Skip to main content

View Diary: BREAKING: Verified Voting Senate hearing, unpublicized, untelevised, convenes (138 comments)

Comment Preferences

  •  the "software total," Rob M? (none)
    Do you refer to the total given at the machine or the total received at the tabulator?

    What if I wrote the machine code and also know the phone number of the tabulator?

    Be all that you can be: Work for peace - - Jesus (Mt.5:9)

    by Upstream Review on Tue Jun 21, 2005 at 10:44:02 AM PDT

    [ Parent ]

    •  Doesn't Matter (none)
      The total given by any bit of software should match the total given by hand-counting the paper receipts that the voter has deposited in a ballot box.

      If someone manages to hack the tabulator and originally wrote the machine code, it'll have no effect on the election because the paper receipts will give an accurate count.

      •  these ballot boxes (none)
        .. are they only in the 1-2% of precincts to be checked, or are they universal.

        If only in 1-2% of precincts, the logistics would require so much advance notice that the machines in those places could be scrubbed by remote before the vote.  - leaving 98% of precincts open to fraud.

        And if universal, why do we need the machines to do what we have been doing for centuries - marking ballots for ballot boxes?

        Be all that you can be: Work for peace - - Jesus (Mt.5:9)

        by Upstream Review on Tue Jun 21, 2005 at 11:29:58 AM PDT

        [ Parent ]

        •  Raise the %. (none)
          to 7-8%.

          Banks do this all the time.  This kind of reconciliation is easy.

          •  Raise it to 100% (none)
            then you don't need the machines.

            Duh.

            •  Well... (none)
              statistically, you don't need more than this.

              However, with this number, randomness would have to be insured, i.e. any advance notice to the precincts to be audited would have to be elimintated.

              If this could not be done, then I agree, paper ballots are the way to go.  Why not HIRE PEOPLE (1 Dem, 1 Repub) in each precinct to help with the disabled community, rather than rely on technology which can break and then need to be fixed (and thus, our county funds are directed to the touchscreen companies and not to our people).

              •  But you don't want election by survey (none)
                you want an election.  One person one vote.  Statistics shouldn't have anything to do with it. You want a population of votes to count, not a sample.

                It is so absurd that in the US you end up "estimating" who won.  I couldn't believe that candidates conceded on the basis of an TV estimate.

                And the money saved on machines could certainly go on people.

                Disabled people here (UK) either use postal votes, or get a proxy (an named individual) to cast their vote.

                The reason for the machines is supposed to be the complicated ballots you have.  But in the UK where we have more than one race, you have more than one ballot paper and the ballots are different colours. Easy to sort and we still have the results next day.

                •  Err...not what I said. (none)
                  You are confused.

                  The "audit" is not the actual vote.  The audit is a way of checking that the results of the actual vote are accurate.

                  The audit check does not have to be a check of each and every vote to be reliable.  It does, however, need to be free of bias.  Hence, the need for blinded, random audits of a statistically significant percent of actual votes cast.

                  •  Not confused, sarcastic ... (none)
                    and not aimed at you!

                    I just think that to have to have sampled quality control to audit your elections (which I agree is probably necessary, given the US's commitment to machine voting) is itself an indictment of the machines.

                    We don't have quality control sampling in the UK because people assume the hand-count is honest and accurate.  If a result is close there is a full hand-recount if candidates request it.

                    I suppose I am still reeling from DU where people still assume that a small survey sample is a better estimate of the vote than the count.  And they may be right, although I no longer hold that view.  But the fact that a substantial number of Americans can think that a survey of 1% of voters could be a better indicator of who won the election than a count of everyone's votes says something pretty serious about how far American democracy has travelled towards losing its legitimacy!

                    But sorry if I caused offence- not intended! I agree that what you suggest is what is required.  it was certainly not what happened in Ohio,

                •  If anyone is going to be in charge... (none)
                  ...of doing the math to "estimate who won," and doing it the right way, I think it should be Febble...

                  Wonderful to see you on my thread, Elizabeth.  I have an enormous amount of respect for you and for your modeling skills.

                  Piss off Frank Luntz: don't use Republican issue frames like "Social Security crisis."

                  by DC Pol Sci on Tue Jun 21, 2005 at 03:59:53 PM PDT

                  [ Parent ]

              •  "statistically," you do need more . . (none)
                   . .  than this because you want to be able to count more than a sample, and do more than an authorized audit.

                You see, an audit (such as you envision) would catch only the errors that turn up randomly across precincts in universally applied code and procedures.

                But deviations or hacks if they occurred would not be random; rather, they would be inserted in hand-picked precinct.

                So a 3% sample is not enough, unless you're lucky to pick they right place.

                Febble is right. You want 100%. A count, not a hit-or-miss audit.

                We need a VVPB law, not a vvpat. A voter-verified paper ballot, not a paper "audit trail."

                A paper "ballot" has full legal standing to be counted, as much standing as the 'recorded' ballot translated by bit-counts. An audit trail is available theoretically, but rarely used, and only used in a sample (unless problems turn up in that sample).

                •  As someone who's been... (none)
                  involved in some challenges to elections, you are absolutely right that we need a VVPB.  It is only a "ballot" that has legal significance.

                  However, it seems you are making the same category mistake as Febble.

                  An election is a vote of the population.  Leaving aside the "audit" terminology, which I agree is a poor word now that Diebold, Dodd, Dickson et al. have so altered its meaning, there has to be a way of determining whether the ballots (whether electronic display images or paper), reflect the intent of the voters.  This follows the election and might be called a check on the accuracy of tabulation.

                  I take both you and Febble to be proposing that every election be followed by a full recounting of all the paper ballots.  This would be a complete, entirely EMPIRICAL check on the accuracy of tabulation.

                  Which is fine and dandy, but it ain't gonna happen.  Election officials would go bananas.  And, as I understand it, it just ain't necessary because every statistics person I have talked to says we CAN sample all the paper ballots in a statistically valid and reliable manner to determine whether the ballots were tabulated properly.  That being said, I further agree with you that 3% is too little.  I have been told that 7-8% will raise the margin of error into the millionths.

                  Of course, if the difference were in this margin of error, then that is a different story.  Presumably, a larger sample would then be dictated.  And, that would be in keeping with the procedures and statutes currently in force.

                  All this is a long way of saying that I agree with you on the need, within limits, for more EMPIRICAL counting in order to meet a high order of statistical certainty, but I don't think you have to go overboard and essentially have a full recount after every election.  

                  •  3%, 7% . . . (none)
                    That's no good, because as I said, certain problems would occur because they were inserted into hand-picked precincts and locations.

                    A random selection will likely not pick up something that is deliberate, or pick up something that is not a uniform error (across jurisdictions) of the reporting or tabulating software. It will more likely pick up uniform, accidental program inaccuracies.

Subscribe or Donate to support Daily Kos.

Click here for the mobile view of the site