An associate raised questions about hand-counting, demonstrating that our local experience with hand-counting over three years has had errors -- and the machine-counts (paper ballots in a central scanner) have been accurate.
Whatever system of voting and counting that will be used must be secure, accessible, recountable, and accurate, now and forever. Oops, _forever_!
History is nice, and a successful past is a terrific plus.
If I were intending to subvert elections and I was the maker or operator of the DRE's or scanner-counters, I would do just what he has noted has been done - make the computerized system work flawlessly-- for several years. Then the officials and the public would let down their guard. (How long would _you_ be willing to wait to gain control of US politics? Twelve years? Eight? Four?) After everybody got lax, and "trusted" the machines, I would trigger the shifty process to cause my desired election "results". Bang! I control the election results, and there is NO WAY to find that out.
End of story. We must have an independent double-check on WHATEVER system is used. I grant his concern that people doing hand-counting might (will) make errors. So, if hand-counting were the norm, how should we double-check them? By another hand-count or a machine count, of course.
Bottom line - there must be a double-check.
So for me, the valid question is: Since it seems desirable/essential that we will have a machine-count and a hand-count (at least partial), how much might the hand-count cost? That's why I developed an estimating tool.
Download a new Excel budget sheet I have prepared (handcount-mb.xls - now Version #11).
The spreadsheet calculates the costs of hand-counting by teams of workers (with some supervisory, training and other costs added) and uses variables like these and others: (Substantially Revised November 2007)
For a typical case in my county, with some optimistic parameters: 70,000 ballots & 30 races, staff at $15/hour and Supervisors at $40/hour, it calculates about $225,000 cost = $3.20 per registered voter.
You are welcome to "double-check" my work for accuracy and completeness - and share your edits. :-)
We should try to compare (input) actual data from Elections Offices - and the formulas might need adjusting.