Re: Not your ordinary barcode

From: David Mertz <voting-project_at_gnosis_dot_cx>
Date: Sun Apr 18 2004 - 16:57:52 CDT

On Apr 18, 2004, at 3:09 PM, Teresa Hommel wrote:
> Beware of forcing voters into dependency on "trust-me" technology
> (can't read
> the barcode? Trust us, it says the same thing as this small print
> where you can
> directly read your ballot choices).

I'm with Alan in STRONGLY urging new members to at least read the FAQ
before asking questions that are discussed there. Some stuff has been
covered on the mailing lists in more detail than in the FAQ
too--fortunately (because I made it so), the mailing lists provide a
nice interface to Google's search engine, so it's easy to find keywords
and phrases.

Anyway, I am also against any kind of invented, specialized barcode
that is "human readable." For one thing, it simply won't be readable
to most humans without quite a bit of prior training. Maybe a yes/no
on the left most race would be visible... but what about if there are
ten candidates; or what about ranked preference votes. Or even what
about the sixth race in, where it's easy to lose track of position (and
hard to remember which race is #6). Even beyond the fact it's not
really human readable, it introduces several more layers for malicious
or buggy software in the code that analyzes the barcodes. Code128 is
well analyzed and support code is already well tested (or likewise for
a number of other common formats). And anyway, if it *was* readable at
a glance, it would defeat the obfuscation that we deliberately
introduce to the barcode.

As to transparency, we've covered that too. The FAQ indicates that we
positively encourage independent implementations of the barcode reader.
  The encoding is part of a documented interface (And the implementing
reference source code is open); admittedly, the version of this API for
the demo is cruder than a generalized one will need to be. In
principle, a voter could bring in a wallet-sized validator (scanners
like that exist to photograph passages at libraries, for research
purposes). More likely, a contesting party can demand a statistical
spot check of barcode integrity. But even failing all that, our FAQ
explicitly indicates that in the case of discrepancy, the
human-readable version governs[*].

[*] There -is- a slight issue here. A blind voter will have verified
only against the barcode encoding, not the "readable" typeface. In the
abstract, you might want that voter's ballot to have the barcode
govern... but of course, you can't determine which is which without
losing anonymity. Still, spot checks and code inspection do a pretty
darn good job of making sure barcodes contain only and all of what they
should.

All that said, some members--including me--have expressed a fondness
for eliminating barcodes, and making the BVA stations be OCR machines
(that only need to recognize one particular font and layout). The
system requirements for this scheme are somewhat higher, but in
principle this could also maintain vote anonymity for vision-impaired
voters (they could perhaps determine the orientation of pages with a
notch on the page, and insert the ballot into the page scanner with
appropriate orientation). If the blind voter scans/vocalizes their
whole ballot page behind a curtain, they still prevent other people
from determining her votes. All of this is in the list archive, of
course.

Yours, David...
==================================================================
= The content of this message, with the exception of any external
= quotations under fair use, are released to the Public Domain
==================================================================
Received on Fri Apr 30 23:17:11 2004

This archive was generated by hypermail 2.1.8 : Fri Apr 30 2004 - 23:17:29 CDT