Re: Have you ever been a poll worker?

From: Steve Chessin <steve_dot_chessin_at_sun_dot_com>
Date: Mon May 17 2004 - 15:30:16 CDT

>Date: Mon, 17 May 2004 02:41:04 -0400
>From: David Mertz <voting-project@gnosis.cx>
>Subject: Re: [voting-project] Have you ever been a poll worker?
>
>> [Ed system] Using Santa Clara County as an example, the third poll
>> worker wouldn't sit behind the table, but would instead escort each
>> voter to a machine and activate it. Could be workable. That poll
>> worker will get a lot of exercise. :-)
>
>I understand from Gayle (my mom) that normally poll workers are
>supposed to rotate roles during the day. At least in Colorado. Still
>some exercise, but each one only exercises for, e.g. 1/4 of the day.

Yes, I assume that the people rotate. But the roles are fixed, even if
the people fulfilling those roles change from time to time. And some
poll workers may not be very ambulatory and would rather sit all day.

>>> David: Each voter given unique PIN to enter on machine from randomized
>>> list. PIN contains code for party specification.
>> No extra poll workers needed, but does require voter to remember a
>> six-digit number, or else have the poll-worker write it down on a slip
>> of paper to hand to the voter (slowing down the process).
>
>Right. I was thinking about whether it might be easier to remember a
>shorter letter/number combination than a plain digit string. It
>depends whether voters are presented with a number keypad or an
>onscreen keyboard (as for write-ins). It might easier to remember
>'35Q2F' than to remember '354821', for example (both contain a similar
>amount of information in total, i.e. enough). That's an empirical
>question, but I think one that people have looked at in other contexts.

That's a minor optimization. (Probably a good one; may even be better
to alternate letters and numbers: 3Q5F2 rather than 35Q2F.)

>Voters *will* be frustrated by entry errors. But they would be allowed
>to try more than once, I presume (I'm not sure whether that means
>unlimited, or up to N). Maybe with the 10 second delay I suggested
>between attempts to limit attacks.

Give them three tries with no delays, then a 30-second lockout with a
screen that says "please ask a poll-worker for assistance". (Another
minor optimization.) The lockout will probably expire in the time it
takes for them to call over a poll worker and for the poll worker to
arrive. (If they can't do it in three times then either they are
innumerate or misremembered the PIN or have some other difficulty.)

>> Identical to current DRE procedures, including what was used in Santa
>> Clara County. (BTW, each polling place was given 25 smart cards, even
>> though there were only five machines, so that if a few went missing
>> voting could still take place.)
>
>Right. But can the cards that "go missing" be smuggled back in later
>to cast illegal votes? Maybe there's a way to guard against that, but
>it sounds like complicated procedure for poll workers to follow (and
>complicated programming to set it up).

Well, with the OVC technique, they'd be printing multiple ballots
that would go into the privacy folder and be handed to a poll worker,
who would probably notice that there was more than one ballot present.

With today's DREs, you're right that they could cast multiple votes,
but (as Avi Rubin noted), these machines make a very audible "click"
when the card is ejected, and a poll worker might notice that a voter
seemed to be inserting a card just after one was ejected.

Also, this is retail fraud, not wholesale, and requires confederates.
The person taking the card could not come back in and revote; they'd
have to take the card someplace, reactivate it, and give it to a
confederate who would then vote twice. (There's a lifetime to card
activation; the Sequoia Edge won't accept a card that was activated
more than a few minutes previously, or was activated in the future.)

You're never going to eliminate retail fraud entirely (for example,
corrupt poll workers casting ballots for no-show voters after the polls
have supposedly closed, or someone voting as a friend's absent neighbor
in a different precinct, and it is known that the neighbor didn't apply
for an absentee ballot). The goal is to make it no more possible than
with existing DRE and optical scan equipment.

--Steve
==================================================================
= The content of this message, with the exception of any external
= quotations under fair use, are released to the Public Domain
==================================================================
Received on Mon May 31 23:17:52 2004

This archive was generated by hypermail 2.1.8 : Mon May 31 2004 - 23:18:16 CDT