The unfortunate side effect of technology is its allure. People advocate using tools of technological complexity to accomplish a task, the general assumption being that the more technologically advanced the tool, the better the tool is for the job. What this does not take into account is the ultimate goal one is trying to achieve.

You can take anything as an example of this, let’s say writing. What is the better tool for the job of writing: a computer or a typewriter? The answer seems obvious. The computer is better. But is it? It depends on two things: what you have available to you, and what your final goal is. For me the computer is better in numerous ways. It’s faster, easier, less messy, more portable, cheaper, etc.

But the typewriter does have a few advantages over the computer. The typed manuscript lasts much longer than the electronically stored one. A typed page could conceivably last thousands of years in the right environment. A hard drive degauesses after a few years of inactivity, and even CDs are estimated to break down in fifty years or so. I do have the option of printing this writing, of course, but if I don’t take the time to do that, there’s a chance it’ll be gone forever in a few decades. On the other hand, a typewriter prints the page as you’re writing it.

Furthermore, a typewriter does not require electricity. If you have electricity available to you, a computer makes sense but if you don’t have electricity, a typewriter is quite literally is the next best thing. Neither a computer nor a typewriter is inherently better than the other; it depends on the environment available to the user and their goals.

Which brings me to the subject of electronic voting machines. The arguments put forth defending their utility are the ones used for computerizing anything: speed. They’re faster, but speed is the only thing they have going for them. They are not easier. Indeed in some cases they are significantly more difficult to use for non-computer users to understand. They are not more secure. In fact, they are significantly less secure, which anyone with the smallest bit of computer experience will understand. But most importantly, they are not more accurate—which I’ll show in a minute.

First, let’s consider what the most important goal of the voting process is. The final goal is to elect someone to office. This can only be a just process if every vote is counted. Of course, regardless of which voting system used, there will always be missed votes. It’s just a fact of life. We should do our best to eliminate errors and minimize the ways they can happen. The most important aspect of an election is the integrity of the count.

“The most important aspect of an election is the integrity of the count.”

Proponents of electronic voting often make the argument that hand tallies are unreliable because people make mistakes. This is understandable; people certainly do make mistakes. People made the electronic voting machines, too, which must be considered. However, precisely because people do make mistakes, measures have arisen to compensate for this, and they’ve worked for hundreds of years before electricity existed to make supposedly “better” computers.

One example of human error correction happens in banking. At the branch I worked at, as long as we balanced to within a certain tolerance, no problem. But if we were off greater or less than that amount, the head teller would recount the drawer, as well. This is standard banking procedure.

Banking has its differences from vote tallies, of course. You start with a known quantity of cash and after accounting for credits and debits, you should have another known quantity of cash at the end of the day. There’s no known quantity at the beginning of a vote count. Rather, I think it’s more important to recognize that the principals involved with handling large sums of money and not having any of it mysteriously disappear can also be applied to hand-counting votes.

Speaking of money, let me address the cost of a hand count. Election judges are typically volunteers or if they’re paid at all, it’s a stipend. I think the most efficient way to handle a count would be to have a team of two poll workers in a separate room, tallying batches of votes collected every half hour or hour—or more often depending on the traffic at the precinct. Every hour or so, have them switch off with another pair to prevent mental fatigue, the original team going out to help run the polling place. In this way the hand count is done expeditiously and costs little to nothing. You could even stick a window on the room and let the voters watch other voters’ votes get counted. That’s just good PR.

As far as insuring the integrity of the count, various techniques to account for human behavior can be used. First, each team should be a pair of people who are not affiliated with the same political party. Every ballot should be counted twice by two different teams. (In a busy precinct, this may delay reporting by a few days, but again, expeditiousness, not outright swiftness, is the key to an accurate count. Furthermore, if this is the case, workers could volunteer for counting shifts on non-election days, reducing the onus on the judges.) Every time, and at what time someone enters or leaves the count room, it should be logged, so that the status of the count room can be reconstructed at any point during the count. And finally, those who count a given stack of ballots should “strap” them (perhaps by placing in a sealed envelope), and sign their names, enter the official count for each candidate, and the date and time. In this way, every counted vote can be traced to one of four individuals (two pairs).

Now here’s how we can use technology to help guarantee the count, rather than muddy it up. Digital video runs about 12.6 gigs per hour. If it takes eight hours to count a precinct, that’s 100 gigs of data. We could store a video of the entire count for less than a hundred bucks. It would be even cheaper if we suggest that the judges call out each tally and make an audio recording of it instead. You could fit the audio of an entire state’s vote tally process on an iPod.

I’m not sure of the exact law, but my understanding is that voting materials used in an election must be stored for about two years following an election. This would give more than enough time to audit every election—something we do not currently do, to my knowledge. Six months (or however many deemed appropriate) after the election, collect a random sample of packets and recount them. Better yet, recount them along with the video/audio recording of when they were counted the first time to determine whether it was an honest mistake or deliberate fraud. This audit would ensure that not only was the initial count accurate, but that the process is working properly.

In a theoretical, perfect environment, the computer would always be 100% accurate, but it is ridiculous to believe this would be the case in real life. Even worse, using a computer assumes the election has certain goals that are more important than the accuracy of the count. In a representative democracy like the United States purports to be, there is nothing more important than the integrity of the voting tally. Period. Like the kernel of a computer operating system, the vote is the most important, precious thing we have, and we should use the system that best maintains its integrity.

But why is a computer not the best way for this? I’ve made the case for how hand counting can be made very accurate, but I haven’t shown that computers aren’t necessarily more accurate. There are many arguments as to how computers can be less accurate than people. Here’s two.

There are many things computers can’t do, but counting is not one of them. Computers count very well. Unfortunately, computers only count pristine data very well. Voter ballots—filled out, as they are, by sloppy human beings—are not pristine data. As an example of this, off the top of my head I can think of maybe two dozen ways to make marks on a piece of paper. When you start to consider different variations of the same mark making material (such as color), the number of unique ways one can mark up a sheet of paper numbers in the thousands.

Out of the scores of ways to put a mark on paper, how many of these can a typical scanning machine read? Maybe five or six? Most are designed to work with only one. It’s probably a common one, of course, such as ball-point ink or #2 pencil graphite, but it’s very few. On the other hand, how many could a human read? Almost all of them. If the mark is visible, a human can read it and assuming their faculties are in order, understand its meaning.

Computers are wicked at counting—as long as the data is presented exactly the way they need to see it. Deviate a tiny bit in either direction and they crash and burn. Computers are accurate, but not flexible. Humans may be less accurate without controls, but they are enormously more flexible. But we can make humans more accurate. We can add checks and balances to reduce error, and we still maintain the benefit of flexibility. On the other hand, we are having a fiendishly difficult time making computers more flexible.

Secondly, a million computers are, collectively, much less secure than a million people. I’m not going to go into a conspiratorial discussion of whether any elections were rigged electronically; that’s beside the point. The point is that an army of computers allows one person to rig an election completely anonymously. It allows fraud on a massive scale. Forget whether or not fraud happens, the point is that it’s very possible and very easy to do with electronic machines, and this compromises the integrity of the count.

The problem today is compounded by the fact that private companies make the machines, they don’t allow anyone else to see how the machines work, and private companies also count the vote. How can any normal person verify the integrity of the election? Without calling for a hand-recount, they can’t.

On the other hand, look what happens when we have millions of teams of election judges counting ballots. Imagine you’re Senator Shmuckatelli, looking to rig yourself an instawin in your home town. You want to influence the count, but all the aforementioned human checks and balances are in place. How do you do it? Even assuming there is no audio/visual monitoring and the counts were done in secret so the public couldn’t watch, you’d still have to bribe every single election judge—and you’d have to make sure not one of them squeals. Possible? Maybe—but it’s sure of a hell less likely that you’ll pull it off.

Even disregarding everything I’ve written here, there is still one more reason that hand counting makes more sense. The reason should be clear: we are already hand counting ballots in the current 2008 primaries. So many questions were raised during the New Hampshire primary election, that both the Democrats and Republicans—and everyone else, too—agreed that a recount was the best course of action in light of the questions being raised. In other words, even the winners didn’t believe the count was accurate.

It doesn’t even make sense to say the electronic machines save money, because voter confidence in them is so low that hand recounts are the only thing that can satiate voters’ apprehensions right now. If you think NH is going to be the only state with its citizens calling for a hand recount, you haven’t been paying attention. But if electronic voting machines are so faulty that they have to be checked by hand, anyway, cut them out of the process all together. Skip the machines and their questions, and just do it by hand the first time.

The integrity of the vote count is one of the most important aspects of a democracy. The only time technology should be introduced into the voting process is if it strengthens that goal. Electronic voting machines weaken the integrity of the vote count and, as such, are a threat to democracy.

-Ted