Calibration drifted to RIGHT, Palm Beach 2004 (Score 1)

I witnessed my wife's attempts to vote for Kerry in precinct 1196, St Edwards Church, Palm Beach FL 33480. about 8am on Sequoia EDGE machines
She pushed Kerry at least 3 times, each time a Bush Vote displayed,
called me over and I suggested that she not push so hard on the
screen, and push DIRECTLY on the X for Kerry and it worked. the
summary at end stated a Kerry vote.
My machine gave no problems, We voted early, to go answer phones for
the Palm Beach county Democratic HQ.

During my stint, I answered 2 calls relating to VOTER COMPLAINTS:
"I Pushed the Kerry Button, and get a Bush Vote"
After the first one, I called Kerry Lawyer pool, and their response
was "seems to be happening everywhere","voter workers have a
procedure to take machine off line, and re-calibrate it". The 2nd
call, I told to relay the information to "demand a re-calibration".

After thinking about this problem, (with 40 years of Computer programming
experience), I thought about, how to debug the program, REQUIRING
RECALIBRATION enough to make it a STANDARD PROCEDURE. Then the though
came to me that it may not be a BUG, but a "DESIGN FEATURE" as we euphemistically call some in the trade.

If your touch-screen routine was designed to properly execute when pushed
lightly in the DESIGNATED SPOT, it would be certifiable. If it was
pushed, elsewhere or HARD enough, what would the program do?
Perhaps, skew to a "Preferred candidate"? Based on proximity to the
DESIGNATED SPOT, perhaps this was calculated on a Pixel basis, and maybe
the size of the finger/footprint. What happens when one pushes farther on
the Kerry name verses the shorter Bush name?
Is the touch-screen map hard coded in pixel ranges, or a bitmap, which
could be modified by a clock routine? Or some other routine, unrelated to
voting such as Windows scheduler?, or the touch interrupt.

I would feel better about this if:

1) PBC elections commissioner had not ruled the no
"outsider" can experiment with the machines, hardware, software,
procedures, because "they are proprietary", AND that "would void the
warranty".
2) I heard ANY (documented or anecdotal) Bush voter complain that her/his
vote was MYSTERIOUSLY changed to a Kerry vote.

3) The Sequoia machine was debugged to not require re-calibration, and
the re-calibration problem was ADEQUATELY explained in the new version
report.

My feeling is that all Bush needed was to get 1 or 2 or more CHANGED votes
from EACH of these machines, allowed by an inattentive voter neglecting to
verify the final summary page, due to time/inattention problems, or
frustratedly let the vote stand without complaining.

This might explain some of the exit poll/verified vote discrepancy.

PS: My Palm PDA only needs recalibration after a CRASH!