Re: Alternatives to a single bar code

From: David Mertz <voting-project_at_gnosis_dot_cx>
Date: Wed May 05 2004 - 10:43:01 CDT

On May 5, 2004, at 5:22 AM, Karl Auerbach wrote:
> multiple bar codes could lead to reader misreads...
> individual bar-codes from different ballots be scanned as if they were
> from the same ballot.

Right-O. One barcode is better for this type of reason.

> I don't see any hope that we will be able to stay within the data
> limits
> of the 1 dimensional bar codes - we are already pushing the limit, even
> without a digital signature message digest or end-to-end check-code.

This is not quite true. Well, obviously if we use Karl's standard of
"worry when we get to 10% of theoretical limit" we're there. But that
standard makes more sense for quantities--say bandwidth--that pretty
much expand monotonically. Elections don't.

The demo merely used an inefficient encoding. You can calculate the
optimal encoding for any election using my election-entropy.py tool.

We can fit the demo data in about 35-40% of the space we actually used.
  For the demo data, 53 bits is what we need. That leaves enough room
for a crypto signature and globally-unique ballot ID to be added (and
even some small ECCs, beyond those inherent in Code128).

I second Alan's opinion that the data requirements are an empirical
question. If 99% of elections can get by with the simpler 1-D barcode,
we should stick with simple (for most jurisdictions). If lots of
elections need 2-D, we should start working towards that.

So first thing would be for someone to locate ONE historical election
that can't fit in Code128. I am unaware of one. For example, the
California Special Gubernatorial contest fits fine. Chicago judicial
retentions with 40 Judges and a handful of Congresscritters, Senators,
initiatives, etc fit fine (assuming a moderate number of candidates for
each contested race).

I'm not particularly a FAN of 1-D codes like Code128. Less so than I
think Alan is (because he's too fond of the el-cheapo CueCats he
found). I'm not an enemy of 1-D either. But before we conclude that
Code128 will inevitably hit a storage limit, lets find some concrete
evidence for that.

*Keeping in mind* that elections, unlike bandwidth, HDD size, CPU
speed, or spam, don't follow a Moore's Law style monotonic curve.
Instead, the range of sizes that exist in the last half dozen decades
is pretty much the same as what we expect to find in the next half
century too.

  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 Mon May 31 23:17:14 2004

This archive was generated by hypermail 2.1.8 : Mon May 31 2004 - 23:18:15 CDT