Vermont SOS: bulk of recount changes have come in towns that use HCPB

[eomer]

I don't know the veracity of this report but am posting it anyway because, right or wrong, it illustrates a principle: that no high-level approach to election systems is a panacea. With HCPB or any other approach there are many ways that an election can be rigged or can fail due to error. In other words, audits, citizen observers, and other forms of vigilance are required with HCPB just as they are with opscan or any other system that can be invented. Emphasis is mine.

-snip-

Secretary of State Deborah Markowitz said a preliminary look at the recounts has shown the bulk of changes have come in towns that use paper ballots, which have to be counted by hand.

Elections officials in those communities split up ballots in lots of 50 or 100 and count them, recording the results on tally sheets. Those totals then are transferred to a master list for a final townwide count.

In the auditor's race, it appears that in a number of communities, votes that should have been transferred from the tally sheets to Salmon's total actually went to Levy, whose name was just above on the ballot.

"What we seem to be seeing is that in towns that hand count their votes, particularly the larger towns that are hand-counting votes, there are some number of Salmon votes that were incorrectly marked in the Levy column," Markowitz said.

She attributed the problem to the design of some tally and summary sheets, which her office likely will consider redesigning.

Hardwick is an example. The official results reported to the secretary of state's office gave Salmon 518 votes and Levy 44. But the more painstaking hand recount gave Salmon 542 votes in Hardwick to 20 for Levy.

"Election workers were working late at night and it's easy to make that kind of error," Markowitz said.

-snip-

http://www.boston.com/news/local/vermont/articles/2006/12/11/ap_ahead_salmon_picks_up_votes_as_clerks_complete_vote_recounts/

_{Edit: typos}

[garybeck]

1. yup, and when I went to observe our "audit"

the Director of elections was quick to point out to me that even in the controlled environment of an audit, there are discrepancies in the hand counts. they'll probably use this against us... but the bottom line is, just because one system has flaws, it doesn't mean anything about another system. the flaws for each system are different and need to be addressed. The fact that people can make mistakes counting or tabulating votes does not mean that proprietary software code is acceptable!

[diva77]

2. it sounds like they weren't using the sort and stack method which is the

most accurate method -- can you get more details on their procedures for HCPB?

[WillYourVoteBCounted]

3. reference for sort and stack?

Hey, I know you have linked that somewhere before,
but I can't find it.

I don't think our recounts and audits are conducted in the
sort and stack, but should be.

[diva77]

4. I think you're talking about the NSF letter

it's in Appendix C of the NSF letter report on Electronic Voting

http://www.nap.edu/catalog/11704.html

Copyright © National Academy of Sciences. All rights reserved.

Letter Report on Electronic Voting

Page 13

THE MANUAL COUNTING OF PAPER BALLOTS

Counting paper ballots is inherently manual, but there are better and worse ways
of doing it. One common method is based on ballot reading and tally marks. One
member of a two-person team reads the ballot, declaring those legal votes apparent
from the voter’s marks. The second team member places a mark on his/her tally sheet
for the candidate receiving a vote. This method involves the possibility of a mistake
because the ballot is examined only once or a mistake because only one person is
doing the tallying. Since this method commonly involves reading through the entire
ballot, the ballot reader's eye and brain are not focused on looking for a single type of
data, and thus the reader must expend mental effort to distinguish among the contests
in which choices are made. (INFERIOR METHOD. I think this is how they did it in VT)


(BEST METHOD) At least one state (New Hampshire), in its state recounts, has been using another
process that seems to be less subject to error. This process, based on the use of ballot
sorting and piles, involves one member of a two-person team picking up the ballots and
placing them in piles corresponding to each choice in a particular race. The other team
member observes each ballot as it is placed in a pile. After the sorting process is
complete, one team member counts each pile in stacks of 25 and then the other team
member recounts each stack. This process enables at least two persons to
simultaneously examine each ballot at least once, and to keep things simple by
identifying choices in a single race at a time. If one person makes a mistake, the other
can catch it. This method is often modified so that each ballot is rechecked during the
stack-counting process. Hence, each ballot can be seen two times by each member of
the team, for a total of up to four views of each mark on a ballot in each race. The ballot
sorting and pile method, which involves as many examinations of the same ballot as
there are contests, is noticeably faster than the ballot reading and tally mark approach.