http://homepage.cs.uiowa.edu/~dwjones/voting/west02/
West02, Jones on End-to-End Accuracy Standards, Slide 1
- Accuracy standards should ask: How close does the outcome of the election come to the actual intent of the voters?
- This is an end-to-end question in the sense Saltzer, Reed and Clark introduced in their 1984 paper. End to End Arguments in System Design.
- Recounts of punched cards and mark-sense ballots frequently exceed agree with the first count to 1 vote in 5000.
- Exit polls frequently predict precinct-level results very well (I have no numbers).
- These are not perfect end-to-end measures, but we can use them!
- Accuracy standards such as the 1 in 500,000 or 1 in 10,000,000 figures cited in the FEC standards are almost irrelevant!
- ... the "machine model," ... counts as valid only those votes that the vote tabulating machine can read and record.
- The machine model ... relies on an objective tabulating machine ... if a vote is properly cast according to the instructions given to the voter, the machine will count it.
11th Circuit, Touchston and Shepperd vs. Michael McDermott, 2000.