Optical Scan Voting Shown To Be Very Accurate In Minnesota Election

from the vote-here dept

We've had so many different stories about problems with e-voting machines and optical scan ballot counting machines, that it's at least worth acknowledging when those machines appear to have actually done their job reasonably well (though, not perfectly). Andrew Appel notes that the hand recount done in Minnesota for the Senatorial election there gave us a chance to look at how well some optical scan machines did, and he suggests they did extremely well, with a net accuracy at 99.99% and a gross accuracy of 99.91%. Of course... both of those numbers mean that the number of ballots incorrectly recorded could have swung the election in one way or another, given the minuscule margin between the two candidates.

Either way, the fact that the machines can be somewhat accurate is hardly up for debate. The issue is about whether or not we know they are accurate, and have mechanisms to easily go back and verify that they're accurate and secure. And, it's on those two issues that e-voting companies are way behind in fulfilling.

Reader Comments

Subscribe: RSS

View by: Time | Thread

  • identicon
    sacamano, 23 Jan 2009 @ 1:44am

    Compared to what?

    So how do we know that manual counting is flawless? I can only think of the "counting again" algorithm. Was the manual count also verified?

    Just to say, manual counting has always had possible flaws too (election rigging did not start with voting machines!), and automated counting can be allowed some leeway too.

    I would say this is close enough for most cases if they keep improving; recounts for small differences will always be needed. But a recount only has meaning if the same method is applied. A manual recount after an electronic election therefore needs to be executed twice.

    reply to this | link to this | view in chronology ]

    • identicon
      Anonymous Coward, 23 Jan 2009 @ 2:53am

      Re: Compared to what?

      i one worked for an election in Montreal.

      at the ballot, there is 1 representative of each party plus 1 independent "supervisor".

      the independent, signs every ballot (in the back) before the voters show up, that makes sure no voter cam in with extras.

      then after the voting is done the votes are counted by the representatives and the supervisor, they also examine if the signature of the supervisor is on the back of each one.

      if there is any disagreement votes are recounted or if any doubts arise about any ballot they debate on them one by one.

      of course its not perfect, but as long as the those ppl are doing there job as intended the you can be assured that the numbers tallied are correct.

      reply to this | link to this | view in chronology ]

      • identicon
        sacamano, 23 Jan 2009 @ 10:29am

        Re: Re: Compared to what?

        Thanks for this insight, that is really different from here in the Netherlands. However, this is a solution to reaching agreement on the outcome of the first counting. It does resolve differences of opinion, but it is not really certain that a second counting (by a diffferent group of officials) would produce exactly the same result.

        Which was sort of my point.. ;-)

        reply to this | link to this | view in chronology ]

  • identicon
    DS, 23 Jan 2009 @ 4:04am

    The issue became (yet again) how to count votes from people who cannot follow directions. There's two problems with this situation. The first one, is when people start trying to be mindreaders, and start judging the vote by "intent" is when you run into issues. If you can't figure out how to fill in a scantron form, your vote should not count.

    The second issue is that it's a method of voting that allows for the users to submit an invalid vote. There either needs to be a verifier machine in the same booth that the card is filled out so the voter can ensure that their vote is valid, OR, to move to a method where the only result is a valid vote.

    The sad thing, is that New York already has machines that are accurate, and (for the most part) foolproof. But, because some people could not figure out how to vote in Florida, we have to get rid of them. The old voting machines are great machines. You either vote for someone, or you don't. You cannot vote for more people that you are allowed to. Easy-Peasy.

    reply to this | link to this | view in chronology ]

  • identicon
    Sean, 23 Jan 2009 @ 6:21am

    Find someone that runs a good lottery...

    I worked for GTECH who provide the tech for the lottery in Ireland. Optical-scanned machines designed to figure out what numbers people picked by scrawling on a card, and do so accurately. Record those numbers, accurately and securely. Print a receipt (and overcome printer-related problems). Process the results. Performs lots of complicated calculations (to check for fraud, cheating, etc.) and spit out the results.
    Now if the lottery people can do that (and survive lawsuits, audits to the nth degree, and fierce competition) why is it so freaking hard for the elections people to do the same?
    Scan the card. Register the selections. Produce a paper copy. Produce a result. Get audited. Hard, but not impossible.

    reply to this | link to this | view in chronology ]

  • identicon
    anymouse, 23 Jan 2009 @ 10:04am

    Follow the $$$

    Lottery people 'can do this' because they have a vested interest in making sure the 'details' are recorded correctly, both to prevent lawsuits and accommodate auditors, and because they are being PAID do to it.

    Voting machine makers only get paid to provide the machine, they have no vested interest in ensuring the accuracy of the results (meaning they get paid when the state/county/government buys their machines, NOT when the machines provide accurate results, although some may argue that they get additional 'pay' for ensuring that certain parties can influence the results).

    Change the way the machine contractswork, make the voting machine manufacturers provide the machines for free, and pay them based on the accuracy of the results produced by said machines (with accuracy tracked based on manual recounts/audits and various other 'trails' that could be included), and things might change. If I'm going to get paid the same regardless of the results of the item I'm selling, what is the incentive to actually sell something that works? They might go somewhere else and get a product from a competitor? Yeah right, it's a government contract that was granted to my company based on my brother's son's wife's neices grandfather's position, good luck finding another 'qualified' contractor....

    reply to this | link to this | view in chronology ]

  • identicon
    ttrygve, 23 Jan 2009 @ 10:04am

    In this case I suspect those numbers mean the machines probably worked exactly as intended. That kind of margin seems more likely to be caused by the inherently and unavoidably less precise human counting. I suspect when the machines are experiencing flaws that affect their vote counts, you'd see a larger differential than that.

    reply to this | link to this | view in chronology ]

  • identicon
    Ed, 23 Jan 2009 @ 7:50pm

    As a Minnesota voter, I can say with confidence that our system is top notch. The ballots are clear, the scanners accurate, and the original ballots are retained in case of issues like we're dealing with in the US senate race.

    While there are still things to fight about, we're still ahead of the curve by relying on a paper ballot that's electronically counted (the first time).

    reply to this | link to this | view in chronology ]

Add Your Comment

Have a Techdirt Account? Sign in now. Want one? Register here
Get Techdirt’s Daily Email
Use markdown for basic formatting. HTML is no longer supported.
  Save me a cookie
Follow Techdirt
Techdirt Gear
Shop Now: Techdirt Logo Gear
Report this ad  |  Hide Techdirt ads
Essential Reading
Techdirt Deals
Report this ad  |  Hide Techdirt ads
Techdirt Insider Chat
Report this ad  |  Hide Techdirt ads
Recent Stories
Report this ad  |  Hide Techdirt ads


Email This

This feature is only available to registered users. Register or sign in to use it.