The Future of Voting Systems in Los Angeles County

ivotedbigApart from my other jobs, I moonlight as a wonk.  In particular, for the last 6 years or so, I’ve been involved with a group called the Open Voting Consortium, much of that on its board and as its CTO. With that hat on, I am enormously excited that Los Angeles County is likely to get much better voting systems in the relatively near future.

Let me give the brief plug: we want to make sure that no one has to vote on proprietary DRE voting machines (or ever does voluntarily, for that matter).  There are two glaring flaws in these systems: the source code is secret (so-called trade secrets), and both accidental flaws and deliberate vote tampering is both possible and has likely happened; a voter has no means to inspect the recorded vote before casting it (other than a machine telling them, “trust us, we’ll put the right electrons somewhere”). The right system is an Electronic Ballot Printer, which is basically to say just a computer-assistive device to help mark a ballot that a voter can inspect physically before casting.  The paper is crucial because voters and poll workers can easily and reliably understand both that and why they are secure and accurate.  Using computers is also important though, because it enables independent and anonymous voting by persons with disabilities (especially, but not only, blind and visually impaired voters), enables multi-lingual ballot presentation, and reduces overvoting, undervoting, and other errors in capturing voter intent.

The Los Angeles County Registrar-Recorder/County Clerk, Dean Logan, held an all-day symposium yesterday, entitled Technology, Diversity, Democracy: The Future of Voting Systems in Los Angeles County.  This was a really wonderful effort that shows the best of our government officials.  Registrar Logan has a commitment to getting input from the range of stakeholders in this process, while simultaneously understanding well the technical and political issues involved.  The meeting was composed of… well, lots of wonks like me, but ones from the right range of walks of life.  The disability rights community was well represented; as were LA-based voter groups (such as advocates for diverse ethnic and linguistic groups that need ballot access); and a good number of the nation’s top cryptography and political science thinkers about voting were in the mix for good measure.

The Registrar-Recorder staff, many of whom I had the chance to speak with, were well prepared and well-informed in their role of facilitating the symposium.  Unfortunately, California Secretary of State Debra Bowen, although scheduled to participate, was unable to attend–it’s too bad because she is really one of the good guys in relation to openness and transparency in government.  It was also nice to hear Election Assistance Commissioner Donetta Davidson speak though, and I was delighted that I happened to have the chance to talk with her at breakfast, before the formal sessions.

This is all a bit technical, as good news goes.  And nothing is announced (or developed) yet, in any case.  But I encourage readers to become informed on this, and bring with the process a dose of optimism that hasn’t been possible for a few years.  Read OVC’s site for background information, and also take a look at the Registrar-Recorder’s website.  Provide feedback to the Registrar-Recorder as this process unfolds (information will be posted over time, and voter feedback is essential to our future democracy).

Addendum: In response to a reader comment below, I think it is worth exhibiting a sample ballot produced for a demonstration election using the OVC Ballot Printer Architecture design.  Something like this would serve as the official ballot that is inspected by a voter, stored for recounts and audits, and so on.

An OVC sample ballot
An OVC sample ballot

16 Replies to “The Future of Voting Systems in Los Angeles County”

  1. Is a difficult thing to comprehend, the problem of source code transparency and the need to keep trade secrets. We need watchdogs but if there is no transparency how could a watchdog protect the system.

    Basically at least the old system of ballots had a “backup” where the ballots could be recounted. Unfortunately the old system is cumbersome. Good luck in explaining how a new system can be foolproof, tamperproof and reliable.

  2. Unfortunately, the OVC system is not much better than current touch-screen systems. It’s also a touch-screen system, which means when the power goes down, people cannot vote. But more troubling is that, as with DRE machines, it’s impossible to verify after an election, that the ballot printed out by it is marked the way voters had intended it to be.

    Open source is fine. Open, verifiable, fully transparent voting is much better. The OVC system fails to meet that bar.

  3. TarzanaDem has missed many of the details of the OVC design. Which is understandable enough, it takes a while with immersion in all the jargon and design of voting systems to understand the subtleties.

    The point of a “Ballot Printer Architecture” is precisely that it is directly voter-verifiable. A machine is used solely for producing a piece of paper (the ballot). That paper is physically and directly inspected by a voter before the voter casts it into the ballot box. No record is stored on the machine itself, only on the paper. Inspection is just a matter of looking for sighted voters, but a separate verification machine is available for vision impaired ones. If a printed ballot, upon inspection, does not reflect voter intent, the voter can and should mark that ballot invalid and produce a replacement.

    The issue of power outage is important, and unfortunately a difficult one. Pre-printed hand-markable ballots can be provided to cover gaps in power availability. And batteries can operate the machines for a certain duration during an outage. But absent electricity, we lose the accessibility advantages during an outage. Most voters can continue to vote without using the Ballot Printer, but a subset risk being disenfranchised during these intervals.

  4. “voter-verifiable” does not equal “voter-verified”.

    Studies have shown that voters do not check their computer vote print out, and other studies have shown that even when voters do check it, they don’t notice that their vote has been printed wrong by the computer more often than not.

    If we cannot know if a vote was cast as intended, nothing has been solved. OVC’s electronic touchscreen scheme failes in this regard. Good intentions, perhaps, but failed execution.

    By the way, they might want to try a test with 10 or 15 YES/NO ballot measures and see how many voters notice when the touchscreen ballot printer has misprinted choices. They would learn what those who study this sort of thing already have.

  5. TarzanaDem seems to have thought about voting technologies, and I encourage him or her to join the OVC mailing list, where all of these issues have been discussed in great depth for years. It sounds like s/he know something in these areas, but I worry a little that s/he comes with a chip on his/her shoulder too, from the tone of these comments. There actually are people who study these things for a living and by doctoral training, and assuming one automatically knows better than the experts isn’t always the right conclusion. OVC is largely composed of “those who study this sort of thing” (I’ve published academically in these areas myself, fairly broadly, but not as much as have some members of our group).

    There is indeed a gap between “voter-verifiable” and “voter-verified”. One of the point I raised at the Registrar’s symposium was exactly that; specifically the need to make ballots as transparently and accurately verified as possible. The OVC’s Ballot Printer is not perfect this way… however, it is better than any existing system, or than any other proposed system. Accurately and actually verifying LA’s current InkaVote+ system is far less likely to happen in practice, though it’s really not a bad system currently used. Other physical ballot marking systems suffer the same flaw (including simple pen-marked ‘X’ in a box), though full-face ballots are more easily verifiable than are compact (punchcard-sized) versions.

    I am confident that the “reduced format ballot” shown in the article above will, in fact, be more accurately verified in practice than are any other ballot systems. Positively demonstrating that is part of what needs to be studied in developing LAs voting system (there are sociologists and cognitive psychologists on board to do this work). The general idea of the “reduced format” is that it is possible to print only the options affirmatively chosen by a voter, and omitting the non-chosen candidates or initiative choices does not reduce the information on a ballot (it just saves space and eases verification).

  6. “TarzanaDem seems to have thought about voting technologies, and I encourage him or her to join the OVC mailing list, where all of these issues have been discussed in great depth for years.” — I am already on their list and well familiar with their proposed touchscreen voting machines.

    “It sounds like s/he know something in these areas, but I worry a little that s/he comes with a chip on his/her shoulder too, from the tone of these comments. There actually are people who study these things for a living and by doctoral training, and assuming one automatically knows better than the experts isn’t always the right conclusion.” — A chip? What makes you assume I’m not such an expert? Or even that it takes such an “expert” to know what can and can’t be verified after an election as being voter-verified?

    Suggesting that an “expert” is needed to determine that, whether I’m an expert or not, sounds like you may have a bit of a chip on your shoulder, from the tone of your comment. Voting systems are for general citizens who they serve. Not for experts. The “experts” are one of the reason we’re in the state we now are.

    Lauding any voting system by saying “well it’s better than inkavote”, is not very high praise. Not to mention that at least with inkavote, there is a way to assure with a very high degree of likelihood, that the votes on the card are the ones the voter actually intended to cast. the OVC computer printed ballots offers no such possibility.

    Among other studies out there, I might suggest you read the ones from both Rice University and from Caltech/MIT on these matters.

    Comparing ANYTHING to L.A.’s inkavote system,

  7. (That last sentence fragment was an accident. Please ignore it. Although, in truth, there’s no way to know whether I intended it to be there or not, is there? :)

  8. Question – what is it that the machine counts, is it that bar code on the side of the ballot?

    If the machine counts the barcode, how does the voter verify that the bar-code on the ballot is correct?

    Is there a reason OVC couldn’t just create an open source optical scan system that lets voters mark the ballots?

  9. Good question, Laura Roslin. Indeed what the OVC demo/prototype code does is use the bar code for scanning. You can follow much more of this on the OVC mailing list (which are all archived as well). Actually, more recent ballot samples use a 2D bar code rather than a 1D, but the idea is the same. We have also done versions of code with OCR instead.

    A number of developers/users (myself included) have expressed concern that the bar code is not directly verifiable. What it encodes is exactly the same information that is verifiable, but knowing that has a worrying “trust us” element to it for average voters. That said, the idea would also be to allow independent companies/voter groups to independently verify the accuracy of the bar code in matching the printed information; and ideally, all elections would have a proof of correlation in a random sample of ballots (if you know statistics, it’s not hard to construct a pretty small sample that makes the odds of tampering negligible… probably less than 1% of ballots needed for this).

    That said, there are a number of striking benefits to bar codes over plain OCR of human-readable print. OCR is much faster to scan, the equipment needed for scanning is much cheaper, and 2D bar codes can contain considerable error-correction within them (allowing scans even if part of it is smudged/torn/etc, a feature likely not present in OCR). From an esoteric point-of-view, we can also include cryptographic anti-counterfeiting information in there. I wouldn’t take the bar code as a deep part of the Electronic Ballot Architecture. However, I suspect that that feature will be included in eventual LA voting systems because both cost and ability to scan rapidly are both high on the design wish list.

  10. Note: Two posts from this commenter with a belligerent tone were indeed removed. The below comment seems unnecessarily angry as well, but in any case, this is the wrong forum for an open-ended discussion of voting systems –Lulu

    Um, Lulu, why did you remove my post? It wasn’t rude, it wasn’t obsene, it was in response to your note and the discussion concerning the absolutely insane idea of using non-verifiable bar codes to count ballots electronically, while ignoring the verifiable (though not necessarily verified) text.

    Is your support for OVC’s system unable to stand up to scrutiny? And what does that say about OVC’s system which, as I described it in the note you removed without explanation, seems to be EPIC FAIL.

    What the heck is going on here? (And doesn’t it show how easy it is to remove something without a trace on a computer?? Will this be removed too???)

  11. As you know, but your readers may not, because you deleted them without notice, until I called it out above, there was nothing belligerent about the posts you deleted, Lulu.

    Concerning your strange suggestion that “this is the wrong forum for an open-ended discussion of voting systems”. Huh? The title of this post, is “The Future of Voting Systems in Los Angeles County”, where you kicked it off with a discussion of your preferred voting system.

    When problems were pointed out with it, as enlightening discussion and debate ensued, you began deleting notes and then claimed this is not the place for a discussion of voting systems.

    If that’s not what this thread is about and why you posted it in the first place then why DID you post it and where IS the forum for such a discussion?

  12. TarzanaDem raises a point that comes up frequently in voting technology discussions. Folks like Bev Harris and her BlackBox Voting group are distrustful of all computer assistive devices, and want voting to be strictly pen-and-paper. I voted happily for a number of years, myself, in Massachusetts, where my county had a strictly pen-and-paper system. However, there are several factors which make this approach a non-starter for Los Angeles, and indeed for any future US voting systems:

    (1) All-paper disenfranchises blind voters and mobility-limited voters who cannot operate pen-and-paper. Computer-assistive devices can enable these voters to vote independently and anonymously. Many of the all-paper advocates feel that it is a reasonable compromise to either let disabled voters vote on “separate but equal” voting systems, or to require them to rely on trusted third parties in casting a ballot. Very few, if any, disability-rights folks find this acceptable; moreover, it is unlikely to be legal under HAVA and the ADA.

    (2) There is no actual evidence (nor probability) that all-paper systems are in fact more accurately verified that are ballots printed on Electronic Ballot printers. It is frequent to mark the wrong bubbles/boxes; to mark outside borders; to ambiguously cross-out an incorrect mark; etc. Computer assistive devices (such as touch screens) can watch out for such failures in expression of voter intent better than can pen-and-paper, and prevent overvotes and accidental undervotes. It is quite true that you cannot force voters to accurately check the ballot produced by an EBP, but the overall rate of actual verification is still likely to be higher than with pen-and-paper.

  13. 1) This is an olde thyme red-herring from the e-voting industry. Setting aside that there are assistive devices for the disabled that don’t require a computer, this is not a discusion about disabled voting. That was something Diebold and friends in Congress and the e-vote lobby used to use to push for requiring all voters to use DREs (touchscreens), and even touchscreens without paper trails, for that matter. The “separate but equal” argument is bogus on its face. If it weren’t, we’d have to outlaw mail-in voting if computer printouts are used in the polling place. Are you suggesting that we do? Or are you willing to all “separate but equal” for some, but not others?

    I will also note, as others have pointed out before me, that we do not require City Hall to destroy the stairs for some to use optionally, just because they’ve installed a “separate but equal” wheelchair ramp for similarly optional use.

    There is nothing in HAVA or the ADA that requires everyone use voting systems that are specifically made for the disabled. It is, as I said, a very red-herring.

    2) Hand marked paper ballots systems are, by their very definition, are “more accurately verified that are ballots printed on Electronic Ballot printers” in that we can know the voter him-herself actually chose the the candidates-initiatives selected on them (presuming proper chain of custody, etc.).

    Whether they marked them “accurately” or not, meaning the way they meant to mark them, is their responsibility. We can’t force someone to do anything. But we know they are RESPONSIBLE for the selection marked on the ballot.

    There can be no way to know that even a single vote printed out by a computer is actually a vote that was intended by the voter.

    Why you continue to push for such a system is beyond me. Do you work with OVC or some other company/group that might benefit from moving to such an unverifiable type of voting system? Not making any allegations here but trying to understand why you are pushing voting systems, with huge long-known failure/flaws in them, on readers here.

    The issues we’re discussing have been long understood by evoting experts and computer scientists and computer security experts, and yet you still seem to be pushing for an outmoded and some might say discredited type of technology that most of America has already begun to move away from.

  14. I’d love to hear about these “assistive devices without computers”; in the last 6 years of reading the literature on voting technologies, publishing in the area, and attending conferences on the subject, I have never encountered them before. Maybe TarzanaDem knows something no one else does though. FWIW, as stated in my original post, I have volunteered for OVC for a number of years (which is like work, I guess, minus the pay).

    I confess I am annoyed by the vacuous totemism of most of the all-paper folks. For them, as in the above comment, representing voter intention doesn’t actually matter, it’s only the ritualistic act of making a physical mark. It starts to resemble the ritual meaning of an “X” made on a contract by a non-literate person (i.e. who cannot read the document). In real life, voter errors using pen-and-paper systems are quite common, and that is a huge failing with those systems; though fortunately it’s one that will ultimately be solved by better systems. Obviously, no voting technology is going to make all voter errors impossible; it’s a question of reducing rates of error as low as possible.

    The famous “hanging chads” from Florida of 2000 are a particularly well known example of a jurisdiction where voters clearly failed to express intention (in statistically important numbers) via a physical modification of paper. It is true that was the (totemic) physical act of punching a hole rather than of making a mark with a pen, but generally the same failures are dangers in any all-paper system. An EBP (properly built) is simply and greatly less prone to these errors.

    It is worth noting as well, that error in tallies come in two parts: (1) errors in representing voter intent; and (2) errors in tabulation. Other than DREs, which potentially have tabulations errors of 100% (either because of tampering or just sloppy programming), hand-marked long-form ballots are probably the most prone to tabulation errors (probably worse if they are hand counted rather than machine scanned). The use of Optiscan cards in many jurisdictions, and the use of bar codes on the OVC sample designs, reduces tabulation errors greatly (with the OVC bar codes being far more reliable than the bubbles of InkaVote+ and other Optiscan-style systems). The other part of the fetish of hand-marked/hand-counted paper advocates is the magical belief that humans cannot make errors during tabulation.

  15. “I’d love to hear about these “assistive devices without computers”; in the last 6 years of reading the literature on voting technologies, publishing in the area, and attending conferences on the subject, I have never encountered them before. Maybe TarzanaDem knows something no one else does though.” — They are referred to as “tactile devices”. There are several. Here’s one that’s being used in Wisconsin and elsehwere currently: Vote-pad.us

    “I confess I am annoyed by the vacuous totemism of most of the all-paper folks. For them, as in the above comment, representing voter intention doesn’t actually matter, it’s only the ritualistic act of making a physical mark.” — I think that’s a very unfair and inaccurate characterization of those who support transparent voting systems. I’ve not heard any of them suggest it’s about the “ritualist act of making a phyiscal mark”. But rather, about the transparency of being able to know after an election that you are actually counting the intention of the voter. Also, they correctly argue, in my opinion, that private corporations have no busines counting the publics vote with secret software. While open source seems, on the surface, to solve THAT problem (it doesn’t, but I won’t get into the open source code myth in voting systems for the moment), it doesn’t defeat the transparency/verifiability issue.

    “voter errors using pen-and-paper systems are quite common” — the good news is that such errors can be discerned by any citizen after any election. There is no way to discern ANY errors on touchscreen/computer-printed systems like OVCs. Every single vote could be “in error”, but there would be no way for anybody to know. Impossible. The “human error rate” business is also a red-herring in my opinion. If there’s a concern, improve ballot design, instruction, etc. Don’t remove the ability to verify ALL votes in the process. It’s an old e-vote industry red-herring like the one you mentioned in your previous note.

    Nobody legitimate is arguing for Florida/2000-style punch-cards, so that is yet another red-herring. With my apologies to you.

    “hand-marked long-form ballots are probably the most prone to tabulation errors (probably worse if they are hand counted rather than machine scanned).” — I am unaware of any information to back up your parenthetical claim there. Please share it. The information I am familiar with shows just the opposite, when hand-counting is done at the precinct on the night of the election. (precinct-based versus centralized counting).

    “the use of bar codes on the OVC sample designs, reduces tabulation errors greatly ” — That’s a presumption, since there is no way to know if a vote cast during an election is the same as the one counted after an election. The bar code scheme is absolutely insane. Again, pardon me, but absolutely crazy and almost as UN-transparent and verifiable as you can possibly be short of DRE voting machines.

    “The other part of the fetish of hand-marked/hand-counted paper advocates is the magical belief that humans cannot make errors during tabulation.” — I am offended by your “fetish” slur, sir. Wher you have delighted comments of mine that you did not care for, obviously I do not have that same ability. But it’s an inappropriate slur, as far as I’m concerned.

    With that in mind, I have heard no “magical belief that humans cannot make errors during tabulation” from such hand-marked/hand-count/transparency proponents. But yes, it is difficult to make many substantive “errors” when there are at least four people checking every vote for accuracy, in addition to an entire community watching and or video taping.

    Errors could happen, I suppose, but it seems they’d be very easily and quickly caught, and very difficult to actually affect an entire election.

    It doesn’t sound like your very familiar with how proper hand-counting actually works. You may wish to educate yourself on that front. I’d suggest looking at New Hampshire’s process where it’s still done for appx. 40% of the votes cast. There are very good booklets available explaining the entire fully-transparent, fully-verifiable (by all!) procedure.

    Here’s a helpful link:
    “Hands-on Elections: An Informational Handbook for Running Real Elections, Using Real Paper Ballots, Counted by Real People,” – http://electiondefensealliance.org/HCPB_election_admin_handbook

Comments are closed.