Larry Moore – TRANSCRIPT
I’m here today to talk about election transparency. And to answer and to bring to your attention a technology that I think will answer a seemingly intractable problem, one that’s very difficult for your mind to even get around and it is this: in a close election where hundreds of thousands of ballots have been cast, what technology could be brought to bear to allow election officials charged with deciding the election to see the result and to know that every vote that was cast was counted as intended.
And to do that in just a few minutes. So how would that be possible? Six years ago, my only connection with elections was as a voter. I was surfing the channels, I had a clicker, and I landed on an HBO documentary called “Hacking Democracy.” And in dark tones, this documentary cast doubts on the integrity and the legitimacy of the 2000 presidential election in Florida. And in a dramatic demonstration at the end, the count of the vote was shown that it could be manipulated and the outcome changed.
Well for me, this was like a small earthquake. And I thought that all of the assumptions that I had about the underpinnings of democracy were at that moment shaken a little bit. And though I now realize that that was a highly sensationalised video, it did get me thinking. But I continued to think about that video. And about two years later, someone pointed me in the direction of an experiment that was done in Humbolt County, California.
There, about 100 miles south of the Oregon border, a team of passionate citizens interested in election transparency teamed up with the county clerk to conduct an experiment with a very interesting idea. So here’s what they did. They scanned the ballots that the voters voted, there was about 60,000 of them, and they put the images on the Internet and from there, they let citizens count the vote. And I was really fired up. And in that “Hacking Democracy” video, that never left my brain, I remembered that there was this one guy, Ion Sancho, who organised this dramatic demonstration at the end of that video.
And so I called him up right out of the blue and I told him about what I’d learned in Humbolt — he was very curious because he had heard about it too — and I started hinting around, wouldn’t it be cool if I could get some of your ballots to scan. I’m pretty sure he thought me nuts. He said, “You know, when you’re ready I’ll give you access to the ballots.” And so in June 2009, I found myself on Railroad Ave in Talihassee, Florida having rented a scanner, hired an operator, and at the end of 8 days, I had a disk drive that had the images of 150,000 ballots from the general election in Leon County of 2008.
And guess what? Nearly every single one of them was worthless. So why was that? Well, it turns out, me not knowing anything about scanners, there was a setting in the scanner software that I didn’t know about that said, “Turn 3-hole drill marks into white space.” Well you know what a fully darkened oval looks like? It looks like a 3-hole drill mark. And so I had a white space in the place of votes. So I was, I guess, bloodied but unbowed.
I started attending seminars and I began to meet the people in the election industry. And here’s what I found: I found the most incredibly dedicated, often patriotic, hard working individuals that were awash in a sea of ballots. These people were drowning in paper. The technology that every other paper-intensive industry had come to enjoy had bypassed these guys.
And so the march of technology had sidestepped the election industry. I pose this question: “Why do we have two methods of counting votes?” You probably don’t even think about this. But in regular elections, we use machines to count the votes. And in about 99% of the US vote. But in close elections, we have to do hand counts. Why is that? Well, it’s because today’s voting systems cannot resolve voter intent. It’s not just counting the ovals, it’s looking at the intent of the voter. And that is the law in almost all but maybe one state, it’s that voter intent trumps in a recount.
So let me show you what I’m talking about. Here we have a fully darkened oval and a machine and a human will agree on that count. But what if the voter misses the mark, and you’d be surprised at how often this happens. Well, the machine may or may not get it, we don’t know. The machine looks like a black box but a human would judge that as just a misplaced oval and give the candidate that got that vote the credit.
But here’s another one, sometimes we call this a hesitation mark. So the guy’s reading a long ballot question and he drops his felt-tip pen and he makes this mark and then he decides maybe he’s going to vote for the other guy. Well, in this case, an election official has to look at all the marks on that ballot to see if that one’s consistent. And if it is, he’ll give this choice the vote.
But now, here’s another problem. Here is what’s called an over vote. On a vote for one, like Yes/No, this voter, to the machine, marked both ovals. But when a human looks at this, they see clear intent and the machine cannot resolve this. And so in a close election, we have to go through this arcane, time-consuming, expensive proposition of counting every ballot by hand.
And what does this look like? What does this look like in a small election, in a small county, in Florida? These are the marginal marks from one election in one county. So what if you could have just one way of doing this? Whether it be close or not close? A friend of mine introduced me to the Secretary of State, Kurt Browning, in Florida.
I visited Kurt and I told him about an idea of an automated independent audit system and how that might be a way for me to get started in bringing the skills that I had to the world of elections. He introduced me to Susan Gill, who is the supervisor of elections, Ion’s peer in Citrus County, Florida, down on the west coast below the Panhandle. And she listened to this idea of the automated independent audit and she paused and then she asked a zinger of a question. She said, “OK, so tell me this, what would happen in the heat of an election if your accurate audit system shows a discrepancy with the voting system? Which one would you ask me to believe?” Well, this was a really tough question and to tell you the truth, I had no snappy answer. Something in the back of my head said if you could see it, you’d believe it.
But it was a couple of months later, and I will never forget this day, that I’m standing in my kitchen and boom, this idea pops into my head. It was not a way of counting the votes differently. You have to get this, it was a way of moving them around so that they would order themselves in a way that you could see the count with your own eyes. We’re going to start with this bag of marbles. So count them up, maybe count them twice to make sure you’ve got them.
But let’s make it easy. I’ll give you a grid and lay these marbles on that grid so that you can see at the bottom that the exact count is nine and you don’t have to count them again. Now let’s lay votes against those where dark marbles get dark votes and white marbles get empty votes. And now we move them around. Let’s sort them by their darkness.
So that we can see that the last obvious vote is at number 7 and that is crystal clear to us. So let me move to the warehouse in Leon County and here’s 293,000 ballots in warehouse shelves 16 feet tall. Like in Humbolt, what we are going to do is scan the ballots into image files. And this is the first step of what we are going to do. And now, believe it or not, we are going to electronically scissor out the vote ovals for a given candidate.
In this case, we’re going to choose Barack Obama. And we’re going to scissor out three of his vote targets, just as an example, and lay them on a web browser window. And now we’re going to do this at scale, and sort them by their darkness. And here’s where in my mind’s eye the votes begin to dance. And so when we get down to the last obvious vote, we can put our cursor on that oval and see the count of that election.
And to preserve transparency we can link that oval instantly to the ballot that it was on. It is said that there are two purposes of an election. I’ll posit a third. The first is to select the winner with finality. But the second is to confer legitimacy on that winner and on the election process.
Some might say that the purpose is also to convince the loser and their supporters that they fairly lost. But the role of technology in elections, I think is not for technology and software to decide an election. Rather it is to assist human judgement in the following ways: To present the evidence of the election as clearly and succinctly as possible to election officials and the widest possible audience. And then to record for others to see the decisions that those election officials were entrusted with the process, the decisions that they made to decide the outcome. So that now when you are looking at a recount like the one that’s going to go on in a few days in Virginia, that you will look back on this talk and you will see that, I hope you will think, that this idea that you heard today is an idea worth spreading.
Thank you very much.