No skulduggery needed
BY HARRY J. SAAL Nov. 19, 2000, in the San Jose Mercury News
FIRST, an admission. I supported Al Gore in this election.
Now, for some background. I entered college
in 1960, and spent my idle hours deep in the
bowels of the computer center at Columbia University. Programming
the giant IBM mainframe systems of the era meant punch cards. Lots
of punch cards. The punching process produced mountains of confetti,
so-called ``chad.'' The stuff billowed out of card punches, fell into
crevices and needed vacuuming with powerful but tiny sucking devices.
Electronic card punches made their holes smartly; but using a tiny
manual pin, as is done in election booths, is quite another affair. I've
always made sure to remove my own ballot, turn it over, and manually
pull off the tiny pieces that flap behind the card. What if I didn't take
this extra step whenever voting? Sometimes the chad will fall off when
the operator fans, snaps and bends the ballots before feeding them into
the reader apparatus. But sometimes the chad just slips right back into
its little square nest. Each time you read, and re-read the deck of
cards, more and more of the chad makes its way off the cards, thus
increasing the total number of votes detected in the process.
That's why the Palm Beach computer recount found more votes for
both George W. Bush and Al Gore than the first time around. And the
manual recount found even more.
Suppose a totally unbiased recount finds 1 percent new, perfectly valid
ballots, compared to a previous count. In a county like Palm Beach,
where the voters preferred Gore to Bush by a two-to-one ratio, of
every three additional valid votes, two will go to Gore and one to
Bush, for a net gain of one vote to Gore. In a county where 450,000
votes were cast, a perfectly unbiased recount yields a 1,500 vote
advantage to Gore. No cheating or skulduggery is needed, because of
the fact that manually punched ballots are systematically undercounted
by machines. According to their various manufacturers, a well-tuned
and adjusted machine may miss from near nothing up to 5 percent of
the ballots; this is not acceptable when differences of far less than one
hundredth of 1 percent are involved.
So all the blather about manual counting being biased, inconsistent or
potentially rigged is just that: blather. The fact is that a perfectly
fair
and unbiased manual recount in Palm Beach County could find over a
thousand net votes for the Democratic candidate.
But what about the other counties? What might they have shown, if
they too were hand-counted?
Based on information released by the secretary of state in Florida,
Gore consistently led in the counties using punch card systems,
whereas Bush won where optical scanners were used. The Volusia
County manual recount confirms that the optical systems in use had far
fewer under-votes than punch card systems. So Gore's true total was
undercounted, whereas Bush's wasn't. A full and fair manual recount,
across all counties, could result in thousands of additional votes to Gore.
So how will we know who really won the election?
That brings us to problem No. 2, the design of the Palm Beach County ballot.
I am convinced beyond a shadow of a doubt that over 10,000
Democratic voters in Palm Beach County spoiled or miscast their
votes as a result of the ballot design, whereas only one tenth as many
Republicans did the same.
Claiming the ballot is simple to follow ignores clear flaws in its design.
For instance, if you want to vote for the second candidate on the
ballot, shouldn't you punch the second hole down? If you want to vote
for the Democratic candidate, don't you just follow to the right of the
word ``Democrat?''
I believe Florida won't take into account the errors introduced by the
unfortunate ballot design in rendering its final, certified count. Indeed
there is no way to prove the intent of a voter after the fact. And while
holding a ``re-vote'' with a new ballot design would avert these
problems, it introduces a whole new set of difficulties.
There is extraordinarily strong evidence from Palm Beach County that
Gore voters sometimes punched either above or below their intended
hole (No. 5), hitting No. 4 or No. 3. Several publications have
examined the Buchanan vote, observing how inconsistent the number
of Buchanan votes in the county was, when compared to other
Buchanan-like surrogates. Buchanan himself has acknowledged his
belief that it was not his popularity that caused the incredible 3,400
votes he received in Palm Beach, and that some portion came by
accident from Gore voters.
Even more remarkable is the number of votes received by the Socialist
Labor Party candidate, David McReynolds. He received roughly 10
times as many votes in Palm Beach as he garnered in any other Florida
county. McReynolds' punch is just below the Gore punch, whereas
Buchanan's is above it. This is not evidence for a vast cabal of radical
voters among the Palm Beach electorate, simply human error at work.
Still not convinced? Look at the double-punched votes as well, where
a voter presumably chose one candidate, decided he made a mistake,
and then re-punched his ``correct'' intent.
Of all the possible two, three, four, etc. multiple punches for president,
only three combinations had more than a very few examples, in a total
of about 4,500 ballots in the 1 percent sample that was hand-counted.
The three common errors were 3-4 (Bush-Buchanan), 4-5
(Buchanan- Gore) and 5-6 (Gore-McReynolds). Curiously,
Gore-Buchanan errors were roughly three times the number of
Buchanan-only votes.
I assume that the 3-4 punches were really meant for Bush, whereas
the 4-5 and 5-6 votes were intended for Gore. For this 4,500 vote
sample, about 101 votes intended for Gore got spoiled, compared to
11 for Bush. It's pretty hard to get it wrong for George Bush with this
ballot design. The difference: 90 votes to Gore. This 1 percent sample
projects to a 9,000 vote preference across the county. Add another
1,000 net to Gore from the accidental Buchanan-only voters, plus
perhaps 300 of McReynolds votes, for an estimated 10,300-vote loss
for the Democrats due to the faulty ballot design.
Let me reiterate my belief that no court or election board will ever
award this 10,000-plus vote pickup to the Democrats. But a scientific
analysis of the outcome in Palm Beach convinces me that this was the
true intent of the voters in that county.
What if the present lead of a few hundred votes for Bush prevails?
Does Mr. Bush take Florida and proceed to assume the presidency?
I think not.
This state of affairs is precisely why our founders set up the Electoral College.
If the Bush statewide lead stays at a few hundred votes -- not even
one hundredth of 1 percent difference -- then the case of the disenfranchised
Palm Beach Gore voters ought to play in the minds of the electors.
The Electoral College, rather than the popular vote system, can
respond appropriately to precisely this situation.
The courts can't undo the disastrous effects of bad design in Palm
Beach County. But the electors can. In those states where they can
legally use their sound judgment, they ought to cast their votes after
a
thoughtful consideration of the above analysis. Clearly, Gore won
Florida, by at least 10,000 votes.
Dr. Harry J. Saal co-founded Network General Corporation (now
Network Associates) and is president of Cultural Initiatives
Silicon Valley.