Re: Compression, encoding, entropy

From: Karl Auerbach <karl_at_cavebear_dot_com>
Date: Sun May 02 2004 - 13:57:57 CDT

We can't forget that one of the potential sources of data errors will be
software programming errors.

The software to do the encoding between a ballot to barcode (and vice
vesa) can and probably will be rather complicated - and there may easily
be bugs lurking in that code.

For this reason I believe we want some simple checksum/digest that can be
computed against the fully-unpacked data representation before encoding,
then embedded in the encodeing, and then validated after unpacking.

This can be something simple - perhaps simply numbering the contests on
the ballot 1..N and, in each contest, turning the choices made into a
simple binary number. Then we simply go through the contests and sum
contestnumber*votenumber. (This is analogous to Fletcher checksum in
ISO/OSI in which the checksum is the sum of the value of each byte
multiplied by its position in the sequence of bytes.)

As I type this, and as I think about the awful ramifications of
discovering a bug in the encoding routines during the evening after an
election, I am beginning to think that the ballot should contain *two*
barcodes (each printed at least twice), with each encoded by a 100%
distinctly separate codepath with distinctly different software
althorithms. An election is like an airplane - redundant independent
systems are a good thing.

                --karl--

==================================================================
= 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 23:17:02 2004

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