I think that we currently have two realistic options for vote counting:

    -- Optical scan with counting in the precinct
    -- Direct Record Electronic (DRE)

Each system has strengths and weaknesses.

Optical scan with precinct counting can provide feedback and a paper trail.  However,
the feedback is in public (not in the voting booth) so that privacy issues arise.
Furthermore, overvotes are still possible with optical scan.  Some voting administrators
also worry that optical scan will not work in large counties with long ballots.  Optical
scan systems, however, have been used for a few decades so that at least some of the
problems have been worked out.

DREs provide feedback in the voting booth which preserves privacy, and they make
overvoting impossible.  But DREs typically do not provide a paper trail, and DREs
(especially touch-screen DREs) are a relatively new technology which means that we still
have a lot to learn about them.

Both optical scan and DREs can be technologically complicated in some instances.  In
fact, it appears that the more complex DREs were the ones with problems in Florida while
the simpler ones had fewer problems.

Thus, no system is perfect or even best.  I think that each jurisdiction has to decide
what trade-offs it wants to make.  An important part of the calculation is making sure
that poll workers and other election administrators are well-trained.  There has been
too much emphasis on technology during this debate and too little emphasis on people.
(The Carter-Ford Commission, however, recognized the problem and the major reason why
they suggested a national voting holiday was to get better poll workers.)

Perhaps the best system would be an electronic one with a paper record.   What is
needed, however, is much more research and development on a better system.  I think that
the Cal-Tech/MIT people are still trying to do that, but it is hard to get support for
the type of research that needs to be done.

Henry Brady
Professor of Political Science and Public Policy
Director, Survey Research Center and UC DATA
University of California, Berkeley