Politics: SnoCo Vote Counting: News

  Home  Evote

Table of Contents

• 2004-12-04

2004-12-04

1. While I am glad Martin Daniels has raised the public awareness of the issue, he and I do not see eye to eye on this topic (and haven't since the spring LATs).

2. The proposed way to do manual recounts for DREs is to print the cartridges out to PDF's and then count those as if they were paper ballots.

This sounds misguided at best, and deliberately misleading at worst. There is a possibility voters will become convinced this constitutes a manual recount of the DRE votes. They will be wrong.

For starters, the fraud may have happened before reaching the cartridges. So the only reason to count from the cartridges is to detect fraud in central tabulation under the precondition that there was no fraud earlier in the chain. Even if we assume the precondition, what we need is an alternative and trusted tally from the cartridges, to compare to the central tabulations.

3. To be a trusted alternative we can't rely on closed source software. That includes DRE vendor or OS vendor software used to convert from the cartridge data format to PDF. It would be trivial to "cook the books" at that step of the audit chain, to create whatever PDF the vendor wanted.

Further, it has been suggested once the PDF's are generated, we could parse them back out to a machine sensible format, and automate the tally. This provides yet further opportunities for error and tampering -- though if this software was open source that would at least be auditable. However the real issue is that it would be an absurd round trip from raw data to PDF and back. PDFs are not easy to parse, so we are essentially left with eyeball tallies of PDFs created from raw data.

4. Instead we need:

a) Assurance that the cartridges are not being tampered with as we speak.

b) Full specification of the data format on the cartidges, and the raw data made available on open postings. This allows anyone (including someone from the Ukraine, where they take voting seriously) to analyze the data and prepare a tally. With dozens of researchers reporting their tallies, and demonstrating how they reached the numbers, we will rapidly iterate to a trusted tally.

5. Even if we got a trusted tally from the cartridges, that does NOT constitute a manual recount of the DRE votes. Those votes are lost irrevocably, in contradiction of law re manual recounts. What we have are some data cartridges prepared by closed source software under suspicious circumstances. With any luck the data on the cartridges is the same today as it was at the end of election day, but that is about it.