Re: [ieee] RE: Hand Recounts of votes recorded on DREs
A poorly calibrated touchscreen is not a serious problem. Unless the voter
is completely spaced out she will notice that the wrong name is lit. If the
voter is experienced with touchscreen voting they will simply touch the name
again to turn it off and vote for the correct candidate. An inexperienced
voter will call the poll worker immediately. If the calibration problem
persists, the voter will notify a poll worker. The poll worker will move
the voter to another machine and, if necessary, recalibrate the voting
station. I say 'if necessary' because in Georgia we only work on a voting
station in the precinct if it is absolutely necessary. We train our poll
workers to simply remove the machine from the line and then accumulate the
votes from it after the polls close.
Brit
----- Original Message -----
From: "Arthur Keller" <ark@SOE.UCSC.EDU>
To: "Andrew Berg" <andrewb@votehere.com>
Cc: <stds-1583-stg3@IEEE.ORG>; <stds-1583-disc@IEEE.ORG>;
<stds-1583-tg1@IEEE.ORG>; <stds-1583-tg2@IEEE.ORG>
Sent: Friday, December 10, 2004 12:49 PM
Subject: Re: [ieee] RE: Hand Recounts of votes recorded on DREs
> Paper trails do partially address touchscreen calibration issues.
> Having an independent voter-verified copy of the vote would help
> voters detect when the wrong candidate was selected by touchscreen
> miscalibration. Of course, the review screen also helps with this
> problem.
>
> Miscalibrations of touchscreens are very serious, since they can mean
> voting for the wrong candidate, or even being unable to vote for some
> at all.
>
> Best regards,
> Arthur
>
> At 7:30 AM -0800 12/10/04, Andrew Berg wrote:
>>Just so that I'm clear on what you're trying to say:
>>
>>Are you claiming that machines tested to a 15000 hour MTBF would
>>_not_ need a paper trail?
>>
>>Also, are you intentionally conflating touchscreen calibration
>>issues (which a paper trail does not adequately address) with data
>>retention reliability issues (which a paper trail might be of some
>>benefit)? I'm no expert on reliability, but my years in system
>>design have taught me to keep seperate issues seperate.
>>
>>-andrew
>>
>>On 09 Dec 2004 19:05:15 -0500, Stanley A. Klein <sklein@cpcug.org> wrote:
>>
>>>Requiring recalibration is a machine failure. It is a loss of function
>>>lasting more than 10 seconds.
>>>
>>>The 163 hours allows as low as a 90.7% reliability during election day
>>>and a 72% reliability during combined election day and pre-election
>>>setup.
>>>
>>>If the systems were 99.9% reliable (i.e., a 15000 hour MTBF instead of
>>>the ludicrously inadequate 163 hours) then you would have a failure
>>>budget of .001 within which you would allocate all failures including
>>>requirements for recalibration during election day.
>>>
>>>Touch screens should be designed that can operate with much lower than
>>>.001 probability of needing recalibration during election day and less
>>>than .002 probability of needing recalibration during system setup. You
>>>need to be much lower because there are other failure modes besides
>>>requiring recalibration. If you can't do that, then touch screen
>>>technology should not be permitted in voting systems.
>>>
>>>The 163 hour MTBF is so inadequate that between 50% and 80% of precincts
>>>will experience some kind of machine failure, visible or invisible,
>>>during election day. This is borne out by empirical observation during
>>>the recent election. Equipment built to that shoddy a standard needs
>>>the backup of a paper trail.
>>>
>>>
>>>Stan Klein
>>>
>>
>>-andrew
>>
>>--
>>andrewb@votehere.com
>
>
> --
> -------------------------------------------------------------------------------
> Arthur M. Keller, Visiting Associate Professor, Computer Science Dept.
> University of California, Baskin School of Engineering, Room 153B,
> 1156 High Street, Santa Cruz, CA 95066-1077
> tel +1(831)459-1485, fax +1(831)459-4829, www.soe.ucsc.edu/~ark
>