M
Michael Griffin
At 11:33 06/12/00 -0500, Mark Hill wrote:
<clip>
>If a simplified "card reader" were provided at each polling station each
>user could slide his card into the reader and verify his vote within
>seconds.
>This could be a simple device that mimics the current Votomatic, but with a
>light source below the card. A light would shine through each punched hole,
>lighting up a small indicator next to a candidates name. The voter could
>turn each page as he does in the Votomatic and quickly verify his
>selections.
<clip>
Since we are looking at this problem from an industrial automation perspective, I will answer the above with "Repeatability and
Reproducability". I'm sure that almost everyone on this list has had to deal with this issue at work. If with a typical machine the chances of achieving perfect R&R are virtually nill, what is the probability of getting two different machines (of different designs even) to correlate perfectly? You haven't introduced a solution, you've just introduced another problem. "You
can't inspect quality into a product" applies in this case as well.
Walt Boyes wrote:
<clip>
>The biggest undiscussed problem with this election isn't the technical end.
>The real issue is the identification of unauthorized voters. There are some
>simple technical solutions to this issue, as well, but for one major
>problem. The databases that must be used to verify who is and who is not a
>legal voter are not being permitted to be used.
<clip>
>But there is no "controlling legal authority"
>requiring that the databases for felons, for illegal aliens, for death
>certificates, for birth certificates be used to determine eligibility of
>voters.
That is kind of interesting, because the biggest story along this line in our election was that an election official in the province of
British Columbia didn't let a girl vote just because she wasn't on the voter's list and didn't have ID that indicated she was a resident of that
riding as she had only recently moved back there. However, under our election laws she should have been allowed to vote because an acquaintance (her father) was there to vouch for her being a resident.
What would your system have done in this case? It would simply have repeated the same error that the election official made. If you go too far overboard in trying to exclude invalid voters, the error in your system will also exclude too many valid voters.
All these databases you mentioned have errors of their own. If you combine them together, all I think you will turn up is just how big of a mess they are in. If we are going to compare election systems to industrial processes, then you need to imagine how you are going to build flexibility into it and to allow for the unforseen.
Ralph Mackiewicz wrote:
<clip>
>The basic problem with punch cards seems to be that it is
>too easy to use improperly leaving the results ambiguous
<clip>
>Regardless of the specific system, is it really possible to come up with a
>system that can count votes to an accuracy of 0.000001% ? This is the
>accuracy required for a national popular election (1 in 100,000,000) in
>the US.
<clip>
If the errors are random, there is no significant effect on the outcome. The problem arises when the errors are systematic, i.e. a number of defective machines or ballots in one or more sensitive areas. Does anyone ever apply any of the standard SPC techniques to verify the functioning of the voting equipment? If I can't haul a machine out of a warehouse and put it into production of auto parts without doing an R&R study, why wouldn't this apply to voting machines? It may not be just the hardware that is
antiquated.
>The only reason that there is such a fuss is that the vote was a statistical
>dead heat and the resolution of a dead heat in the US constitution doesn't
>involve a direct public vote. Some people seem to have a problem with that.
To continue the industrial analogy, the process is not very robust. The system tends to magnify errors or perturbations rather than supress them. Perhaps what you really need is a constitutional monarchy with a prime minister chosen from the party with the most seats in parliament.... (ha-ha!).
But seriously, no one has really explained just why these machines are necessary in the first place. Plenty of countries get along quite nicely without them. Other people have mentioned that these machines are so old because no one feels its worth spending money on equipment that gets used so seldom. So why not just toss them out and count the ballots by hand? Where is the return on investment here?
**********************
Michael Griffin
London, Ont. Canada
[email protected]
**********************
<clip>
>If a simplified "card reader" were provided at each polling station each
>user could slide his card into the reader and verify his vote within
>seconds.
>This could be a simple device that mimics the current Votomatic, but with a
>light source below the card. A light would shine through each punched hole,
>lighting up a small indicator next to a candidates name. The voter could
>turn each page as he does in the Votomatic and quickly verify his
>selections.
<clip>
Since we are looking at this problem from an industrial automation perspective, I will answer the above with "Repeatability and
Reproducability". I'm sure that almost everyone on this list has had to deal with this issue at work. If with a typical machine the chances of achieving perfect R&R are virtually nill, what is the probability of getting two different machines (of different designs even) to correlate perfectly? You haven't introduced a solution, you've just introduced another problem. "You
can't inspect quality into a product" applies in this case as well.
Walt Boyes wrote:
<clip>
>The biggest undiscussed problem with this election isn't the technical end.
>The real issue is the identification of unauthorized voters. There are some
>simple technical solutions to this issue, as well, but for one major
>problem. The databases that must be used to verify who is and who is not a
>legal voter are not being permitted to be used.
<clip>
>But there is no "controlling legal authority"
>requiring that the databases for felons, for illegal aliens, for death
>certificates, for birth certificates be used to determine eligibility of
>voters.
That is kind of interesting, because the biggest story along this line in our election was that an election official in the province of
British Columbia didn't let a girl vote just because she wasn't on the voter's list and didn't have ID that indicated she was a resident of that
riding as she had only recently moved back there. However, under our election laws she should have been allowed to vote because an acquaintance (her father) was there to vouch for her being a resident.
What would your system have done in this case? It would simply have repeated the same error that the election official made. If you go too far overboard in trying to exclude invalid voters, the error in your system will also exclude too many valid voters.
All these databases you mentioned have errors of their own. If you combine them together, all I think you will turn up is just how big of a mess they are in. If we are going to compare election systems to industrial processes, then you need to imagine how you are going to build flexibility into it and to allow for the unforseen.
Ralph Mackiewicz wrote:
<clip>
>The basic problem with punch cards seems to be that it is
>too easy to use improperly leaving the results ambiguous
<clip>
>Regardless of the specific system, is it really possible to come up with a
>system that can count votes to an accuracy of 0.000001% ? This is the
>accuracy required for a national popular election (1 in 100,000,000) in
>the US.
<clip>
If the errors are random, there is no significant effect on the outcome. The problem arises when the errors are systematic, i.e. a number of defective machines or ballots in one or more sensitive areas. Does anyone ever apply any of the standard SPC techniques to verify the functioning of the voting equipment? If I can't haul a machine out of a warehouse and put it into production of auto parts without doing an R&R study, why wouldn't this apply to voting machines? It may not be just the hardware that is
antiquated.
>The only reason that there is such a fuss is that the vote was a statistical
>dead heat and the resolution of a dead heat in the US constitution doesn't
>involve a direct public vote. Some people seem to have a problem with that.
To continue the industrial analogy, the process is not very robust. The system tends to magnify errors or perturbations rather than supress them. Perhaps what you really need is a constitutional monarchy with a prime minister chosen from the party with the most seats in parliament.... (ha-ha!).
But seriously, no one has really explained just why these machines are necessary in the first place. Plenty of countries get along quite nicely without them. Other people have mentioned that these machines are so old because no one feels its worth spending money on equipment that gets used so seldom. So why not just toss them out and count the ballots by hand? Where is the return on investment here?
**********************
Michael Griffin
London, Ont. Canada
[email protected]
**********************