RE: Wait just a friggin' minute!

["AlKolwicz" <AlKolwicz@xxxxxxxxx>]

==========================
ADK
Surely you are not saying that the security and reliability of DRE's are not
limited by physical constraints.  You seem to infer so.  Surely you are not
saying a DRE or a memory card is safe from theft or a firebombed voting
place.  You seem to infer so. 
==============================
NB
Yes, that is what I am saying. An electronic ballot does not exist in 
the physical world, so it doesn't have physical constraints. You can 
make as many duplicate ballots as you want and save them (physically) on 
myriad media (hard drive, memory card, remote drive, minidisc, etc) in 
multiple formats. You might be able to destroy one physical copy, but 
destroying 10 digital copies spread out physically and implemented in 
several ways and checksummed, is next to impossible.

========================
ADK
Please tell me more.  

How is it that a digital ballot is not represented on a physical medium?
And how is it that a physical medium is more impervious to destruction or
theft because it contains a digital ballot?

==================================


-----Original Message-----
From: Nicholas Bernstein [mailto:Nicholas.Bernstein@xxxxxxxxxxxx] 
Sent: Wednesday, February 11, 2004 7:39 PM
To: AlKolwicz
Cc: 'Evan Daniel Ravitz'; cvv-discuss@xxxxxxxxxxxxxxxxx
Subject: Re: Wait just a friggin' minute!


>
>Surely you are not saying that the security and reliability of DRE's are
not
>limited by physical constraints.  You seem to infer so.  Surely you are not
>saying a DRE or a memory card is safe from theft or a firebombed voting
>place.  You seem to infer so. 
>
Yes, that is what I am saying. An electronic ballot does not exist in 
the physical world, so it doesn't have physical constraints. You can 
make as many duplicate ballots as you want and save them (physically) on 
myriad media (hard drive, memory card, remote drive, minidisc, etc) in 
multiple formats. You might be able to destroy one physical copy, but 
destroying 10 digital copies spread out physically and implemented in 
several ways and checksummed, is next to impossible.

>
>Have you answered the basic question?  How does a voter VERIFY that the
>digital ballot has recorded exactly what the voter intended and only what
>the voters intended?
>
Yes I have. Check out my whitepaper at 
osl-www.colorado.edu/~bernsten/dualvoting.html

>We discuss the problem of verifiability in two parts -- voter recording and
>vote counting.  The ballot box is the link between the two.
>
There are more parts than that. Vote "counting" should be split up into 
many parts.
1. interpreting the ballot.
2. recording the intepreted ballot
3. summing the results locally
4. reporting the results to larger area (repeated up through district, 
city, county, state, and national levels)

A voter can only 'verify' voter recording with a paper ballot. In a DRE, 
steps 1 and 2 are eliminated because the votes are already in the form 
that you need them

>Only the voter can verify what is recorded -- after all we do believe in a
>secret ballot.  Only the voter can verify their votes and only the voter
can
>place their verified ballot into a secure ballot box.  Clearly, the
security
>system that prevents stuffing the ballot box is an exposure -- but is
>manageable.  The possibility of a voter voting YES, and getting a receipt
>that says YES and the DRE recording NO is real, and if we follow the press
>is happening.  A voter verified ballot does not say that it is counted
>correctly, only recorded correctly.
>
Read the white paper. I address this issue.

>It is definitely possible and relatively easy to verify that votes recorded
>on paper ballots are counted correctly -- as long as it is agreed what
votes
>are marked.  This is a function of process design and transparency -- as
you
>well know.  
>
This may be logically simple, but it doesn't mean it's trivial. You've 
got lots of votes to count. Keeping track of numbers is easy for 
computes, hard for humans.  How many tries does it take you to ballance 
your checkbook so that it agrees with what the bank says?

>Unfortunately, the votes on ballots are not always unambiguous.
>Where absentee ballots account for 20% of the vote, there is often
>considerable ambiguity.  The accuracy of the vote count is complicated by
>this ambiguity.  Recounts that do not take the interpretation of ambiguous
>ballots into account are flawed.  
>
Thanks for supporting my point. There is no abiguity in a DRE, it is 
digital. That's a good thing.