Diebold Admits That Its E-Voting Machines Drop Votes

Voting Machine manufacturer Diebold admitted earlier this week that there is an error in its voting machines that could potentially lead to dropped votes.

A voting system used in 34 states contains a critical programming error that can cause votes to be dropped while being electronically transferred from memory cards to a central tallying point, the manufacturer acknowledges.

The problem was identified after complaints from Ohio elections officials following the March primary there, but the logic error that is the root of the problem has been part of the software for 10 years, said Chris Riggall, a spokesman for Premier Election Solutions, formerly known as Diebold.

The flawed software is on both touch screen and optical scan voting machines made by Premier and the problem with vote counts is most likely to affect larger jurisdictions that feed many memory cards to a central counting database rapidly.

Riggall said he was “confident” that elections officials through the years would have realized votes had been dropped when they crosschecked their tallies to certify final elections results and would have reloaded cards so as not to lose votes. Ohio Secretary of State Jennifer Brunner has said no Ohio votes were lost because the nine Ohio counties that found the problem caught it before primary results were finalized.

The article doesn’t seem to indicate that this has affected any elections. But on the other hand, how can we be sure? If it took Diebold a decade to fess up to this error (which apparently still plagues its machines), what other errors are out there in our e-voting systems?

(link via Engadget)

Image Credit: Joseph Hall

FILED UNDER: US Politics, ,
Alex Knapp
About Alex Knapp
Alex Knapp is Associate Editor at Forbes for science and games. He was a longtime blogger elsewhere before joining the OTB team in June 2005 and contributed some 700 posts through January 2013. Follow him on Twitter @TheAlexKnapp.

Comments

  1. Michael says:

    Nothing like closed, unverifiable, unaccountable solutions to preserve our democracy. Somebody is clearly doing their job horribly wrong.

  2. Anthony says:

    Can someone explain to a Brit why you don’t just use an old school voting-slip/pen/cross-in-the-box system? I understand that on your side of the pond it can be relatively complicated, but our local elections involve increasingly long ballots with multiple boxes to tick and people seem to manage it ok using the old ways and we’re still able to get our vote counts done promptly.* Is this just a matter of technology for technology’s sake?

    *Our postal vote system, on the other hand, is coming apart at the seams.

  3. Bithead says:

    Nothing like closed, unverifiable, unaccountable solutions to preserve our democracy

    Ummm. Wait a tick.
    Look again:

    Riggall said he was “confident” that elections officials through the years would have realized votes had been dropped when they crosschecked their tallies to certify final elections results and would have reloaded cards so as not to lose votes.

    In other words the reason the errors were not found until now and teh reason Diebold didn’t know about the bug until now, was because the people responsible for cross checking the votes, didn’t.

  4. Triumph says:

    Is this just a matter of technology for technology’s sake?

    No, dude–its just easier to rig the elections with these systems.

    If David Cameron had any sense, he would deploy this voting method once the tories get back in power.

  5. jeff b says:

    Anthony: we had a debacle in 2000 when the election counting in Florida was botched. They used an antique mechanical system which didn’t work properly and the outcome of the election was in doubt for a month. It was finally settled at the Supreme Court rather than at the ballot box, causing a great deal of animosity amongst the voters for the defeated party. To distract the nation from that, the government passed the “Help America Vote Act” which requires any precinct still using obsolete vote counting machines to upgrade to something else, usually touch screen machines or paper ballot scanners. Unfortunately, HAVA also requires this of any precinct where the old manual counting mechanisms work perfectly well. The result is that every place in the USA now uses optical scanning or touch screen direct entry.

    Thankfully the scanners are much more popular. This is a good compromise which allows the easy computerized tabulation of votes, alerts the voter to any mistakes made (skipping a race, or voting twice in the same race), and can easily be counted by hand in case of problems. It is unconditionally superior to hand-counted paper ballots.

    Like many other parts of the law, HAVA was a mistake made under the pressure to “do something” about the Florida fiasco. It will be difficult to get rid of it, but it might be easy to amend it for the better. For the next couple of years we will unfortunately be busy with higher domestic priorities, I’m afraid.

  6. Michael says:

    In other words the reason the errors were not found until now and teh reason Diebold didn’t know about the bug until now, was because the people responsible for cross checking the votes, didn’t.

    Hmmm, I read it to say that the errors happened, were noticed, the process was repeated (perhaps more carefully) and succeeded without error.

    Either way, though, the question of the accuracy of the original tallies, what was produced when they were all loaded, and everything happening in between, remains.

    I think the process went something like this:
    1.) People enter votes
    2.) ???
    3.) Tallies produced
    4.) Upload something to central computer
    5.) ???
    6.) Final results produced
    7.) Profit!

    I don’t know about you, but steps 2 and 5 bother me.

  7. od says:

    Why not just use pencil and paper … it seems to work well enough in every other western democracy, and has the added benefit of being almost completely verifiable.

  8. Tom Traubert says:

    The former Diebold has been making ATM machines for years. So they have always had the ability to make verifiable, accountable machines. The underlying technology isn’t all that different. The idea that they can’t produce a voting machine that prints a receipt is preposterous. To insure the secret ballot, the receipt would be placed in a secure box, and the paper results subject to random reconciliation and accounting.

    The Diebold and ESS electronic scanners suffer the same fundamental flaws as the touch-screen machines. There is no way to independently verify the counts, and no paper trail.

    Imagine a situation where you were forced to use ATMs and credit card scanners that didn’t offer you a receipt, or any method whatsoever to verify your transaction. Imagine a situation where your bank had only one way to count up the number of transactions, the Diebold way. There would be no accurate accounting possible, no way to recognize errors, no paper trail to backtrack either missing transactions. excessive transactions, or misdirected transactions. The banking industry and their customers would never put up with such a system.

    Yet that’s exactly the situation that exists with Diebold and ESS voting machines. Is that any way to run a democracy?

  9. just me says:

    Pencil and paper aren’t used, because you can’t count those votes quickly in really large jurisdictions.

    A lot of small towns though still use pencil and paper. Up until the presidential primary the town I live in has used pencil and paper.

    The town now apparently uses optical scanners now, but at least with optical scanners there is a way to check and compare votes since there is a paper ballot produced.

    I think the fallacy is that we somehow want a perfect system that has no flaws, when there are going to be flaws in any system of voting. I think it is important that when using electronic systems to vote that there be a paper record to back up the electronic one, but I don’t expect a flawless voting system. I want a fair system-but I don’t expect perfection.

  10. Bithead says:

    I don’t know about you, but steps 2 and 5 bother me.

    Well, look, let’s examine one more point, here.

    The flawed software is on both touch screen and optical scan voting machines made by Premier and the problem with vote counts is most likely to affect larger jurisdictions that feed many memory cards to a central counting database rapidly

    Who is it, that overwhelimingly runs the voting process in “larger jurisdictions”? Democrats.

    Paper and Pencil

    It was a system pretty well nigh on that that caused Florida 2000. And may I remind us all again, who it is that ran that mess? Again, Democrats.

    Gee… almost like there’s a pattern, here.

  11. Michael says:

    Who is it, that overwhelimingly runs the voting process in “larger jurisdictions”? Democrats.

    Oh right, I forgot about the magical powers of blaming Democrats.

    It was a system pretty well nigh on that that caused Florida 2000.

    Actually most of the problem was from punch card systems, not paper and pencil systems. My district used the pencil and paper optical scanners, we had no problems.

    And may I remind us all again, who it is that ran that mess? Again, Democrats.

    Oh right, sorry I forgot about the magic blame again, thanks for clearing everything up for us.

  12. Bithead says:

    Everything? Of course not.
    But should we deny the pattern developing, here, simply because Democrats are involved?

    The problems with eaither system, comes down to the people responsible for RUNNING them.
    In one case, The numbers were supposed to be cross checked. They were not. Do you blame the system or those who were supposed to be cross checking it’s output?

    Similarly, in Florida, the orignal problem when it’s boiled down, is the people responsible for running the vote, failed.

    In both cases, it was Democrats. Sorry, That’s not partisan sniping, that’s simply the facts.

    Seems to me that what the peopel sniping loudest about electronic voting systems are trying to get us to do is make an idiot proof system, while providing us with ever better idiots.

    And that’s the kinder of the readings one might give; the other, of course, involving intentional acts. which I won’t address for the moment.

  13. Alex Knapp says:

    Bithead,

    This information was allegedly discovered by Diebold in the course of a lawsuit filed by the state of Ohio. On the state level, the past few elections in Ohio have been run by Republicans.

    The problem with Diebold machines is that they do not leave a paper trail, making it impossible to verify whether votes are being tabulated correctly. My jurisdiction uses e-voting machines from a different company, which allows for a paper trail and each vote is recorded on a card that can only be used once by one voter. This is a better system. I don’t have a problem with e-voting per se. I have a problem with Diebold.

  14. od says:

    How fast do you need results? I’m not even sure its correct that its slower in larger centers – you just break the vote into more station (each one with the same number of volunteers because they service the same population).

    Does anyone know how votes are counted in large centers like London (England), Paris, Berlin, Tokyo etc? I know Canada still uses pen and paper, and some of the cities are at least reasonably large (greater Toronto is about 5 million) and they seem to get results within a day anyway.

    But I agree that if voting machines gave a paper receipt that would be good enough … it would give a paper trail.

  15. Bithead says:

    This information was allegedly discovered by Diebold in the course of a lawsuit filed by the state of Ohio. On the state level, the past few elections in Ohio have been run by Republicans.

    But not the folks actually charged with running the machines in say, Cleveland, for example. That falls to the county boards, who are invariably dominated by Democrats in the larger munis.

    The problem with Diebold machines is that they do not leave a paper trail, making it impossible to verify whether votes are being tabulated correctly

    You’re correct that they don’t have a paper trail. But since when is a paper trail needed to do a cross tab run? And clearly, from the wording of the article, there are ways to do it… ways the local BOE’s never employed. Doesn’t it give you pause to wonder why?

    And given the technology, can’t a paper trail be forged, too?

    I suppose given your level of access to this blog, you can see the IP I’m posting from. I’d just as soon this not be common knowledge for contractual reasons, but cross reference it and see where I’m coming from, when I say with no question in my mind… “yes, it can”.

  16. Arcs says:

    It was finally settled at the Supreme Court rather than at the ballot box, causing a great deal of animosity amongst the voters for the defeated party.

    Actually, no. The 2000 presidential election was settled at the ballot box. All the Supreme Court did was stop the unconstitutional recount plan of the Florida Supreme Court and allow certification of the vote.

  17. just me says:

    The problem with Diebold machines is that they do not leave a paper trail, making it impossible to verify whether votes are being tabulated correctly.

    Not so with optical scan. Or at least the ones I have used. Those you fill in the line with a special marker next to your candidate choice, and the ballot goes into the scanning box. There is both an electronic trail and a paper trail, and there weren’t apparently problems with this type of machine as well.

    Once again the problem here seems to be human error-how fast the results are fed and the will to take the time to compare-rather than with the actual computers/scanners.

    I don’t think there should be any form of vote counting that doesn’t come with a paper trail of some sort, but I am also not a believer in there being a fail proof system of counting votes.

    Even pencil and paper votes can be counted incorrectly, because you have a high probability of human error.

    It is going to happen-the goal is to reduce the error as much as possible.

  18. od says:

    Even pencil and paper votes can be counted incorrectly, because you have a high probability of human error.

    The main reason for a paper trail is that you can do a recount afterwards, especially in close elections. In the case of voting machines, if they spit out a piece of paper, the voter could check it to make sure it was what they’d picked before putting it into the ballet box, and afterwards, if necessary, people could sit down and manually count the returns.

    I don’t think this is a partisan issue, I don’t think either party is trying to “steal” elections. But anyone who works with software or hardware knows there’s almost always bugs and glitches, and it would cost almost nothing to put a paper backup in place (ie they have them on ATM’s for the same reason).

  19. Michael says:

    But should we deny the pattern developing, here, simply because Democrats are involved?

    No, we should deny the pattern because it only exists as an artifact of your insistence on blaming the Democratic party for any and all problems.

    The problems with eaither system, comes down to the people responsible for RUNNING them.
    In one case, The numbers were supposed to be cross checked. They were not. Do you blame the system or those who were supposed to be cross checking it’s output?

    Again, the report seemed to indicate that they _were_ cross checked, and that the error _were_ found by those who were supposed to be checking them. But because the errors went away after re-loading the data, it was not reported back to Diebold. At least that’s Diebold’s claim.

    Similarly, in Florida, the orignal problem when it’s boiled down, is the people responsible for running the vote, failed.

    Uh, no. The problem was that the ballots and machines were not definitive on whether and which candidate received the vote. The problem was _not_ with the people running the vote, because they are not allowed to check the accuracy of the ballot when it is cast. They could only check _after_ the ballot was cast, in which case they had to try and determine the preference of the voter, using only an non-definitive ballot for information. Again, the problem was the system, not the people.

    In both cases, it was Democrats. Sorry, That’s not partisan sniping, that’s simply the facts.

    There were representatives from both parties at both the initial voting locations, and during the recount. Oh wait, you’re using the magic blame card again, aren’t you?

    Seems to me that what the peopel sniping loudest about electronic voting systems are trying to get us to do is make an idiot proof system, while providing us with ever better idiots.

    What we want is a verifiable system. We recognize that there will be idiots casting votes, we just don’t want idiot software doing the counting, especially when we have no way to verify it’s accuracy.

    You’re correct that they don’t have a paper trail. But since when is a paper trail needed to do a cross tab run?

    How else are you going to check that what the computer says it counted is the same as what the computer _should_ have counted?

    And given the technology, can’t a paper trail be forged, too?

    It could be. The intention isn’t to prevent fraudulent votes, it’s to ensure that legitimate votes are counted.

  20. Michael says:

    Even pencil and paper votes can be counted incorrectly, because you have a high probability of human error.

    The difference is that, after the election, when somebody asks “Was the count accurate?”, you have a way of verifying that yes, it was or no, it wasn’t. Without a paper trail, the only answer to the above question is: I don’t know.

  21. Bithead says:

    No, we should deny the pattern because it only exists as an artifact of your insistence on blaming the Democratic party for any and all problems.

    Oh, please, spare me.
    Look at the DATA, Mike,a nd tell me what other conclusions there are to draw.

    Again, the report seemed to indicate that they _were_ cross checked

    Impossible, else the errors would have been located. Example; checking total votes vs total number of voter signtures. (We DO get people to sign a statement, so signitures can be compared, right? No?

    Uh, no. The problem was that the ballots and machines were not definitive on whether and which candidate received the vote.

    And who designed that system, and apporved it’s use?

    There were representatives from both parties at both the initial voting locations, and during the recount. Oh wait, you’re using the magic blame card again, aren’t you?

    If the treatment of Republican reps during the recont process is exemplary of the treatment we saw they got during the vote, I think we may be able to exonerate the Republican reps…. clearly they had no say at all.

    What we want is a verifiable system. We recognize that there will be idiots casting votes, we just don’t want idiot software doing the counting, especially when we have no way to verify it’s accuracy.

    There is no software that exists that’s any better than the people running it; in thsi case, not the voters, as you seem to imply here, but rather those responsible for it.

    And given the technology, can’t a paper trail be forged, too?

    It could be. The intention isn’t to prevent fraudulent votes, it’s to ensure that legitimate votes are counted.

    Isn’t the idea to make sure the votes counted are ACCURATE?

  22. Michael says:

    Look at the DATA, Mike,a nd tell me what other conclusions there are to draw.

    My first conclusion: Whenever you have a process that involves large numbers of people without strict organization and management, things go wrong. Strangely, my conclusions don’t require any preconceived opinions about which political party is better.

    Impossible, else the errors would have been located.

    Why would they be located, when they went away all by themselves?

    And who designed that system, and apporved it’s use?

    The county’s supervisor of elections. I have no idea who that was when the punch-card systems were first put in place. I somehow doubt that the person’s political affiliations really mattered much in that case. But hey, if you want to keep using your magical blame card, don’t let mere logic stop you.

    If the treatment of Republican reps during the recont process is exemplary of the treatment we saw they got during the vote, I think we may be able to exonerate the Republican reps…. clearly they had no say at all.

    Of course, you’re right, the Democrats are always solely at fault. Republicans never make mistakes. Why, if I only changed my voter affiliation to R, I’d be beyond blame too.

    There is no software that exists that’s any better than the people running it; in thsi case, not the voters, as you seem to imply here, but rather those responsible for it.

    The people running the elections locations did not half-punch ballots, they could not check if the voter half-punched ballots, so the people running the process did not have any control over the quality of the ballots cast. The poor quality of the ballots cast was the primary problem in the 2000 Florida recount. Only the voter and the voting machine had any influence over that.

    Isn’t the idea to make sure the votes counted are ACCURATE?

    Yes, accurate, but not necessarily legitimate. Voting machines can only make sure that a vote _is_ counted, they can’t determine if they _should_ be counted.

  23. Bithead says:

    My first conclusion: Whenever you have a process that involves large numbers of people without strict organization and management, things go wrong. Strangely, my conclusions don’t require any preconceived opinions about which political party is better.

    No, you go out of your way to avoid the obvious facts in front of you. I’m not suggesting you’re biased, here so much as going beyond neutrality, in your effort to avoid assigning blame.

    Why would they be located, when they went away all by themselves?

    And what, in the article give you THAT idea?

    The county’s supervisor of elections. I have no idea who that was when the punch-card systems were first put in place. I somehow doubt that the person’s political affiliations really mattered much in that case. But hey, if you want to keep using your magical blame card, don’t let mere logic stop you.

    When they were first put in place is irrelevant. That they were approved by the people currently running the thing over a period of decades, is another matter. And that’s who I’m blaming.

    The people running the elections locations did not half-punch ballots

    Arguably true… the voters did. (And I’ll get to that part.)

    Yet, interestingly, the systems employed failed to meet the standard being set for the electronic systems under discussion in Alex’s post… (I would argue that despite there being a paper trail, the votes were unverifiable in many instances.) …There seems a lot of blame being placed on Diebold, (et al) yet there’s no blame to be placed on the local BOE, for their failure. Why?

    Oh. They are of the group that cannot be named… err… BLAMED…

    And just a side thought as promised… and at the risk of ‘the blame card’ charge being lowered again, let me point out that in the case of Florida, the claim of the Democrats was that people who punched Buchannan, really WANTED to punch Gore. Is it reasonable to assume that most of the people who wanted to vote Gore but did not, and punched the wrong hole, were in fact Democrats?

  24. Michael says:

    I’m not suggesting you’re biased, here so much as going beyond neutrality, in your effort to avoid assigning blame.

    Oh I have no problem assigning blame, I’m just not going to categorically blame an entire voting block when their political preference has absolutely nothing to do with the fault.

    And what, in the article give you THAT idea?

    From the quoted portion of the article:

    Riggall said he was “confident” that elections officials through the years would have realized votes had been dropped when they crosschecked their tallies to certify final elections results and would have reloaded cards so as not to lose votes.

    Emphasis mine. Mr. Riggall at least believes that the problems were worked around by the elections workers you accuse of failing to spot them.

    When they were first put in place is irrelevant. That they were approved by the people currently running the thing over a period of decades, is another matter. And that’s who I’m blaming.

    If you are going to claim that the current officials made a mistake approving the machines, you have to also claim that all previous officials who approved the machines made the same mistake.

    I would argue that despite there being a paper trail, the votes were unverifiable in many instances.

    That’s pretty much the point I’ve been making, the the system was the problem, not the officials who maintained them.

    There seems a lot of blame being placed on Diebold, (et al) yet there’s no blame to be placed on the local BOE, for their failure. Why?

    Everyone I’ve seen blaming Diebold has also been blaming the election officials who continue to use them even after their faults have been revealed.

    And just a side thought as promised… and at the risk of ‘the blame card’ charge being lowered again, let me point out that in the case of Florida, the claim of the Democrats was that people who punched Buchannan, really WANTED to punch Gore. Is it reasonable to assume that most of the people who wanted to vote Gore but did not, and punched the wrong hole, were in fact Democrats?

    That was only one of several ballot issues in 2000. The “hanging-chad” debacle effected both parties nearly equally, which was of course why even after an initial hand recount, the debacle continued for months.

  25. Bithead says:

    Oh I have no problem assigning blame, I’m just not going to categorically blame an entire voting block when their political preference has absolutely nothing to do with the fault.

    OK, show me where we’ve had problems of this magnitude in a district not run by Democrats.

    Emphasis mine. Mr. Riggall at least believes that the problems were worked around by the elections workers you accuse of failing to spot them.

    Beleived. Past tense. Clearly, that wasn’t the case, the totals were NOT being checked…. else the problem would have been found.

    If you are going to claim that the current officials made a mistake approving the machines, you have to also claim that all previous officials who approved the machines made the same mistake.

    Approving the machines, there’d be no way to know of a problem. IN operation is another metter. Of course there’s been no substantial change in the makeup of the local BOE’s in question that I’m aware of, no massive turnover, that would indicate a shift of blame during the period these machiens have ben in operation, so the point seems rather moot.

    That’s pretty much the point I’ve been making, the the system was the problem, not the officials who maintained them.

    No.
    Look; if your demand is for a paper trail, so these things can be cross checked, would, by the same token, at least a electronic cross check of the data be of equal import? If the measures available to the BOE’s and their minions are not used, how is that the fault of anyone but operators of the machines? Whose failure was it that such did not get performed?

    Everyone I’ve seen blaming Diebold has also been blaming the election officials who continue to use them even after their faults have been revealed.

    Again, the point is missed. It’s not that they were used, but that they were MISused… which I would certainly consider not cross checking the totals to be. Is there another way I can get this across?

    That was only one of several ballot issues in 2000…The “hanging-chad” debacle effected both parties nearly equally

    Which of course is why the Democrats kept it up in the courts. But you know, you and I unknowingly slide past a great parallel that just now occurrs to me;

    Is it logical to assume the problems we saw in 2000 in Florida are the first time such problems popped up? The implications here, when held up against the electronic systems issues are interesting.

  26. Michael says:

    OK, show me where we’ve had problems of this magnitude in a district not run by Democrats.

    We’ve only had problems of this magnitude once.

    Beleived. Past tense. Clearly, that wasn’t the case, the totals were NOT being checked…. else the problem would have been found.

    You and I still read this differently, which explains our disagreements over most of the rest of your post. So we’ll move on.

    Is it logical to assume the problems we saw in 2000 in Florida are the first time such problems popped up? The implications here, when held up against the electronic systems issues are interesting.

    Of course that wasn’t the first time the problem with punch-card systems came up. It was, however, the first time that the number of undecisive ballots was greater than the margin of victory, and so it was the first time they mattered. In past elections, questionable ballots would be ignored because, regardless of what those voters intended, they would not have changed the results of the election. If Florida wasn’t so close in 2000, those machines would probably still be in use.

  27. Bithead says:

    OK, show me where we’ve had problems of this magnitude in a district not run by Democrats.

    We’ve only had problems of this magnitude once.

    Exactly so.
    Any other questions?

    Of course that wasn’t the first time the problem with punch-card systems came up. It was, however, the first time that the number of undecisive ballots was greater than the margin of victory, and so it was the first time they mattered.

    True. Now; was that the case in the primary where the electronic problems were found? No? seems to me the problem here was discovered under much less serious conditions. On that basis, I guess one might argue that the Diebold systems are a bit better than the old, despite the problems. (Shrug) I duuno as I would myself, but there it is, anyway.

  28. Bandit says:

    Without knowing anything about the source db and application architecture I would say that every transactional record (vote) writes to a verifiable log and that the percentage of dropped votes is probably less than .001. I don’t think we should use any technology – I think that even x’s on scraps of paper are too risky – I say use colored pebbles – they won’t burn.

  29. Michael says:

    Without knowing anything about the source db and application architecture I would say that every transactional record (vote) writes to a verifiable log and that the percentage of dropped votes is probably less than .001.

    I believe they use a combination of Windows, VB, and either MS Access or an embedded SQL server. Though the problem seems to be with the upload process, not the actual vote casting process. I’m guessing there is some kind of database merge that has race conditions if you’re uploading from more than one source at a time.