response 162 of 216:

Nov 12 07:26 UTC 2000

They just had on cnn, the meeting of the Palm Beach county board of
elections.  They counted 1% of the votes by hand and showed a net plus to
Gore of 19 votes.  If you assume that 19 votes is an average for each one
percent, counting all 100% of the votes could produce an extra +1900 votes
for Gore.  Based on the fact that such extra votes could swing the
election, the board voted (vote was carried live) to order a manual
recount of all ballots in all precincts.  

response 164 of 216:

Nov 12 08:17 UTC 2000

I'd guess an important part of the recount would be to *also* count the
spoilt ballots, and perhaps try to take a stab at figuring out if there
is any sort of systemic fault in the spoilt ones.  If after a recount,
the margin of victory (for whichever side) is larger than the spoilt
ballots, then they can declare a victor and rest.  If the # of spoilt
ballots is larger than the margin of victory, and especially if there's
evidence of systemic bias in the errors, then that would seem to be
strong evidence that a re-election is necessary.

Presumably, there is no lack of hands that would be willing to count
ballots, so the real question may be a matter of figuring out how many
and which hands to trust, and coordinating the whole mess.  If I were
them, I'd probably be starting by asking how much time I had to do the
count in, figuring out how many people were needed to do that count, and
go from there.  I suppose if it turned out it was going to take longer
to do the hand count than to do a re-election, then that might be
another good argument to do so.

I haven't heard anything on those overseas ballots - things could get
*very* interesting if there's a net gain to Bush of, say, just under
2000.  Do you suppose we might end up having a 1 vote difference
determining the next president?

response 166 of 216:

Nov 12 20:43 UTC 2000

News that was glossed over a bit in last night's Palm Beach recount, that will
come back to haunt us, perhaps: They looked at 100 or so "overvote" ballots
-- where people had cast multiple votes. 80 of them were Buchanen-Gore (and
noone else), which is statistically significant support for the "confusing
ballot" theory.

response 168 of 216:

Nov 12 22:20 UTC 2000

On the grounds that, well, even though that's the law, it's not fair, dammit.
After spending a week telling Gore to lump it and live with what the rules
are, even if they don't seem fair, Bush is now turning around and (in my
opinion) whining that the legal but unfair rules may cost him the election.

My understanding, and the understanding of just about every one of CNN's
pundits, is that Gore hast he right to ask for a manual recount, and the
jurisdictions in question have a right to proceed with as much or as little
of a recount as they wish to, and neither the Secretary of State of Florida
nor the Federal Government has much to say in the matter. If the manual
recount differs from the machine count, then it becomes a matter of dispute,
but I beleive the last count (manual) becomes the "official" one.

Now the State of Florida (proxy for Jeb Bush) is rattling ITS saber and
insisting it'll simply ignore anycounty certifications it receives after
Tuesday. Given that Palm Beach County's hands have been tied by an injunction
not to certify, it'll be interesting to see where this falls out.

response 170 of 216:

Nov 12 23:21 UTC 2000

I think all of this backs up my contention that this is something that needs
to be settled by the courts.

I suppose Bush's lawsuit is going in that direction, but the comments that
go with it certainly seem more divisive than useful.

response 172 of 216:

Nov 13 00:17 UTC 2000

lets see...bush says this shouldnt be decided in the courts, then he
files the first lawsuit...

bush says there shouldnt be handcounts, yet his people agreed to a
handcount of some ballots in seminole county.  And he signed the law in
Texas that states explicitly that handcounts are a logical solution to
balloting disputes.

does the word "hypocrisy" come into play here?

The fact is that had this situation been reversed, and it was Gore who was
ahead by three hundred votes, you KNOW Bush would be pushing for
handcounts in counties where he thought his vote might be undercounted.  

response 174 of 216:

Nov 13 02:27 UTC 2000

If that 1% of ballots were chosen at random, and found an additional
19 votes for Gore, an estimate for the standard deviation in that
number is about 4.4, so the 95% interval is about 6 to 32, or, for the
full count, 600 to 3200. Looks worth recounting, as the null hypothesis
is rejected at the 5% level. 

response 176 of 216:

Nov 13 04:29 UTC 2000

#174> Except that the general media opinion seems to be that the precincts
for the 1% were selected because of their concentration of Democrats, not
randomly. All the same, if the manual recount translates to about 1000 votes
for Gore, which it easily could, that would tie everything up.

#175> Palm Beach County's in a bad spot, legally. They have a judge telling
them they can't certify the results just yet, and the state telling them they
must certify the results. Unless the judge or the state bends, their window
for certifying (legally) is only an hour or so.

response 178 of 216:

Nov 13 14:30 UTC 2000

I'm sorry...but I would want to know who really won if I was a candidate. 
And I wouldn't point the finger at anyone else.  Bush is afraid, Gore is
afraid.  I just don't care anymore.  if counting the votes will determine a
winner, then we will count the frickin votes.  But an injunction to hand
counting the votes because it would prevent, possibly, a bush win, because
of either the outcome, or the "human" error of counting?  God, if you're THAT
worried, hire computer people or accountants to do it.  They're used to
numbers, if you can't trust others to do it.

I didn't like EITHER of them.  And I like them less.  And I don't think that
this whole fiasco is as important as they are letting on, since the electoral
college wouldn't have voted yet anyway, and the new pres isn't until January.
We're obsessed with the luxury of "predicting" the winner by a landslide, that
when it is actually a close race and have to wait "it's all falling apart!!!!"

response 180 of 216:

Nov 13 15:39 UTC 2000

The questionable accuracy of hand counts is not because humans can't count.
It's because of all the nebulous determinations of the "voter's intent".

  Ballot 1:  Hole for Bush is unpunched, but dimpled slightly.  Nothing else
     punched.   No vote or Gore vote?  Let's say "no".

  Ballot 2:  Hole for Gore is partially punched out, with the chad still
    hanging by two corners.  No vote or Gore vote?  Let's say "Gore".

  Ballot 3:  Hole for Gore is partially punched out, chad hanging by two
    corners.  Hole for Bush cleanly punched out.  We just said the same
    sort of Gore punch was a Gore vote, so is this an overvote or a Bush
    Vote?

If you try to read the voter's mind by examining the ballot, you can probably
achieve a pretty accuracy rate, but certainly not perfect.  It's partly guess
work, which wouldn't be accurate even if the counter is perfectly accurate
and unbiased.

So the Bush campaign has a legitimate beef about the accuracy of manual
counts.  I'd guess manual counts are "more accurate" than machine ones,
but not all that much.

response 182 of 216:

Nov 13 17:43 UTC 2000

Since the machines are designed by humans, they have *some* built in
features to reduce counting errors, but *by definition* they cannot have
more than humans. Only fraud can reverse that. 

response 184 of 216:

Nov 13 18:07 UTC 2000

By the way, I hope both candidates are hard at work behind the scenes drafting
The Mother of All Concession Speechs.

Eventually, there is going to be a winner declared, and the other guy needs
to give a speech that will close the book on the debate over who is president
as much as possible.  It has to be done without whining and without
resentment, congratulating the other guy on a close race, and completely
backing his right to hold that office.

It's interesting that the first best chance for a candidate to really show
his fitness to be president by selflessly bringing the nation together in a
time of stress will go to the loser.

response 186 of 216:

Nov 13 18:35 UTC 2000

I'm sorry, but in designing the machines, engineers chose dimensions and
operating protocols so that, in their opinion (or by experimentation)
errors would be reduced. These are all "built in features to reduce
errors". A machine is NOT "just is" - it was designed. However all design
is compromise, so not all possible errors will be addressed in the design.
Hence, "they have *some* built in features to reduce counting errors, but
*by definition* they cannot have more than humans".