Re: OVC-discuss Digest on "dumb scanners"

From: Marc Baber <marc_at_botworks_dot_com>
Date: Wed Dec 06 2006 - 17:07:45 CST

To: Charlie Strauss and Richard Johnson OVC-Discuss list
Fr: Marc Baber
Re: Dumb Scanner proposal

Thank you both for reviewing my proposed "dumb scanner" system ( per Jerry Lobdill's request.
I look forward to learning more about the OPS product from OVSI from the
link you provided. I'm guessing that the technology is similar but the
logistics are significantly different. My proposal is that in addition
to an official ballot scan, representatives from competing campaigns
and/or parties would be able to dumb-scan precinct ballot stacks
independently and post their raw data on the web. The raw data is not
accumulated, but actually contains a section for each ballot noting
which ovals were filled in. Thus, the scanners do absolutely no counting
or incrementing of memory locations, they merely produce a raw data file
which, in essence, says "Here's a list of all the ballot documents I saw
and what ovals were filled in on each one". Machine totals are never
accumulated and, so, are never cleared. By deferring all
accumulation/totalling to subsequent phases where information from all
precincts can be combined, preference ranking support is preserved. The
non-dumb element is the open source software that combines the raw data
file(s) with the appropriate Ballot Definition Files (BDFs). Each ballot
type or "style" has a unique BDF which assigns candidate names, yes/no
choices and ranking (for IRV) to the ovals on the ballot. The ballot
type or "style" ID, which indicates which BDF to use for a given ballot
is part of the scannable data on each ballot. There is no sorting and
stacking of ballots and, so, no "nightmares of sorting/merging issues,
discovery and undoing of mistakes". One polling place equals one stack.
The ballot stack from each polling place might contain ballots of
multiple types or styles, but they are all kept together as an auditable
unitary record of the ballots received at a given polling place.
Software later does the sorting and stacking logically, to avoid the
chain of custody issues and discrepancies that would likely arise
otherwise. A smart scanner is provided in each polling place that has
knowledge of the BDF files (so it isn't "dumb") for the express purpose
of checking for overvotes and undervotes (so undervote/overvote
protection is provided). This polling place scanner also produces a raw
data file in dumb mode at the end of the day, providing a record of what
the pollworkers actually delivered to the couriers who carry the ballots
to central county facilities. At the county facilities redundant,
independent dumb scanners representing multiple competing interests in
the election are each given the opportunity to scan one precinct stack
at a time. A subsequent message will detail this process further. There
is nothing about chain of custody or ballot mutilation in this proposal
that would necessarily be handled any differently than today's optical
scans. I do propose a stricter model of witnessed couriers from polling
places to county-wide counting facilities using buses and vans, however
the technology doesn't require that-- it's just a separate issue. Ballot
mutilations are handled, as usual, by having observed pollworkers
transfer the markings from the mutilated ballot to a fresh ballot,
notarizing said copy and keeping the original mutilated ballot in an
evidence bag for later verification, if necessary. The fresh ballot is
scanned in place of the mutilated one. I think I've addressed all of the
concerns in your posts. Please let me know if you still have any
reservations whatsoever as to the viability of my proposed redundant
dumb scanner system. Thank you for the reference to the OVSI OPS system
and for taking time to review my proposal. Best regards, Marc Baber P.S.
I would welcome employment opportunities related to elections system
design and development if anyone knows of opportunities. Message: 3
Date: Tue, 5 Dec 2006 18:54:54 -0500 (EST) From: charlie strauss
<> Subject: Re: [OVC-discuss] dumb scanners? To: Open
Voting Consortium discussion list <>

Hart(of the hart intercivic) is marketing a scanner based paper ballot reader. The non-dumb element comes in the software that reads the ballot image files.


I have not yet read his proposal but there would need to be some means of discriminating totals for different ballot styles (i.e. different legislative districts). It's a non-trivial issue so I assume he may have addressed it. In precint voting one could possibly get away with one scanner per style as in most locale's a precint only has a few styles (usually 4 or less during primaries) In early voting there can be hundreds of styles. Bernallillo county had to used 43 opscans for each polloing station, each programmed for multiple styles, to accomodate early voting, and bernalillo county would be podunk comapred to precints in major cities.


Another issue would be overvote protection. The scanner would have to know something about the structure of the races to do that, not just xy positions. Likewise, reports of simple totals by XY posiiton would not suffice for Ranked Choice voting.

the trickiness in the opscans is the assinment of these by ballot style and the overvote protection, which requires some additional logic in the ballot definition.


Okay I just read his lengthy document. In a nutshell the scanners are "Scantron test scoring scanners used in schools for over three decades". He requires all the ballots to be gathered centrally then sorted into stacks by preinct and ballot style. He does not say so but presumably the machine totals would be cleared between stacks and the new-xy totals recorded for each stack. (This would not even be legal under most states current laws but that could be changed)

The system would not work as described for the reason I gave in my previous two posts. e.g. No over vote protection, no ranked preference voting. Plus it creates new problems of chain of custody before the ballots reach the sorting room, nightmares of sorting/merging issues, discovery and undoing of mistakes, and confusion when dealing with things like mutilated ballots that won't scan (no easy way to enter XY positions). In short the propsed system needs a lot of patching to make it robustly workable in any real USA style election (it would possibly work in simple elections like those found in parliamentary countries) --- yet it's the added sophistication inherent in those patches would become the real points to criticize.

The genius of the OVC system is not it's basic concept, it's the clever simplification of the intracies and self checks on the process. I dont' think anyone here disputes that "dumb" = "good".


Message: 6
Date: Tue, 5 Dec 2006 17:30:09 -0800 (PST)
From: "Richard C. Johnson" <>
Subject: Re: [OVC-discuss] dumb scanners?
To: Open Voting Consortium discussion list <>

Marc's proposal is not all that far from the OpenPrecinctScan (OPS) product of Open Voting Solutions, Inc. (www. OPS uses an off the shelf Kodak i40 scanner (with Open Source Linux drivers) to process hand-marked paper ballots. OPS then uses Open Source software to tally the records, first counting the mark (or not) in each checkbox on the ballot. There are a few more features, but that is the gist of it.

The scanner merely helps to count the ballot selections. There are a few more features after the marked boxes get counted, but this system is quite close to Marc's suggestion.

-- Dick

 Marc Baber 
 The Bot Works, Inc. 
 P.O. Box 5008        Phone: 541-485-8446
 Eugene, OR 97405     FAX:   541-485-8446

OVC-discuss mailing list

= The content of this message, with the exception of any external
= quotations under fair use, are released to the Public Domain
Received on Sun Dec 31 23:17:06 2006

This archive was generated by hypermail 2.1.8 : Sun Dec 31 2006 - 23:17:16 CST