[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: openvotingconsortium

Hey Pete,

Maybe you could try to sound a little less condescending? Hmmm?

If my message wasn't clear to you, I would be happy to provide a clarification--without the petty assertion that I don't "get" your side of the argument.

Imagine, if you will. A ballot that initially was marked with a vote for "candidate A."
That computer made mark is crossed out and a mark in the box for candidate B is hand written.

Simple, right? The person meant to vote for B.

OK, what if the box beside candidate B is partially erased with the word "stet" next to it? How do you grade that one?
Not clear is it. Is that vote ignored or does the judge take his best guess?

I'm not saying that the bar code should take precidence over the hand written code. But I do have a problem with your assertion that a bar code would completely muck up the works. In fact, my GUESS is that if there were a bar code, people would be more likely to ask the voting official for a new ballot if they mess up rather than trying to fix it by hand. If the bar code and the written response don't match, well, a voting official can make that judgement--but at least he/she has all the information.

You stand by your assertion that there should be one official vote count--the paper ballots, because, just like having two independant marks of the vote (bar code and plain text) having a paper and electronic copy allows the possibility of discrepancies between the two.

I ,on the other hand, would rather see those discrepancies up front. We know that there will be an error rate associated with both paper and and electronic ballots. Using BOTH methods simultaneously at least provides a way to discern when one of the methods has failed. If your counts match perfectly, you can be pretty confident that there was a 0% failure rate. If your paper count registers 100 fewer votes than your electronic count you know you have a problem. You can either look around for a missing ballot box or check your computer for signs of tampering. But in any case you KNOW you have a problem. Using any ONE method may not reveal a failure.

You argue that only one method should be used. But I'm guessing that you're pushing for at least a sample hand counting of any electronically recorded vote in Boulder. Well, you can't have it both ways. Either you think this redundancy is good or it is not. But you can't claim that redundancy is bad when an electronic count is challenging the status quo but good when a hand count is doing so.






Pete Klammer wrote:

You're being silly; is it deliberate?  "... inviting the possiblity of
discrepant count" ?  C'mon!

Or maybe you're finally getting it, the final stage of enlightenment: we are
not arguing for a "paper trail" to verify "electronic votes".  You obviously
see the problem: if the plain text or "receipt" says one thing, and the bar
code or "digital record" say another, the election directors, or the courts,
will have to rule, well, the computer recorded those bits, and humans are
fallible, so the bits must be right!"  Right?

The ballot as marked and placed in the ballot box is the only record that
counts!  If a voter alters his marks before he casts his ballot, then the
altered marks are counted.  Why should this be hard to understand?  Oh, I
get it, you're still assuming that the machine which printed the ballot
should have priority over the human who casts it.  Get over it.

There is no discrepancy between a marked ballot and itself; the possibility
of discrepancy arises when you set up two competing, ambiguously-vested
representations of one thing: for example, a plaintext list and a bar code,
or an electronic record and a receipt, or any other two or more records, one
voter-verifiable and the other not.

If the open-source PC system is a non-recording, non-counting ballot printer
or marker, then there is no need for a bar code, or any other "second
channel" of information.  Marks in boxes are unambiguous, countable, and
scannable.

And if your naughty voter "alters" his ballot after it's printed and before
he casts it, then maybe he's completing some undervoted races with
hand-marking, or maybe he's overvoting, or maybe he's erasing a printed mark
to replace it with a hand mark.  Maybe those actions ar inadvisable; maybe
the voter should request a replacement ballot and call this one spoiled;
maybe even the ballot should be presented behind glass, so the voter has
only two choices, to cast or to do over; but it's his ballot, and his only
ballot once it is cast.

It is only by creating additional, superfluous, buried, encoded, or secret
records or counts, that anyone is "inviting the possiblity of discrepant
count".

-----Original Message-----
From: Nicholas Bernstein [mailto:nicholas.bernstein@xxxxxxxxxxxx] Sent: Thursday, April 08, 2004 11:06 AM
To: pklammer@xxxxxxx
Cc: paul.tiger@xxxxxxxxxxxx; 'Cvv-Discuss@Coloradovoter. Net'
Subject: Re: openvotingconsortium

A bar code is also not hand-modifiable. This reduces the possiblity that a voter will try to change his/her ballot after it is printed.
And your argument about "worst case scenario" doesn't hold water, because it is the exact opposite argument as the one that is used against electronic voting in the first place. I.e. If you require a paper trail (and allow a sample hand count) you can check for mistakes by looking for anomolies. Why is inviting the possiblity of discrepant count a benefit in one instance but a hinderance in another?

On another note. Joe, could you explain exactly what you think is unconstitutional?

Nick

Pete Klammer wrote:



Stepping into a question, Pete - why not a bar code?




A bar code is not hand-countable, not hand-count auditable. Now we have to
"trust" yet another layer of electronics to translate what I thought I


voted


for into a tally and a count.  At the very least, this is another
opportunity for coding error.  Above that, it asks the voter to suspend
skepticism a notch, and trust the technologists.  Worst case scenario: text
says one thing, bar codes say another.

Occams' Rule: keep it simple, stupid!  If it can be done either with a bar
code or without a bar code, then get rid of the bar code!

Pete











--
If you're not outraged, you're not paying attention.