Page 1 of 1


Andrew Wright

From:     Chuck Logan [clogan@co.hartic.com]
Sent:      Wednesday, March 31, 2004 1:02 PM
To:          'Neil McClure (E-mail)1

Cc:          'Michael Lawless (E-mail)1; 'Victor Babbitt (E-mail)'; 'Sandy Green (E-mail)'; Travis Harrell (E-mail);

awright@co.hartic.com

Subject: RE: Boulder County Questions for Hart
Travis provided the following three scenarios which are in response to BoCo questions regarding the tabulation validation of BN.

In all options, it will be very important to record how ballots are resolved so that the results are the same between the hand-counted
tabulation and the Tally reports. We may also wish to screen the ballots for

1 have included my comments on each option below. For options 2 and 3, a question I have is that is it possible to tabulate the
validation ballots into a separate Tally database?

Chuck

Tabulation Validation Process

First option...hand count entire precinct and compare with Tally precinct
report...is least disruptive, requires only procedural measures and no
programing impact.

To facilitate locating the ballots for hand-count, the BN "Scanned Ballots by Precinct" report can be used to identify the batches that
the ballots are in. The report needs to be run on each BN system.

Second option...identify batches for hand recount and separate Tally
tabulation...can be done using procedural measures with Sys 3.X incremental
MBB functionality of BN...can't currently accommodate the special report
issue of including the "Validation Sample" column in the reports...I would
suggest the least disruptive procedure short of Option One is to hand count,
scan, resolve, write CVRs, tabulate, report and validate results of the
validation batches as the first step in the scanning and tally process.

This option requires a separate MBB be used to carry the scanned ballot CVRs for the hand-counted ballots. This option works best if
we can tabulate this MBB into a separate Tally "Validation" database, and run reports against that. Then the MBB can be returned to
the normal workflow and scanned as usual with all the other MBBs into the "live" Tally database.

However, if the validation must take place on the "live" Tally database, then I agree with Travis that the validation should be done as
the first tabulation step so a report can be run showing just the results of the hand-count MBB. This option has the potential to disrupt
the beginning of the tabulation process, because all the hand-count ballots would need to be not only scanned, but resolved as well so
they could be written to the MBB. Then, if discrepancies are found, further tabulation may be placed on hold until the discrepancies
are cleared up.

Third option...random batch checks during the scanning process...this could
be done procedurally in Sys 3.X also...again, would require a separate MBB
to record CVRs of a specific batch, tabulate the batch and compare the
incremental change in the totals with the previous totals, and that should
equal the hand count.

This option seems to be the most complex. It also requires a separate MBB. It could be quite time-consuming and error-prone to
identify the incremental changes between different Tally reports and cross-reference to the hand-counted ballots.

Any special reporting of the "Validation Sample" results in Option Two or
Three would require engineering support.

4/2/2004