Re: A question about how votes are cast (protecting against multiplesheets of paper)

From: Keith Thompson <kst_at_mib_dot_org>
Date: Tue Jan 20 2004 - 09:46:56 CST

On Tue, Jan 20, 2004 at 08:51:24AM -0600, Douglas W. Jones wrote:
> On Jan 19, 2004, at 5:26 PM, Karl Auerbach wrote:
> >But I have a lingering question:
> >
> >As I understand it, the OVC ballot will be placed by the voter into a
> >folder so that only the bar code is showning. How does a pollworker
> >ensure that there are not other sheets in the folder without opening
> >the folder?
>
> There was an extensive discussion of this some months ago. We talked
> about watermarks, for example (ghosted in background images or marks).
> Among the suggestions was to use the government artwork (the official
> seal of the state), and the name of the county.
>
> The idea was to make it so that a ballot that was not official would
> jump out at someone who's leafing through a stack of them. So, for
> each precinct, the location of each of the background watermark design
> elements is randomized (say, in one of 4 locations, top or bottom left
> or right). The number of alternative locations required is small
> (for the example, with two design elements, we get 12 non-overlapping
> alternatives), since what matters is that a pre-printed ballot is
> highly unlikely to get these right. It is, however, easy to increase
> the number of alternatives while leaving it easy to visually detect
> substitutions. Randomly selecting between state seal and textual
> name of state doubles the number.
>
> All ballots printed in the same precinct for the same election would
> have these elements placed identically, in an arrangement that is
> selected at random from among the possible alternatives at some point
> in the setup process.

A cheater could go to the precinct in the morning to find out how the
ballots are printed, the go home and run off a stack of authentic-looking
fakes to be dropped off in the afternoon. Or, if there only a dozen or
so alternatives, he could pre-print as many as necessary. (Perhaps this
was already discussed; I haven't looked through the archives yet).

> You'd have to search the E-mail archive to get more detail. The term
> watermark definitely came up in the discussion and would be a good
> search keyword.
>
> We also talked about much larger cryptographically based validity
> coding on the ballot, I should mention.

That could raise the possibility of legitimate ballots being rejected due
to a glitch in the system. Ideally, a ballot is legitimate if and only
if it looks legitimate to the human eye; this is especially important
for manual recounts.

-- 
Keith Thompson (The_Other_Keith) kst@mib.org  <http://www.ghoti.net/~kst>
San Diego Supercomputer Center           <*>  <http://www.sdsc.edu/~kst>
Schroedinger does Shakespeare: "To be *and* not to be"
==================================================================
= The content of this message, with the exception of any external 
= quotations under fair use, are released to the Public Domain    
==================================================================
Received on Mon Jan 10 00:48:12 2005

This archive was generated by hypermail 2.1.8 : Mon Jan 10 2005 - 00:48:14 CST