I am taking this to the "Support" list where I think
this discussion should take place.
Lesley,
First, as for any error that was
displayed on election night, I was informed that an "SEH Error" was reported but
I have not been able to find out any information about when this occured or any
other information about this error. Rob had indicated he though the error
might have been the result of a memory card being removed before the transfer
was complete, i.e. while the AVTS was trying to write the audit log
entry, since the upload did actually complete.
Secondly, a laymans description of
the problem. During "Transfer" the AVTS unit, for performance
reasons, stores race and candidate information for the vote
center in memory (caches) rather that reading from the memory card. When a new
card is inserted for "Transfer" this data should have been cleared but was not.
Therefore when the next memory card was inserted the cached data was incorrect,
if the memory card was for a different vote center, and therefore the race
and candidate identifiers for the counts were potentially incorrect. The
problem did not accure when using the "Accumulator" since only the memory cards
from a single vote center were transferred during a "Transfer"
session.
I have added additional comments
below.
Tab
If you continually recieve error
messages on an operation this is a very good indicator of some problem and
should be investigated.
> > CENTRAL UPLOAD ONLY: I am also concerned, because due to this incident, > JoCo is opting to NOT modem in results from regional substations, much less > precincts. They are planning to set up parallel GEMS server systems to > upload all PC cards centrally, twice. That way they have two reports from > the same data loaded into separate systems to compare and audit. Is there > no way we can provide some kind of report off the Regional upload units that > can be compared to some kind of report from GEMS to be accountable that an > "anomaly" has NOT occurred? What is our "standard" check and balance proof > for returns on election night? Not doing modem uploads from the regions seems like a
strange response to their problem since the region uploads did not have any
problem. As for doing duplicate uploads, this would not have found the problem
since, if they uploaded the memory cards in the same order on each machine, they
would have identical (erronous) results in both databases.
> > ELECTION SETUP DATABASE TECH REVIEW: During the discussions around the JoCo > April 2 election, I have come to understand that a "typical practice" used > by technical support reps is to review the customer's database prior to the > initial ballot order generation. I understand that JoCo > (Connie/Kris/Debbie) sent you the May all-mail election database for review. > They were concerned that you turned around your comments back to them in > less that "5 minutes." Exactly what in the database do we look for when > reviewing? Are there certain checkpoints or typical mistakes that you look > for? Do you run ballot styles off the database to make sure they work? Do > we have a laundry list of review elements? I am being asked to provide > documentation describing what our election setup database review covers, can > anybody help me with the basic elements? How do we know a customers > database will actually perform as intended? Do we also cover the election > night report styles the customer wants to generate election night so we can > make sure the database is setup to accommodate their expectations? I'm just > trying to understand. > > LOGIC AND ACCURACY PRETEST METHODOLOGY: A reference has been made to be > sure that a regional upload client be sure to include that functionality in > the logic and accuracy pretest. Do we have documentation on all the > functionalities that should be included in our logic and accuracy testing? > Is there a checklist to use to prepare the test and then to run the test? > For the optical scan test deck, why are we not sending out the algorithms > being used to created the test deck? If we at least send out "how the test > was prepared" we could account for NOT sending out a "Test Deck Report" to > compare the L&A report to. > > Thanks. Les > > > > > -----Original Message----- > From: owner-bugtrack@gesn.com [mailto:owner-bugtrack@gesn.com]On Behalf > Of Jeff Hintz > Sent: Thursday, April 25, 2002 4:28 PM > To: bugtrack@gesn.com > Cc: stevem; barry > Subject: RE: Ballot Station v4.0.11 "anomaly" > > > Hi Lesley, > > There was no Error Message received for this problem that occurred on > Election Night. The transferring of the results to the GEMS host computer > went through without any error messages displaying a problem. The problem, > as Tab wrote up in the e-mail that Frank sent out to Connie, is that during > the direct transferring process, the information on the first Vote Center > that was transferred, was stored in the Cache memory of the AVTS unit, when > the second, or subsequent, Vote Center was inserted and then transferred, > the AVTS unit was then transferring those new totals, but with previously > stored Race ID's as well as the new Race ID's, and GEMS being confused, > simply put the totals into the Races where it thought the totals should go. > > That is the best explanation that I can give you. For a better and clearer > picture, you will have to refer to Ken or Tab. > > Jeff > > -----Original Message----- > From: owner-bugtrack@gesn.com [mailto:owner-bugtrack@gesn.com]On Behalf > Of Lesley Thompson > Sent: Thursday, April 25, 2002 2:31 PM > To: BugTrack@gesn.com > Cc: stevem; barry > Subject: Ballot Station v4.0.11 "anomaly" > > > Jeff Hintz, Ken, Tab, or Robert Chen... > > What were the ballot station error code messages that were received during > the central upload of the election day vote centers? Johnson County still > has questions and I am trying to finalize the Report of Findings. Also, can > you explain to me what the statement: > > "This problem only occurred when the second, or subsequent, Vote Centers > being uploaded during the same transmission session, had the same Race as a > previously uploaded Vote Center." > > And, why would the Leawood Mayor's race, which was not on any of the vote > center PC cards uploading centrally, end up with errant votes? And, where > did the 300 votes go that were not accounted for in the initially released > report? > > Can someone draw me a picture? > > Thanks! Les > >
|