I Want My ATM-Style Voting

Jacob Sullum | February 2, 2007

Florida is ditching its touch-screen voting machines just a few years after buying them, opting for machine-scanned paper ballots instead. Various jurisdictions around the country have made similar decisions, and a bill that has strong congressional support would encourage others to follow suit. As one of the last remaining fans of ATM-style voting, I'm disappointed by its hasty abandonment. For flexibility and ease of use, no other method comes close, which is especially important if you're worried about incomplete and erroneous voting, Florida's two main concerns after the 2000 debacle of hanging chads and accidental votes for the wrong candidate. Despite all the jokes about how the elderly can't handle electronic devices, the uncluttered options and built-in verification ("You have selected Pat Buchanan. Is that what you meant to do?") of a well-designed electronic ballot avoid confusion and promote voter confidence. My most pleasant voting experience by far was with touch-screen machines in Virginia, where the state Senate recently approved a bill that would phase them out.

It's hard to believe that the two main concerns about electronic voting—vulnerability to tampering and the lack of a paper record—can't be addressed without abandoning touch-screen machines. When I use a self-checkout aisle at Home Depot, the machine produces a paper receipt as well as an electronic readout. These machines are heavily used and seem to be pretty reliable (except for occasionally obscured bar codes, which would not be an issue in voting). How hard is it to use something similar for voting, with the receipt dropping into a secure box? Some jurisdictions are adding printers to touch-screen machines to create a paper trail, which seems like the logical solution, although ideally the printer should be part of the original design. In addition to facilitating recounts, a paper record also would help avoid fraud, as would data backup and restrictions on access to each machine's memory. The scanners they will be using in Florida also store tallies electronically, so it's not as if they eliminate the need for data security. With proper precautions, it seems to me, the touch-screen method should be at least as secure as the alternatives.

Help Reason celebrate its next 40 years. Donate Now!

Try Reason's award-winning print edition today! Your first issue is FREE if you are not completely satisfied.

|2.2.07 @ 11:35AM|#

Your comments are sensible which is probably one reason they don't feature in the public discussion of this (easy) issue. The only question I have, is what good does the paper trail really do anyway? Does anyone believe that a manual recount of paper ballots is less likely to produce errors than the original computer count? Can't be. If the issue is computer error, this is something that presumably is easily subject to audit before, during or after voting.

The only upside to this issue is that it matters so little. Politician time spent focusing on this is time not spent on doing real harm somewhere else.

Highway|2.2.07 @ 11:46AM|#

The only touch-screen system I've heard of that comes close to all the benefits of the electronically scanned paper ballots was one that someone posted about that outputs a paper tape within an enclosed container, but with a clear window for you to verify your selections afterward. This allows for voter confirmation of a lasting record if there is some issue with the machine at a later time that would require ballot counting.

The thing that boggles me is that some people who are concerned about machine shenanigans (programming-wise) seem to be perfectly happy with the idea that the machine would print you out a 'receipt' that shows who you voted for, for you to take with you. How that does ANYTHING to prevent tampering with the machine is beyond me (just have it print out what you voted, and change the votes AFTER that step).

Personally, I prefer the optical scan ballots because they combine the best features of the systems: Can be well laid out for ease of use, have almost infinite capacity (just get more pens), instantly report spoiled ballots so the voter has a chance to recast their vote, have the ORIGINAL ballot stored for recounts, and those recounts can be also done electronically. I was disappointed when Maryland moved from optical scan to the touch-screens, and would welcome a shift back.

|2.2.07 @ 11:48AM|#

The only question I have, is what good does the paper trail really do anyway?

Deterrence. If you tamper with the machines, you also have to tamper with the paper trail, else the tampering may be detected.

Christopher Monnier|2.2.07 @ 12:03PM|#

> Personally, I prefer the optical scan ballots because...

Definitely! The ballot problem has been solved, and optical scan ballots are the solution.

tomWright|2.2.07 @ 12:05PM|#

Touch screen self checkouts are run by people with a self interest in honesty.

It is in their interest to NOT cheat their customers, and also to not have the customers cheat them.

Not so in politics, since those running the machines, (the state), are largely immune to retaliation, those with an interest in cheating, politicians, can mostly insulate themselves from direct contact with any cheating and, if in power, can cover up or obfuscate any evidence of cheating.

As a software developer and budding sysadmin I can tell you there are so many ways to game a computer system, via hardware, firmware and software, that the less you use it for systems where cheating is in the interest of those that run those systems, the better. If there is compelling enough reason to use them, the strictest, most open, least secret methods of control and accountability should be used.

If I had my way, all votes would be on hand marked paper ballots, counted by people in a glass booth, on a stage with the underside visible, with bleachers around all sides for spectators to watch.

And even then, I want James Randi and his disciples to investigate it.

Larry A|2.2.07 @ 12:09PM|#

Personally, I prefer the optical scan ballots because they combine the best features of the systems:

Security-wise it doesn't matter whether the voter marks a scannable ballot or uses a touch screen that prints a ballot, the computer does the calculation either way. As long as the ballots are rescannable and stored properly you have the necessary backup.

The advantages of touch-screen are:
1. One choice at a time voting, with as much space as needed to make the choices clear.
2. Instant feedback. "You have marked two choices for this office."

The Wine Commonsewer|2.2.07 @ 12:18PM|#

Highway, the ballot you describe is what we have here in my county. Can't pave the roads but we have hi-tech voting machines. I like the paper output for review but I was perfectly fine with the predecessor with on screen review.

|2.2.07 @ 12:21PM|#

Yes, it's quite possible to fix the security problems of the touch-screen systems and add a paper record. It's even easy. So that makes it all the more incomprehensible that Diebold, the primary maker of these machines, has obstinately fought any and all demands to improve them, thus dooming their own product line.

It's been flat-out weird. Any other manufacturer, faced with dissatisfied customers and a public making well-reasoned complaints about the product, would scramble to improve the product to gain wider acceptance. "I like the idea of an automobile, but I'm unhappy with its reliability and tendency to catch fire." Diebold's response has been a sort of belligerent contempt for its detractors. "You people are too ignorant to understand the excellence of the product. We're going to keep making the things the way we feel like making them and you're going to buy them because we say so."

The comparison to ATMs is apt. ATMs are a widespread technology handling a far more complicated transaction. More people probably hit up ATMs every day than ever bother to vote. The failure rate is low and it generates a paper receipt. So why couldn't the touch-screen systems be fixed? Because Diebold just didn't want to, that's why.

|2.2.07 @ 12:29PM|#

It's hard to believe that the two main concerns about electronic voting-vulnerability to tampering and the lack of a paper record-can't be addressed without abandoning touch-screen machines.

The solution is so obvious, I am amazed I don't see it more often: The touch-screen voting machine should output the paper ballot itself!

You check in, get your ballot from the poll worker, put your ballot in the machine, vote on the machine (with all the attendant assistance technology can provide), tell the machine you are done, get your ballot spit back at you, and take the ballot back to the poll worker to be put in the ballot box.

All the advantages of optical scanning. All the advantages of electronic voting.

Even though an electronic voting machine makes it easier to tally votes outside of human vision, it doesn't mean it has to.

|2.2.07 @ 12:43PM|#

Optical scan is pretty hard to beat.

The voter writes directly on the ballot, so there's no way to interfere with vote registration electronically.

The software (counting dots) and data collection (printing ballots) are very well separated. It can be done with touch-screen too, but it's harder to enforce.

It's a lot cheaper, and it doesn't lead to long lines when turnout is heavy.

That Dude|2.2.07 @ 12:44PM|#

Never voted on a touch screen, but I never really understood all the voter fraud scare stories. Let me tell you folks a little story:

In the 1996 election I was a "Republican poll watcher" (don't remember my exact title) in NYC, due to a college community service requirement. There (as in many other jurisdictions) they used those big old mechanical machines to count votes. The ones where you walk in and a curtain swings behind you and you move switches to choose your votes and then you move a big lever which records your vote and swings the curtain back out, and then on to the next person. Well, these things were wrong all over the place. Numbers didn't add up (i.e. different parts of the machine would say a different total of people had voted etc. etc.). I went up to the head of the polling station and pointed this problem out. He gave me a look like, "Yeah? Whaddaya want me to do about it?". And it was just kinda ignored.
The Republican poll watcher supervisor-guy came by later, I told him about it and he said it happens all over the city, so they note it on some little complaint form that nobody ever looks at. (I must say he was also quite dismayed that a rather questionable looking longhair like me was a the GOP representative, but I digress....)

So with that long and pointless story, I'm just trying to say.... Really, are these electronic things so much worse than what we got?

Warren|2.2.07 @ 12:51PM|#

another check in the optical scan collumn

|2.2.07 @ 1:00PM|#

That Dude: my mother had the depressing experience of being involved in a recount and said much the same thing. I think the difference is the assumption that screwups, broken mechanical devices and the like are impartial: they will even themselves out over both parties, but that the electronic machines are inherently vulnerable to a planned, centralized attempt to jimmy the vote. Rightly or not, that's how people feel.

Back when I was in school, an engineering professor told me that he had been asked by a sportswriter after a close football game to estimate the likely error in measuring the proper "spot" of a football after a play. After reviewing some tapes and going down to the field with chains and all, he told him that the nearest reliable accuracy was plus-or-minus one foot. In other words, any "measurement" with the chains of less than one foot could just as easily gone the other way.

But the point of the system is not to create perfect accuracy (they aren't building a piano) but to create an agreed-upon system of resolving the dispute of whether or not it's first down or fourth down. The system works because after the measurement the coaches and players, ninety-nine percent of the time, solemnly accept the judgement of the officials and the game goes on.

So it is with voting. Whether or not it is completely accurate (it never has been) or whether or not this is some kind of minor-league fraud going on (there usually is) the primary purpose of the system is to create confidence in the mind of the public that the system is fair. The electronic voting machines failed that critical test and I maintain that Diebold's eager, politically-connected insistence on early, widespread rollout of machines that simply weren't ready yet doomed the whole project.

|2.2.07 @ 1:03PM|#

another check in the optical scan collumn

So that's zero checks in the optical scan column and seventy-five checks for touchscreen without paper trail.

tomWright|2.2.07 @ 1:03PM|#

For creating the ballots, an electronic machine can be useful. It can print out clearly marked ballots that avoid the mismarked, double marked and hanging chad problem.

They would be checked by the people that created them, the voters, who have a self interest in mkaing sure they are correct.

But for counting them, I want people. For the simple reason that a person can be held accountable. A computer system though, can not. The programmers, administrators, designers, users, manufacturers, specification writers, and everyone else involved can point fingers at all the others. A machine is impersonal and can be made to take the blame for those that suborn the process.

It is harder to do that with people. A person that knows they can be blamed for a miscount will count more carefully, and will be harder to use as a tool of a cheater. That is not to say it would not happen, looking at history says it likely will. But at least we know it was people that can be held accountable if we wish to do so.

Highway|2.2.07 @ 1:20PM|#

metalgrid|2.2.07 @ 1:28PM|#

That's funny - the self-checkout machine at the Home Depot closest to me keeps insisting that I picked debit when I picked credit and to make matters worse, the number 5 on the touch screen when I go to enter my debit password refuses to register until I press on it several times. At which point I lose track of how many times it registered 5.

At least the receipt always matches the value on the screen at time of purchase, but I'm quite sure such a machine could be tampered with to register one thing and print out something completely different.

thoreau|2.2.07 @ 1:30PM|#

I'm fine with a touch screen that prints out the paper ballot, and the voter can examine the paper ballot before deciding to put it in the box.

You get all the advantages of each method. You get redundancy, which is always good from a security standpoint. Sure, paper ballots can be stolen or tampered with, but I'll take redundancy any day. Besides, a process with paper ballots can be more transparent (to the non-expert) than an all-electronic process.

|2.2.07 @ 1:30PM|#

Does anyone know how the blind cast their ballots in districts with touchscreen machines? Or, now that I think about it, anywhere? Are they all going the absentee ballot route?

That Dude|2.2.07 @ 1:32PM|#

James: That's a perfectly reasonable point. For the record I'm sure the screwups where I was were mechanical. But there was constant allegations among the election watchdogs that some machines that were really wrong (really wrong as in off by hundreds of votes, mine were off by 10 or 20) had fallen victim to election fraud. Machines with big problems happened to usually be in poor, minority areas... whether all this was Republican racsim-lite blowhardness, or a real problem with basis in fact, or just because poor areas got the cruddiest machines, I dunno. In the end, this was Manhattan and went, like, 90% for Clinton so who cares. Just sayin'.

Also for the record: I never understood why people accepted the whole football ball-placement situation.

That Dude|2.2.07 @ 1:40PM|#

Son of a!: At my now infamous lower Manhattan polling place we had a blind person come in. What they did was send me and the Democratic poll watcher into the machine with her. She said who she wanted to vote for and we pulled the levers for her. So much for a private ballot, I guess!
Sketchy stuff like this is why I'm so skeptical with all these people who see the polling place as this font of pure, sacred democracy about to be ruined by evil touch screen machines. It's just another government place where everythings kinda effed up and people muddle through.

So the moral of the story is don't volunteer at a polling place if you want to continue to trust democracy.

Jack|2.2.07 @ 1:54PM|#

TomWright is, well, correct. From a governance standpoint, the system security plan for an application like this would be outrageously difficult to implement. It's tough enough to keep the bad guys out of your system, but when you want to defend against the adminstrators? I don't even know where to start on that--it's probably undoable. This would be such a huge project to run, and the cost would be immense.

I like having a technology solution facilitate the voting--controls that say, as stated above by a few people, "You have chosen George W. Bush for President. Are you f'ing stupid?" or something along those lines covers governance well, and it ensures that you are getting the desired output when you vote. From there, well, spit out the paper and go back to old days of counting them up by hand--that way, you know exactly what you have.

Sure, maybe you can use technology to count the output paper votes, but then you still have as a back-up the paper trail. ATM style voting, from a technology standpoint, is probably prohibitive.

Something needs to change, though, since where I live in New York, the gargantuan, curtain drawing, lever pulling machines are so old, the company that built them is long closed up. We had long lines because so many machines have broken over the years, with no opportunity for replacements, that some stations had only one machine.

|2.2.07 @ 1:58PM|#

ATMs have wide acceptance because there are multiple means of verifying their accuracy. First, I have the option of printing out a receipt. Second, I get a monthy statement where I can review all transcations. This gives me the ability to challenge any transactions that I don't agree with. In addition, the law requires the bank to cover any losses that occur due to fraud, so the bank has a very strong interest in proving that all my transactions are accurate.

Given all this, I have co-workers (in the engineer field) that refuse to do online banking because the authentication processes are still immature.

There is no evidence of any kind that says that the electronic voting machines have the same level of integrity as electronic banking systems. More importantly, I have no way of verifying that my vote was counted the way I intended to be counted after the counting has been completed.

So even though I am an engineer with a great deal of respect for electronic information systems, I will refuse to use an electronic voting system. If my county starts using them, I will start voting by absentee ballot.

|2.2.07 @ 1:59PM|#

MikeP, the significant drawback to your idea - using touch-screen machines to generate paper ballots that are then scanned - is the investment required. Now you need to have both the touch screen machines AND the optical counting machines (purchase, replacement, and maintenance).

Yes, it's more expensive than optical-scan only. But I don't think it's more expensive than electronic voting only. With electronic voting, some centralized machine needs to accumulate all the results off all the flash memory or whatever mechanism performs the distributed electronic collection. In the electronic marking scheme as I describe it, the collection mechanism isn't a flash memory, but paper ballots. So instead of accumulation through plugging in flash memory, you have accumulation through optical scan. The price difference should not be significant.

And yes, it would have to be optical counting machines, because I'd want that review step for the voter, so that they can see that the ballot is correctly filled out. No barcodes or other machine code.

Absolutely. The paper ballot marked by the machine would be completely human readable.

To clarify: All the electronic voting machine does in this scheme is offer an interactive interface. The final ballot is identical to what you would get had you simply said to the poll worker, "I'd like to mark mine by hand, thank you" -- which would always be an option.

liberaltarian4kucinich|2.2.07 @ 1:59PM|#

Well this proves it- Gore is going to run.
The Repukes are bringing back the paper ballots so they can steal Florida fronm him again

|2.2.07 @ 2:00PM|#

Computer-generated paper hardcopy is just as susceptible to fraud as pure electronic systems are. Same goes for the scanner itself, which may not scan what you feed it. If you know what the acceptable error rate is, you can even design pseudo-random error so it's harder to detect.

Read Ken Thompson's 1984 paper _Reflections On Trusting Trust_, published in the August 1984 issue of _Communications of the ACM_, on why a review of the voting machine's source code isn't enough to ensure it hasn't been compromised.

|2.2.07 @ 2:15PM|#

Computer-generated paper hardcopy is just as susceptible to fraud as pure electronic systems are.

How so?

I would say they are just as susceptible to fraud as pure hand-marked optical-scan systems are, but they are much less susceptible than pure electronic systems.

|2.2.07 @ 2:16PM|#

The primary advantage of optical scanning is that it produces an human-readable record that can be preserved and protected. If the output of a scanner is challenged, the paper records can be run through a different scanner that has been maintained in a way to guarentee its integrity.

Electronic voting systems could be improved if they produced human-readable outputs that can be preserved. It's not necessary to actually read them at the time of voting, so long as they can be read by a dissimilar system than the original voting system.

However, as Bob Smith points out, if the original voting system can be spoofed in a way that is not easily recognizable, then the paper records are of no value.

|2.2.07 @ 2:25PM|#

However, as Bob Smith points out, if the original voting system can be spoofed in a way that is not easily recognizable, then the paper records are of no value.

Which is why it should be the case that the paper records are the output.

No flash memory, no disk, no network, no paper trail, no verifiable receipts, no spoofing, no nothing. The only output of the electronic machine is a single ballot per voter, as assuredly as if it was marked by hand. In fact, those who'd rather not use the machine can mark it by hand.

Larry A|2.2.07 @ 2:27PM|#

What we could do is manually count voters.

1. Everyone who wants Hillary Clinton for President fly to Hawaii.

2. Everyone who wants John McCain for President fly to Alaska.

3. Seal the borders.

|2.2.07 @ 2:29PM|#

OMR is the obvious winner. And James, I think an instant replay challenge of the spot of the ball is nearly as dangerous to civilization as the recount-mania we got thanks to Al Gore. You shouldn't allow the system to be doubted despite the flaws.

|2.2.07 @ 2:38PM|#

No flash memory, no disk, no network, no paper trail, no verifiable receipts, no spoofing, no nothing. The only output of the electronic machine is a single ballot per voter, as assuredly as if it was marked by hand. In fact, those who'd rather not use the machine can mark it by hand.

Unless you intend to count more than 120 million ballots by hand, then you have not eliminated the opportunity for subverting the counting process.

My guess is that the probability of an incorrect outcome from handing couting ballots is higher than the probability of conspiracy to alter the outcome of electronic voting (once the appropriate means of verifying the count is in place).

Matt|2.2.07 @ 2:39PM|#

As someone who remembers using optical-scan technology in high school (it was only a few years ago), count me firmly in the touch-screen camp. If it can't even grade tests completely accurately, I don't want it counting votes.

Plus, I've seen the touch-screens they use in Florida and I really don't see the problem.

|2.2.07 @ 2:52PM|#

That Dude:

Thanks for the info. Here in Ohio, secret ballots are already a thing of the past, though. When I went to vote, my touchscreen machine had no curtain, and the screen was facing the line of people waiting to vote.

Jack|2.2.07 @ 2:59PM|#

Simply put, the reason that you can't rely on technology for this is that there is NO way to verify that the vote you cast is the vote that is counted. Your "receipt" would be NO indication that it is also the same vote that is counted. ANY IT system could be coded in an almost innumerable amount of ways to shave votes off, dis-count votes that fit a certain pattern, or all sorts of devious things.

The key thing is that there is no independent verification and validation without an artifact that demonstrates compliance with any fair voting standards. Any technology that doesn't produce that artifact--in this case a paper ballot is most obvious--would not be a viable solution.

The most you can hope for is to use technology to facilitate the voting under human goverance.

Jim Lippard|2.2.07 @ 3:16PM|#

Paul McMahon asks, "what good does the paper trail really do anyway?"

Well, for one thing, it would have immediately allowed us to distinguish whether the cause of the undervotes in Sarasota County was user error due to ballot design or a problem with the voting machines themselves.

|2.2.07 @ 3:24PM|#

Setting costs aside, I can see a purely electronic voting system that would be trustworthy.

First, the voter goes to his/her registration place and is issued an ATM-style smart card once he/she has provided proof of age and residence. The smart card is anonymous providing only a psuedo-random voter ID number and a precinct.

When the voter goes to the polls, he/she inserts the smart card into the voting maching. A lovely graphical user interface walks the voter through the voting process with multiple confirmation points and a final summary page. At the end, the voters inputs are recorded on the smart card.

The voter then swipes his/her smart card through three separate card readers developed independently by three separate companies.

The results of the voting process are accumulated in three independent locations, and the vote totals are transmitted in human-readable only format to the person responsible for validating the election results.

If the three vote totals differ by a statistically significant amount, it automatically triggers an investigation. Severe civil penalties are impossed against any company showing signs of incompentence. And most importantly, criminal charges are brought in cases of fraud.

This process would preserve the intellectual property rights of the companies that provide voting equipment, while providing a means to verify election results and punish "bad acts".

|2.2.07 @ 3:33PM|#

It's really not that hard. You want electronic voting? Here's one easy way:

1) Vote on an electronic machine.
2) Machine spits out human-readable receipt and tallies vote electronically.
3) Receipt is placed in "Machine X" locked bin next to voting machine X, after voter glances at paper to ensure vote is correct.
4) Each election cycle has a random audit of 1% of all precincts. The audit is done by comparing the individual machine-reported tallies to a hand-count of the paper receipts. Then the precinct total is compared to the electronic precinct total.
5) If there is a significant discrepency between the machine tally and the paper audit tally, the company that made the electronic machines pays for a revote.

thoreau|2.2.07 @ 3:43PM|#

carrick-

1) What about the machine that puts the data onto the smart card? If that machine is rigged, the 3 competing readers will all give the same response.

2) What if a voter, for whatever reason, doesn't swipe through all 3 machines?

The second concern is admittedly less significant. But the point of the first concern is that you still have a choke point, a machine that puts the data onto the smart card.

The machine can say "I'm writing onto your smart card that you voted for Candidate A? Is that correct?" The voter verifies, and then the machine writes "Candidate B" onto the card instead, due to either malice or incompetence (or a cleverly malicious supervisor who knowingly selects an incompetent vendor, whatever).

The basic fact is that a voter has to submit SOMETHING with information on it. Punch card, scantron, smart card, stone tablet, chad, Florida license plater, whatever. Somebody else has to read it. The goal should be to put that information in a medium that is not easy to modify without leaving a trace (leave aside chain of custody issues, for the moment). An electronic storage medium is (usually) easy to modify without a trace.

Paper is a good bit harder to modify, especially in large quantities, without leaving a trace. Sure, given enough time you can forge just about anything, but if you have a good chain of custody you can leave would-be fraudsters with very little time to do mischief unsupervised. (Unless the supervisors are themselves corrupt, but with enough corruption the storage medium is probably irrelevant.) Electronic data storage is generally easier to mess with, because computers can handle data on microsecond time scales or smaller, while pens and erasers and white-out and whatever else move a bit more slowly in human hands.

This is why I'm a big fan of paper ballots. Sure, chain of custody is an issue, but chain of custody will ALWAYS be an issue.

One way to reduce chain of custody issues is to feed the paper ballots into a reader at the polling place as they are submitted and store the results on a hard drive. At the end of the day, the data on the hard drive is immediately transmitted by some appropriate means. The hard drive and paper ballots are delivered to the election officials by separate people. You now have 3 records: The hard drive, the transmission made BEFORE transit, and the papers (carried by people who didn't have the hard drive). Redundancy of media, "before" and "after" counts to deal with chain of custody issues, and a medium that both the voter and the counter can easily read without any intervening technology (e.g. an electronic device that somebody could secretly program).

That Dude|2.2.07 @ 3:49PM|#

Last post for me I swear... But one thing I've never heard discussed with electronic machines is write-in candidacies. As someone who frequently dislikes everyone running, this is something I mildly care about. If every precinct has electronic ATM style machines would this heavily discourage people voting for write in cadidates? If so would this be bad? My personal answers are "yes definitely" and "kinda sorta".

|2.2.07 @ 3:53PM|#

The machine can say "I'm writing onto your smart card that you voted for Candidate A? Is that correct?" The voter verifies, and then the machine writes "Candidate B" onto the card instead, due to either malice or incompetence (or a cleverly malicious supervisor who knowingly selects an incompetent vendor, whatever).

Can be solved by the reader displaying a summary page with an request to confirm the information.

What if a voter, for whatever reason, doesn't swipe through all 3 machines?

Only an issue if a significan number of people don't swipe at all three readers. Which can also be monitored by the staff at the election site. "Excuse me sir/madam, I think you missed a reader"

The goal is not to produce a 100% foolproof system, but to produce a system that would require a major conspiracy to defeat. Where said conspiracy would be discoverable by other means.

Gray Ghost|2.2.07 @ 3:59PM|#

I was waiting for Thoreau to chime in...good.

Add Morat20's 1% random audit step to Thoreau's process and that would seem to be as good as we can get. That Dude and Thoreau: since I've read that both of you have worked in the elections process, do you know if there is a random audit process similar to the one that Morat describes, and if so, what happens when it indicates chicanery/ordinary error?

I'm sympathetic to the idea of there being a run-off election whenever the top candidates' totals are practically indistinguishable (say 0.01%).

thoreau|2.2.07 @ 4:08PM|#

Gray Ghost-

I only worked there on Election Day. I know very little about audits after the fact, except that there were apparently quite a few of them done as a matter of procedure. I don't know how thorough these audits are in practice, and I assume they vary significantly from one locale to another.

The one thing I do know is that, contrary to what some said in 2000, they don't stop counting ballots once the winner of a state-wide contest is clear. The networks may stop paying attention at some point, but the count goes on. In large part for the sake of completeness (i.e. check everything, so that it's harder for fraud to slip through), but also for local races: The statewide results may make it clear that candidate A won the electoral votes by a huge margin, but a few precincts could make all the difference in who the next member of the local School Board is. So the counting does not stop.

Gray Ghost|2.2.07 @ 4:44PM|#

Thanks for the info, Thoreau.

I agree that all of the ballots should be counted, for the reasons you've described. I'd add to your reasons the idea that fully counting helps absentee voters feel their votes have value, which should help with turnout, which is a good thing.

In case I was unclear, my comment was that there should only be a run-off if the top candidates' totals were practically indistinguishable and that the run-off should only be for that particular race.

I wonder who would have won Florida in 2000 if there had been a run-off to decide what was practically, a statistical tie?

That Dude|2.2.07 @ 4:48PM|#

Gray ghost: OK, I lied about the last post thing. I, too, only worked on election day so I don't know what happens after that. But I don't see how any kind of real audit can be done with those mechanical machines.
This was 10 years ago so someone with more recent experience may be able to correct me. But as I remember there is a manual count of how many people use each machine, then each machine has a mechanical count of how many people used it, as well as a count for each candidate (and a count for people who choose to not vote in a particular race). So, in each race the votes for all the candidates (plus any non-votes in the race) should equal the machine's count of how many people used the machine which should equal the manual count of how many people used the machine. (These numbers never agree, often aren't even close which was my original point).
At the end of the day the number of votes for each candidate is written down and sent up to the main office.

I don't see a way to audit original voter intent in this process. Meaning, yeah, you can go back and double check that you got the correct numbers off the machine (assuming you save the machine as-is) but there's no way to verify the machine recorded a Clinton vote for Dole or vice versa. You'd have to break the machine open and see if theres signs of mechanical tampering, I guess.

|2.3.07 @ 2:50AM|#

It is really astounding to me that people in this day and age still think tabulating data can be made more accurate by adding a manual step in that involves people carrying paper around...

Computers work people. Get over it.

|2.3.07 @ 9:00AM|#

Computers are fast. They aren't terribly transparent, however.

|2.3.07 @ 12:59PM|#

I'm continually suprised that localities have been tricked into spending so much money on these atm-style machines. They make no sense from just a fiscal standpoint.

In my locality, we use the scanned cards. Here is the deal with these scanned cards : when a polling place has a lot of voters and is backing up, with a scanned card system, the cost of increasing the throughput of voters is one more folding table plus some more pencils, and maybe another set of volunteers. with an ATM machine, you need another machine, which you don't have.

|2.3.07 @ 3:07PM|#

I'm continually suprised that localities have been tricked into spending so much money on these atm-style machines. They make no sense from just a fiscal standpoint.

Actually, the high price tag is the one reason I'm not surprised by any of this...

|2.3.07 @ 8:40PM|#

Paper systems can be broken, as long as you can do it in plain sight. Just witness the election of Christine Gregoire (governor - Washington state). They just kept recounting, and finding "missing" ballots, and recounting, and finding more ballots, until she had enough to win. Then they stopped recounting.

|2.4.07 @ 6:26PM|#

I think Diebold is great!

Having only one key that works in all voting machines across the country is very helpful -- if California loses their key, they can borrow Florida's!

It's interesting how much more tamper-resistant ATMs are than voting machines. It's as if Diebold thinks it's important to secure the bank's money but not important to secure the country's democracy.

Add to that its special abilities to register negative votes, overwrite its software by inserting a PCMCIA memory card and pressing a button, etc, and you begin to wonder if it was made tamperable on purpose...

It's OK, though, lots of people feel like ordinary people really can't be trusted to pick our elected officials anyway. It's really comforting to know that Diebold is looking out for us. Now, back to American Idol.

|2.5.07 @ 12:34PM|#

Wow. I implore the brain-stormers here to educate themselves. Smart people have been thinking about this problem for many years. It's the cornerstone of our democracy. The fact that we do such a lousy job at elections and vote-counting is a national disgrace. Here's a bit from Rebecca Mercuri's homepage (does anyone here even know who she is?):

I am adamantly opposed to the use of fully electronic or Internet-based systems for use in anonymous balloting and vote tabulation applications. The reasons for my opposition are manyfold, ...

Why have none mentioned Canada? Simple paper ballot and they hand-count the entire country in about 4 hours. Count me firmly in the paper ballot category. Optical mark-sense ballots are a decent technology.

Reason Magazine

Get Reason E-mail Updates!

Site comments/questions:

Media Inquiries and Reprint Permissions:

Editorial & Production Offices:

advertisements