Problem

bmcculley at rcn.com bmcculley at rcn.com
Wed Apr 21 11:01:01 EDT 2004


>From: "Thomas M. Albright" <talbright at albrightent.com>  
>> Mark Komarinski wrote:
>> > I'm not sure if this is MA wide (I forget what other
towns did)
>> > but my town has scantron-ish forms that you may remember from
>> > the SAT.  Get the paper ballot, fill in the ovals with a
marker,
>> > run it through the scanner on the way out and the ballot goes
>> > into a vault.  No idea if the scanner makes sure the
ballot is
>> > valid, given that I fill it out properly each time.  I think.
>> 
>> That was kind of my point. How do you know those things are
not flawed 
>> in some way, or fixed? How do you know if it counted your vote 
>> correctly? Forget about new fangled computers, I'm talking
about 
>> exsisting systems. What if something happened and it
switches Dems for 
>> Reps or something like that.. Unless a recount is demanded
we would 
>> really never know.
>> 
>That's kinda the poit though: With a paper trail there *can
be* a 
>recount. With a computer-only system, all the data would be
corrupt. 
>Personally, I would not feel secure without that safety net.
>

Couple of relevent points.

1) All systems have some residual error rates, which are
somewhat difficult to measure in a secret ballot system, but
there are generally accepted values.  From what I've seen and
heard from credible sources, paper ballots are around
double-digit errors, optical mark/sense readers are around
4.5% error rate, DREs (direct recording electronic) are down
at 1% or less.  I forget the numbers for mechanical voting
machines but they were somewhere between optical mark and
paper, so perhaps 6 to 7 %.

2) Many of the errors are not reproducible.  I counted paper
ballots for the 2000 presidential election, and I know that
the guy sitting next to me could have counted the same stack
of ballots three times and gotten three answers none correct!
 Errors in mechanical and optical mark are also generated from
sources that may not be entirely deterministic (dust and dirt
anyone?).  What's a recount worth if it's unreliable?

3) As Tom says, "I think".  As others pointed out, there may
be unrecognized problems with existing machines.  Now that
there is a recognized problem that will also be reviewed, but
expect that it will primarily motivate replacement of older
machines with new ones - which is why focusing on the new
designs is important.  However, the installed base of recent
designs is becoming substantial, and it's not likely they will
all be replaced quickly (it's your taxpayer dollars that will
have to be spent if they are!  you want to take the lead in
arguing that the two-year-old expensive voting machines need
to be replaced with expensive new ones? ? )



More information about the gnhlug-discuss mailing list