Popular in Formal Sciences

Formal Sciences

Diameter rigidity for Kähler manifolds with positive bisectional curvature

Indian Institute of Science Ved Datar

Formal Sciences

Rank-Based Multivariate Sarmanov for Modeling Dependence

McMaster University Anas Abdallah

Formal Sciences

The Hybrid High-Order Method for Polytopal Meshes

Monash University

Formal Sciences

Financial Informatics: An Information-Based Approach to Asset Pricing

University of Surrey

Formal Sciences

Multiplicative orders of Gauss periods and the arithmetic of real quadratic fields

University of Newcastle

Formal Sciences

Some Numerical Methods for the Hele-Shaw Equations

University of Massachusetts at Amherst

Formal Sciences

Geometric stabilisation via p-adic integration

University of Toronto

A Study of Ballot Anomaly Detection with a Transparent voting machine

View more insights in Computer Science - Formal Sciences

View more figures from University of Florida

View the page of Juan Gilbert

Read the Study

Juan Gilbert discusses ballot anomaly detection and the use of a transparent voting machine

Image courtesy of interviewee. November 27, 2023

Generating Workspace

Export as PDF

This video video discusses the importance of ballot marking devices in ensuring the integrity and confidence in elections, highlighting their role in facilitating accessible voting and improving the accuracy of the vote tally process. It describes how these devices function as touchscreens that allow voters to select their choices, which are then printed on paper to serve as the official ballot.

The speaker emphasizes the necessity of implementing risk-limiting audits to verify the accuracy of election outcomes. Additionally, research is cited demonstrating that traditional hand-marked paper ballots can lead to confusion and misinterpretation of voter intent. A new transparent voting machine design is introduced, which aims to enhance the detection of any changes made to a voter's selections.

In studies conducted using this technology, a higher percentage of participants noticed if their selections were altered compared to previous methods, indicating the effectiveness of the new device in promoting election integrity. The conclusion asserts that the transparent voting machine design not only improves usability for voters but also enhances security by making it easier to identify discrepancies in the voting process.

Language:

Export as PDF

X-TIMESTAMP-MAP:LOCAL=00:00:00.000,MPEGTS=900000

00:00.000 --> 00:03.285
Well, it's an important area that is critical

00:03.305 --> 00:04.365
to our democracy,

00:05.805 --> 00:09.625
and I think it's essential that we have to do this work

00:10.165 --> 00:13.345
to ensure the integrity of elections, confidence

00:13.345 --> 00:16.585
of elections, and it's necessary.

00:17.165 --> 00:18.465
So that's why we do it.

00:19.165 --> 00:23.525
So a ballot marking device, think of it as a computer

00:24.265 --> 00:27.165
that's a touch screen where you can go in

00:27.225 --> 00:31.395
and touch the screen, essentially marking a ballot,

00:32.315 --> 00:36.335
and it prints on paper the your selections

00:36.675 --> 00:39.615
or the actual ballot, that is the ballot of record.

00:40.755 --> 00:43.565
That ballot then goes into a ballot box

00:44.725 --> 00:49.385
or scans at some point and used as part of the tally.

00:50.645 --> 00:52.145
So with that said,

00:52.675 --> 00:56.465
after you do that, you can do a risk limiting audit,

00:57.435 --> 01:01.825
which is a statistical audit, a specific type of audit

01:02.215 --> 01:04.025
that allows you to ensure the integrity

01:04.965 --> 01:07.145
of the actual tally machines that they,

01:07.205 --> 01:08.825
or the scanners that it goes through.

01:09.515 --> 01:12.145
There are places where you can use a ballot marking device.

01:12.675 --> 01:16.265
There are certain states in the United States that, some

01:16.265 --> 01:19.065
of them, the whole state uses a ballot marking device.

01:20.045 --> 01:22.825
Uh, there are other places that,

01:22.875 --> 01:24.065
don't have it statewide.

01:24.635 --> 01:27.985
There are places that still require you to mark on paper.

01:28.285 --> 01:30.665
If you have a disability, for example,

01:30.685 --> 01:31.985
you're visually impaired

01:32.085 --> 01:36.975
or blind, you can ask to use an accessible voting machine,

01:36.975 --> 01:38.815
which may be a ballot marking device.

01:39.875 --> 01:43.695
So these ballot marking devices are being used in elections

01:44.025 --> 01:45.615
nationwide, in the United States

01:46.075 --> 01:47.455
and in other places in the world.

01:48.115 --> 01:49.455
So they are quite popular.

01:50.235 --> 01:53.175
If you allow people to write on a ballot,

01:54.445 --> 01:56.735
they take the privilege of writing on the ballot,

01:57.465 --> 02:00.995
meaning they can mark any way they want, any how they want.

02:01.375 --> 02:03.675
And then what do you have? You have a debate.

02:04.015 --> 02:06.035
The two, in our case, the United States,

02:06.095 --> 02:08.115
two parties looking at the ballot, arguing.

02:08.335 --> 02:10.835
That's not of actual vote, that's not what they intended.

02:11.015 --> 02:15.555
You have to interpret someone's ballot selection.

02:16.365 --> 02:18.705
We do not want to be in that situation.

02:19.565 --> 02:21.465
So that's one of the negatives

02:21.685 --> 02:25.225
of allowing a hand marked paper ballot is the ability

02:25.405 --> 02:26.825
for people to mark any way they want.

02:26.825 --> 02:29.105
And then you have to interpret those marks

02:29.125 --> 02:33.185
and those interpretations may not translate to

02:33.185 --> 02:34.545
what the voter's intent was.

02:35.285 --> 02:36.465
So that's an issue.

02:37.005 --> 02:41.145
The transparent voting machine was motivated by research

02:41.145 --> 02:43.305
that was done at the University of Michigan

02:43.725 --> 02:44.945
and Rice University,

02:45.355 --> 02:47.385
where they looked at ballot marking devices

02:47.965 --> 02:49.505
and they were printing a ballot.

02:50.405 --> 02:53.905
And when they print the ballot, they would switch one

02:53.905 --> 02:55.505
of the selections that the voter made,

02:55.925 --> 02:58.225
and they wanted to see if people would notice.

02:59.485 --> 03:03.505
And, in their studies, they were reporting that,

03:03.765 --> 03:07.665
for example, in Michigan, only 6.6% of the voters noticed,

03:08.485 --> 03:09.865
that their vote was flipped.

03:10.135 --> 03:13.145
Well, in that case, that's an insufficient number

03:13.725 --> 03:15.705
to really detect the outcome

03:15.705 --> 03:17.465
of an election could be changed.

03:18.605 --> 03:23.235
So with that in mind, I designed this transparent,

03:23.455 --> 03:24.515
voting machine.

03:25.185 --> 03:26.645
The idea is that

03:27.555 --> 03:31.455
if you could turn the paper almost into a touch screen such

03:31.455 --> 03:35.545
that you could touch the screen,

03:35.805 --> 03:40.225
but it almost is like interacting with the paper in a way

03:40.295 --> 03:42.825
that every time you make a selection, you have

03:42.825 --> 03:45.545
to actually touch that selection and verify it.

03:46.525 --> 03:50.625
And in doing so, we were able to run a study

03:50.925 --> 03:55.545
and we saw a significant improvement over

03:56.585 --> 03:59.745
previous numbers, in, in, in, in this area.

04:00.365 --> 04:05.225
In fact, it, it was almost, ironic that in our,

04:05.245 --> 04:09.145
in my study, there was only 6.6% of the participants

04:09.145 --> 04:11.505
that could not identify the change.

04:12.745 --> 04:16.205
So way more people were able

04:16.205 --> 04:18.325
to identify the change and notice the change.

04:18.515 --> 04:22.805
This technology, if it was being used, you could not

04:23.595 --> 04:26.285
flip votes in a way to change the outcome

04:26.285 --> 04:28.125
of the election such that it would be undetected.

04:28.465 --> 04:32.805
So we did this study in 2021, essentially right

04:32.885 --> 04:34.165
after the presidential election.

04:35.695 --> 04:39.075
So what I did was take the, I'm in the, at the University

04:39.075 --> 04:40.595
of Florida in the state of Florida.

04:41.215 --> 04:44.955
So I took the 2020 Florida presidential ballot

04:45.855 --> 04:47.915
and, used that as the ballot.

04:48.675 --> 04:50.895
And we told the participants

04:51.365 --> 04:55.175
that we were just doing a survey with this new technology

04:55.715 --> 04:58.015
to see voter sentiment after the election.

04:59.285 --> 05:01.865
So they didn't know what we were actually studying.

05:03.285 --> 05:07.305
What happened was then we would randomly flip one

05:07.305 --> 05:11.375
of their selections, and we wanted to see if they notice.

05:12.555 --> 05:15.695
And so the voter would, you know, interact

05:15.695 --> 05:17.375
with the transparent voting machine.

05:17.895 --> 05:19.095
I flipped one of their votes

05:19.715 --> 05:22.055
and there was different scenarios that played out.

05:22.945 --> 05:24.235
I'll walk you through a couple of them.

05:24.775 --> 05:26.715
Uh, one is I flipped their vote

05:26.715 --> 05:28.395
and they immediately notice and they say something.

05:30.235 --> 05:34.615
The other is they notice, but they didn't say anything.

05:36.035 --> 05:39.215
So I, I started to see that I flip a vote and,

05:39.315 --> 05:41.825
and they do this, they make a face

05:41.885 --> 05:43.785
or there was a reaction, but they didn't speak up.

05:44.165 --> 05:45.225
So they'll finish up.

05:45.405 --> 05:47.545
And then when they were done, I say, how did it go?

05:47.545 --> 05:49.305
They say, fine. And I say, well, by the way,

05:49.305 --> 05:50.785
did you notice I flipped one of your votes?

05:50.805 --> 05:54.105
And they say, yes. I say, well, why didn't you say anything?

05:54.415 --> 05:55.985
It's just a study. It doesn't matter.

05:56.525 --> 05:59.445
But, but it was a real election. They would've spoken up.

05:59.945 --> 06:01.325
So that was a scenario.

06:02.145 --> 06:04.525
Uh, so there was these different scenarios

06:04.525 --> 06:06.685
of which they would identify the flip,

06:07.345 --> 06:09.365
and in all cases, they saw it

06:09.545 --> 06:11.725
and they identified it at a record number.

06:12.175 --> 06:15.125
Those who actually, so I'll give you,

06:15.245 --> 06:16.885
I start in reverse basically,

06:17.145 --> 06:20.205
but those who noticed the flip occurred

06:20.225 --> 06:24.005
and spoke up, that was, roughly 36%,

06:25.275 --> 06:28.375
but then 41% noticed, but didn't say anything.

06:28.835 --> 06:31.095
But they were able to identify what was flipped.

06:31.395 --> 06:33.495
So they, they actually saw, but they didn't speak up.

06:33.765 --> 06:37.775
Okay, so that's 77% right there,

06:37.955 --> 06:39.895
who noticed immediately.

06:40.945 --> 06:43.365
So that number is my bottom number.

06:44.145 --> 06:48.605
And then there were some who, say they didn't notice,

06:49.105 --> 06:50.325
but they looked at the paper

06:50.385 --> 06:54.885
and correctly identified the change that was roughly 16%.

06:55.465 --> 07:00.085
And then again, there was only 6.6% that did not notice

07:00.385 --> 07:02.645
and could not identify the flip or change.

07:03.305 --> 07:05.085
It tells me that this is

07:06.185 --> 07:08.965
an improvement on existing ballot marking devices.

07:09.515 --> 07:11.765
This technology that we created,

07:12.005 --> 07:16.195
a transparent voting machine, does not exist.

07:16.535 --> 07:20.205
We created it. Invented it. Now, why am I saying that?

07:20.745 --> 07:24.405
No person in this study, 151 participants, none

07:24.405 --> 07:26.765
of them would've ever interacted with a device like this.

07:27.975 --> 07:29.595
But you know what, when they did, when they voted,

07:30.335 --> 07:32.235
not a single person said,

07:32.835 --> 07:34.075
I don't understand how to use this.

07:34.335 --> 07:35.915
How does this work? How do I do it?

07:36.655 --> 07:39.675
So we know we created something that was usable

07:40.585 --> 07:41.965
for all voters.

07:42.945 --> 07:44.365
We know it's secure

07:44.915 --> 07:48.205
because we had a record number of votes

07:48.765 --> 07:50.245
identified as being flipped.

07:50.705 --> 07:53.565
So advances the current state of technology

07:54.185 --> 07:55.765
in a positive direction.

00:00 - Well, it's an important area that is critical to our democracy, and I think it's essential that we have to do this work to ensure the integrity of elections, confidence of elections, and it's necessary. So that's why we do it.

00:19 - So a ballot marking device, think of it as a computer that's a touch screen where you can go in and touch the screen, essentially marking a ballot, and it prints on paper the your selections or the actual ballot, that is the ballot of record. That ballot then goes into a ballot box or scans at some point and used as part of the tally. So with that said, after you do that, you can do a risk limiting audit, which is a statistical audit, a specific type of audit

01:02 - that allows you to ensure the integrity of the actual tally machines that they, or the scanners that it goes through. There are places where you can use a ballot marking device. There are certain states in the United States that, some of them, the whole state uses a ballot marking device. there are other places that, don't have it statewide. There are places that still require you to mark on paper. If you have a disability, for example,

01:30 - you're visually impaired or blind, you can ask to use an accessible voting machine, which may be a ballot marking device. So these ballot marking devices are being used in elections nationwide, in the United States and in other places in the world. So they are quite popular. If you allow people to write on a ballot, they take the privilege of writing on the ballot, meaning they can mark any way they want, any how they want.

02:01 - And then what do you have? You have a debate. The two, in our case, the United States, two parties looking at the ballot, arguing. That's not of actual vote, that's not what they intended. You have to interpret someone's ballot selection. We do not want to be in that situation. So that's one of the negatives of allowing a hand marked paper ballot is the ability for people to mark any way they want. And then you have to interpret those marks

02:29 - and those interpretations may not translate to what the voter's intent was. So that's an issue. The transparent voting machine was motivated by research that was done at the University of Michigan and Rice University, where they looked at ballot marking devices and they were printing a ballot. And when they print the ballot, they would switch one of the selections that the voter made,

02:55 - and they wanted to see if people would notice. And, in their studies, they were reporting that, for example, in Michigan, only 6.6% of the voters noticed, that their vote was flipped. Well, in that case, that's an insufficient number to really detect the outcome of an election could be changed. So with that in mind, I designed this transparent, voting machine. The idea is that

03:27 - if you could turn the paper almost into a touch screen such that you could touch the screen, but it almost is like interacting with the paper in a way that every time you make a selection, you have to actually touch that selection and verify it. And in doing so, we were able to run a study and we saw a significant improvement over previous numbers, in, in, in, in this area. In fact, it, it was almost, ironic that in our, in my study, there was only 6.6% of the participants

04:09 - that could not identify the change. So way more people were able to identify the change and notice the change. This technology, if it was being used, you could not flip votes in a way to change the outcome of the election such that it would be undetected. So we did this study in 2021, essentially right after the presidential election. So what I did was take the, I'm in the, at the University of Florida in the state of Florida.

04:41 - So I took the 2020 Florida presidential ballot and, used that as the ballot. And we told the participants that we were just doing a survey with this new technology to see voter sentiment after the election. So they didn't know what we were actually studying. What happened was then we would randomly flip one of their selections, and we wanted to see if they notice. And so the voter would, you know, interact with the transparent voting machine.

05:17 - I flipped one of their votes and there was different scenarios that played out. I'll walk you through a couple of them. one is I flipped their vote and they immediately notice and they say something. The other is they notice, but they didn't say anything. So I, I started to see that I flip a vote and, and they do this, they make a face or there was a reaction, but they didn't speak up. So they'll finish up.

05:45 - And then when they were done, I say, how did it go? They say, fine. And I say, well, by the way, did you notice I flipped one of your votes? And they say, yes. I say, well, why didn't you say anything? It's just a study. It doesn't matter. But, but it was a real election. They would've spoken up. So that was a scenario. so there was these different scenarios of which they would identify the flip, and in all cases, they saw it

06:09 - and they identified it at a record number. Those who actually, so I'll give you, I start in reverse basically, but those who noticed the flip occurred and spoke up, that was, roughly 36%, but then 41% noticed, but didn't say anything. But they were able to identify what was flipped. So they, they actually saw, but they didn't speak up. Okay, so that's 77% right there, who noticed immediately.

06:40 - So that number is my bottom number. And then there were some who, say they didn't notice, but they looked at the paper and correctly identified the change that was roughly 16%. And then again, there was only 6.6% that did not notice and could not identify the flip or change. It tells me that this is an improvement on existing ballot marking devices. This technology that we created, a transparent voting machine, does not exist.

07:16 - We created it. Invented it. Now, why am I saying that? No person in this study, 151 participants, none of them would've ever interacted with a device like this. But you know what, when they did, when they voted, not a single person said, I don't understand how to use this. How does this work? How do I do it? So we know we created something that was usable for all voters. We know it's secure

07:44 - because we had a record number of votes identified as being flipped. So advances the current state of technology in a positive direction.

Faculti AI can make mistakes