"sfb" wrote in message news:8K4tf.9457$3Y3.8158@trnddc02...
Please explain what "auditing" a sample of the paper receipts establishes
since the only thing that matters is the total votes cast for each
candidate.

Please explain what "auditing" a sample of taxpayers establishes since the
only thing that matters is the total revenue paid by all taxpayers.

An election isn't a production line making a gazillion identical widgets
per day where sampling will tell you something about meeting
specifications.

Actually, in some respects it is exactly like a production line making a
gazillion identical widgets per day. The main difference is that when
you're making widgets, usually you don't have to worry about someone
intentionally screwing it up.

But otherwise, the concept of a random check for some subset of the total
production is identical. By auditing, one can have some degree of
confidence regarding the complete "production output", even without
double-checking 100% of your production output.

A election produces a different product for each candidate on the ballot.
The only way to know how many votes each candidate got is counting every
single vote.

The auditing isn't about knowing "how many votes each candidate got".

Early in the day 2004 exit polls predicted a Kerry win only to be proven
wrong by the actual votes since the sampling was apparently biased to only
asking Kerry voters.

Non-sequitur, and besides there is still ample disagreement regarding the
true reason(s) behind the discrepancy you're describing.

Pete