Re: Can we talk details about adding Humboldt-style scanning to the OVC project?

From: Arthur Keller <voting_at_kellers_dot_org>
Date: Mon May 25 2009 - 11:47:26 CDT

Dear Jim, I have been supportive of publishing ballots (like the
Humbolt approach), for several years now. I have been an advocate of
scanning and publishing results of all ballots at the source of
scanning, dating back to before my OSCON 2007 talk. Please read my

Best regards,

At 12:32 AM -0700 5/25/09, Jim March wrote:
>As I understand the current OVC system model, the plan is to do a
>precinct-level tally by scanning the ballots at the precinct as an
>end-of-day procedure. This would match up with California law
>requiring such a tally be posted.
>The plan at present is to do the scanning via the barcode printed on
>the ballots and a barcode scanner.
>What I want to discuss is doing the end-of-day scan as graphic images
>stored on optical media. In other words, "Humboldt-style".
>At the precinct, we'd be talking about a small extra expenditure in
>hardware. The precinct-central scanning station would need a basic
>scanner worth about $200, with auto-feed, and a second optical drive
>in that computer. The first drive could (and I would argue, should)
>be a read-only basic CD drive, while the second would be a full
>"writer" (likely DVD). This can burn multiple copies of the data for
>bystanders, with at least two cut for the county election HQ.
>The second drive would be about $40 tops, less in bulk.
>Against these hardware costs, we actually get some advantages as far
>as end-of-day procedures. The scanner will have a "hopper" of at
>least 30+ sheets, and auto-feed. So the pollworkers end up doing less
>work on scanning as opposed to hand-scanning each sheet.
>This would of course also apply to central tabulation, with a bigger
>scanner of course. Likely two or more to increase throughput and
>cover some redundancy failure. The advantage here is that with bigger
>hoppers of up to 500 sheets, handling of the mail-in vote becomes far
>more efficient than hand-waving bar code scanners.
>The Humboldt project itself used Fujitsu scanners worth about $18k.
>They had small dot matrix printers on board that could implant a code
>sequence on some unused area of paper just prior to scanning, so you
>could link a given graphic scan to a given sheet of paper later if
>That won't be practical in precinct scanning I don't think, not unless
>we issue pollworkers some kind of hand-inker stamp thingie that
>increments one number on each "kachunka". That would add an extra
>step and I don't think it's critical.
>The other option as opposed to precinct graphic scanning is to haul
>the precinct ballots back to election HQ and run them through the same
>scanners that handled the mail-in vote. The advantage is that they'd
>get the same "sequence number stamp". The problem is that you then
>have to watch your paper transfers like a hawk.
>I prefer precinct scanning. I know for a FACT we're not going to get
>enough activists to watch every precinct paper transfer.
>All of this is about "worst case scenarios", when the elections
>managers are basically crooks. Which does happen.
>I know Alan thinks this is overkill. I've done too much election
>observation where the officials are malevolently evil to agree, and I
>hope to begin a rational conversation on this.
>Jim March

Experienced advisor to leading edge startups and
accomplished expert witness on patent infringement cases.
Arthur M. Keller, Ph.D., 3881 Corina Way, Palo Alto, CA  94303-4507
tel +1(650)424-0202, fax +1(650)424-0424
OVC-discuss mailing list
By sending email to the OVC-discuss  list, you thereby agree to release the content of your posts to the Public Domain--with the exception of copyrighted material quoted according to fair use, including publicly archiving at
= The content of this message, with the exception of any external 
= quotations under fair use, are released to the Public Domain    
Received on Sun May 31 23:17:04 2009

This archive was generated by hypermail 2.1.8 : Sun May 31 2009 - 23:17:07 CDT