When the reader doesn't calibrate on the leading
edge of the ballot, the reader sends zeros to the data stream as it is
supposed to.
Right.
The readers fail to calibrate on perfectly good ballots. Often.
[Ian
S. Piper]
Readers can miss a calibration,
but it's usually very rare. Obviously, you have seen an exception.
If the ballots are perfectly good, then you probably have a reader that needs
to be serviced. Can you be more specific as to the units that you have
seen fail, where they were and when they failed, and under what conditions
they were operating? Otherwise send the unit in and we'll take a look at
it.
You'd think Global could handle a simple thing like
that with its central count firmware as it has done with its precinct count
firmware (PC 1.94p and greater).
Tell me how you would like us to handle a
stream of 0s from perfectly good ballots, and I'll make the firmware change
monday. I think you'll find its pretty hard to determine voter intent
from a stream of 0s. Except in Florida perhaps.
[Ian S.
Piper]
How about changing the reject message from "BLANK
BALLOT" to "BALLOT NOT READ" or " INVALID BALLOT" or
"READER CALIBRATION ERROR"? Or something else that wouldn't be confused
with a blank voted ballot. Maybe then the ballot can be re-fed instead
of everybody staring at the ballot with a quizzical look on their
face.
It
sounds like we should be sacrificing to the gods to save us from our
software.
The gods would settle for working
readers.
[Ian S.
Piper]
I'll ask them next time I see
them.
Ken