FW: Allegations of reliability problems when there is a paper trail
From: Sanford Morganstein [mailto:s.morganstein@populex.com]
At 11:16 AM -0600 12/10/04, Vern Williams wrote:
>Stan,
>I think you need a reality check. The most failure prone component
>of the "paper trail" system you seem so obsessed with is the printer
>with it's mechanical parts and ink delivery system. You state many
>of your pet opinions as fact which is far from true.
<snip>
According to this post, Vern writes that the paper trail is the most failure
prone component.
We don't think so for the following reasons:
Take a precinct count optical scan on which the voter can have the ballot
"verified". Such a system may help a voter determine if an erasure was
complete enough, if there were any stray marks interpreted as a vote or if
there were out and out overvotes. I won't get into in this post that such
"verification" is not really private (although it could be) the reality is
that many election jurisdictions turn off undervote checking because of its
lack of privacy.
But here is the point on reliability. If this scanner goes down, the whole
precinct goes down. Votes cannot be verified, and if the system uses
so-called precinct count the tabulation is down although it can be
reinstated after the voters leave by opening the ballot box. But, the point
is, a single point of failure, on a mechanical device puts the whole
precinct down.
Now...on to DREs. Some are networked in the polling place with a
master/slave or client/server arrangement. Some use memory devices to pull
out the results from the DRE. Despite redundant memories in the DRE...there
are several single points of failure.
Now consider a non-networked electronic machine that prints a paper ballot.
If it goes down, one station goes down. The precinct is still in business.
A non-networked machine going down is not great news...reliability standards
should be tough. But, such an approach is intrinsically more reliable than
the DRE or optical scan.
Sandy Morganstein
Populex