[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: Hart InterCivic User WARNING!
Neal
What are they doing when the NO vote is circled, or otherwise not marked as
instructed but definitely marked?
Will there be an audit trail that shows that the number of over votes in the
original count is equal to the increase in NO votes plus the number of over
votes in the modified count?
Etc.
Al
Al Kolwicz
2867 Tincup Circle
Boulder, CO 80305
303-494-1540
AlKolwicz@xxxxxxxxx
-----Original Message-----
From: Neal McBurnett [mailto:neal@xxxxxxxxxxxxxxxxx]
Sent: Monday, October 31, 2005 3:48 PM
To: cvv-discuss@xxxxxxxxxxxxxxxxx
Cc: John Gideon
Subject: RE: Hart InterCivic User WARNING!
The article below is great, but it is wrong when it says that in
Boulder
> any [ballots] with a crease through an option box will be
> individually resolved.
They are not doing individual (manual) resolution if the crease goes
thru a box that is voted by the voter. Also, if the voter votes for
one box, and a fold goes thru the other box, individual resolution
will only happen if the system reads the fold as a vote, and reports
the contest as _overvoted_.
They are only doing special resolution in the unusual case where a
fold crosses a box in a race which has no vote for either option (an
_undervote_). Those ballots will be run in a separate batch for which
they will configure the system to "manually resolve undervotes", to
prevent the system from counting the fold as a vote. This was
reported before, as I recall, in the papers. It was also confirmed
for me today by Patty Stahl, while I was doing some pollwatching.
This strategy relies on the ability of their normal practice of doing
manual resolution of _overvotes_ to catch cases where the fold crosses
an empty box, the other box is filled in, and the system mistakenly
counts both the fold and the filled-in box, and reports an overvote.
FYI, Joe dug up some more info on the Yakima county problems with a
Hart system, and I've included part of his email on that below.
At least Boulder County doesn't do what Yakima did! They autoresolved
damaged contests, and didn't re-scan when doing a recount, and didn't
do a hand-count audit (until forced to in a third round of review....).
But of course there is plenty more for us to work on fixing
locally....
-Neal
----- Forwarded message from Joe Pezzillo <jpezzillo@xxxxxxxxx> -----
Date: Mon, 31 Oct 2005 14:34:22 -0700
From: "Joe Pezzillo" <jpezzillo@xxxxxxxxx>
This is a link to the original Yakima info (also by John Gideon) which
includes an image of the problem and the discussion with Hart, as well as
the Kodak follow-up note.
http://www.votersunite.org/info/yakimaproblemreport.asp
Joe
--- ... Forwarded from VotersUnite.org, via Al Kolwicz... ---
----------------------------------------------------------------------
Voting Systems User Warning
Voting System: Hart InterCivic Ballot Now optical scan system, using
Kodak i800 Series Scanner
Problems Found by Election Officials: During a hand recount of
ballots counted in Yakima County, Washington, it was discovered that
24 properly marked ballots had been counted as undervotes and had
not been tallied. An investigation was carried out by Hart
InterCivic employees who looked at the ballot images. The
examination revealed that 24 ballot images contained a white
vertical line spanning the entire length of the ballot. The line ran
through the left portion of all option boxes in the 4th column of
each ballot.
Hart reported that since the white line *whited out* a portion of
each option box, Ballot Now was unable to detect at least 90 percent
of each *target box* and therefore classified each contest in the
4th column of each ballot as a Damaged Contest. In this case, the
4th column included contests for Congress, Governor, Lt. Governor,
Secretary of State, and State Treasurer.
Hart also confirmed that during processing of the batch in question,
the option to Autoresolve Damaged Contests was selected.
Consequently, Ballot Now (in Autoresolve mode) confirmed and
recorded the damaged contests as undervoted.
In another reported incident, officials in Boulder County, Colorado
discovered that when a fold in the ballot covers an option box, the
machine may misread the ballot. Of 429 test ballots read, seven
(1.6%) were misread because of such a fold.
Solutions: The Hart representative recommended a regular cleaning
program for the scanner during heavy use periods. The *Kodak Series
i800 Scanners User*s Guide* recommends thoroughly cleaning the
scanner after every 8 hours of use. According to the troubleshooting
section of the manual, the white line that caused a miscount of
ballots in Yakima County can be directly attributed to dirty imaging
guides. VotersUnite and VoteTrustUSA agree with the Hart
representative but, because people*s votes are involved we recommend
a thorough cleaning more often than recommended by Kodak.
The Boulder County problem may not be as simple to fix. The county
is proposing that all ballots be inspected, and any with a crease
through an option box will be individually resolved.
In all cases, VotersUnite and VoteTrustUSA strongly recommend that
the *Autoresolve Damaged Contests* option not be selected. Any
damaged or questionable ballots must be inspected by an election
worker to determine why the system identified the contest as
damaged.
VotersUnite and VoteTrustUSA also point to both of these situations
as substantial evidence for why audits MUST be conducted by hand
counting a percentage of the original paper ballot batches and
comparing that tally to the computer*s tally of scanned images of
those ballots and to the total from the tabulator.
by John Gideon
Information Manager, VotersUnite.Org and VotetrustUSA.org
----- End forwarded message -----