier to audit than a more complex program. In the case where the system is taking the user input, recording that input into a database, and then printing the paper record, there are a lot more places where something could go wrong and record the wrong vote, or, worse, print a different vote than the one selected.

On the other hand, a Scantron system’s programming is so simple as to be practically bulletproof.

Second, if you separate the ballot-printing and ballot-recording into two different machines, then the ballot isn’t final until after the voter has reviewed the paper ballot. So, even if something goes wrong in the first half of the process, the voter can confirm that the printed version is correct, and, if it’s not, then no vote has been recorded and the voter can manually fill out a paper ballot with their actual choices, to be tallied by the Scantron.

It’s no easier to verify, true, but it’s much harder to tamper with, and much easier to audit the code itself.


This is how it was done the last place I voted - I got a receipt showing my vote, and if anything was weird, there were a couple volunteers I could bring it to. Might not scale easily in big cities, but the next major voting bill should make election day a holiday so more can volunteer. I know I would.


This is a reasonable argument for voter verification and preventing the need to revert data. I’m still not sure it’s more secure though.

I sincerely doubt the code for these proposed systems will ever be audited directly.


You can offer a voter an opportunity to “verify” a ballot generated by a computer but you can’t know that any voter actually did that.

The simplest, most direct connection between a voter’s intent and a mark on a ballot is if the voter is the one who marks the ballot.

Any additional machinery inserted into the process is an opportunity for shenanigans.

In an adversarial context, where there is no rational reason for anybody to trust anybody else, the only way to get an accurate ballot count is if all the adversaries count the ballots together, in the same room, at the same time, with their eyeballs, and agree on the results.

Any additional machinery inserted into the process is an opportunity for shenanigans.

Electronic voting is a non-solution to a non-problem. It’s not necessary to automate a fundamentally human activity.


That still leaves the Virginia elections. If that one voter “crossed out” one delegate choice, did they also “cross out” their governor choice?


The problem is that of unintentionally spoiled or miscast ballots. Which do happen. People are generally not good at following instructions.

I remember - I think it was in the 2000 election - a ballot by someone who had forgotten their glasses at home, and so they wrote their choice in the margin on the side of the ballot, since they couldn’t see the ovals to fill in. The ballot was counted as spoiled, despite the choice being clear.

If you can create a written ballot without unintentional spoilage then yes, I’ll agree with you that there is no problem that using a machine to electronically fill out your ballot can solve.


The machine does not fix that problem. A voter can also unintentionally vote wrong using the machine. On the contrary, the machine hides the evidence of the problem, making it worse.

The other problem the machine is supposed to solve but doesn’t? Disabled people who can’t vote by themselves? BY DEFINITION, this class of voter needs somebody to help them vote. If you wouldn’t trust a human attendant to help somebody vote while maintaining the secret ballot and getting the ballot counted correctly, then there is NO REASON TO TRUST THE MACHINE EITHER. Trusting a machine means trusting all the people who built the machine, and everybody who touched the machine since then, and everybody who will touch the machine between the time the voter votes and the time the vote is counted.

It’s really a kind of magical thinking. It’s like the '80s pop-culture concept of what a computer is. There are these edge cases that we don’t feel good about, so we’ll put a screen and a keyboard in front of it and that’ll make our icky feelings go away.

1 Like

The machine I describe above, where it prints a Scantron-style ballot for the voter to review, formatted correctly, which is then either counted manually or fed into a machine with incredibly simple programming to count, would fix the problem of unintentionally spoiled ballots.


Didn’t want to open up a whole new thread for this, and this seemed like the most similar one.

The issue was complicated by the fact that Georgia’s legislature recently passed legislation directing that the state develop a new election system based on ballot-marking devices—electronic voting machines that print out a paper ballot the voter can examine.

Georgia has signed a contract with a vendor for these new machines and plans to start testing them in a few cities in this November’s elections. The state aims to start rolling the new system out statewide in time for next March’s presidential primary. Under that timeline, the state would stop using its current, insecure machines before the end of the year.

The problem, critics point out, is that the state may not be able to roll out the new system in time for next March’s election.


Last Week Tonight addressing the problem of paperless voting machines.