Saturday, October 16, 2004

Voting System Standard needs work

My IEEE P1583 Voting System Standards friends at,

In my professional opinion, it will take much more time and effort to try to live with this poor specification than it will to fix it before it is released.

What the document needs is an abstract architecture of a voting system, a tight set of requirements for every element, a specification of primitive functions, a set of metrics, and a set of acceptable methods to prove compliance.

The document should not be wedded to any implementation. Technology evolves too quickly. The doc is currently wedded to DRE voting equipment that will by my guess be outmoded within the five to seven years it will take to repair this document once it becomes approved by the EAC, incorporated into state statutes, learned and implemented by vendors and independent testing authorities, embodied in test scripts, and so on and so on … The document is totally inadequate when it comes to paper ballots, and does not address classes of existing, new technology such as that of Vogue Election Systems. The document addresses device-level concepts rather than system level concepts. To be useful, all of the elements of the election system must interoperate and be tested as a system to prove secure, accurate and verifiable functionality.

This document, to be useful, must speak to high level rather than implementation level ideas. The architecture and primitives should be used to construct two or three reference implementations (at the high level) and include complete tests that verify each implementation.

Short of this, I believe that we will have failed.