[House Hearing, 107 Congress]
[From the U.S. Government Publishing Office]
HEARING ON TECHNOLOGY AND THE
VOTING PROCESS
=======================================================================
HEARING
before the
COMMITTEE ON HOUSE ADMINISTRATION
HOUSE OF REPRESENTATIVES
ONE HUNDRED SEVENTH CONGRESS
FIRST SESSION
__________
HEARING HELD IN WASHINGTON, DC, MAY 24, 2001
__________
Printed for the Use of the Committee on House Administration
BOB W. NEY, Ohio, Chairman
VERNON J. EHLERS, Michigan STENY H. HOYER, Maryland, Ranking
JOHN L. MICA, Florida Minority Member
JOHN LINDER, Georgia CHAKA FATTAH, Pennsylvania
JOHN T. DOOLITTLE, California JIM DAVIS, Florida
THOMAS M. REYNOLDS, New York
------
Professional Staff
Paul Vinovich, Staff Director
Bill Cable, Minority Staff Director
TECHNOLOGY AND THE VOTING PROCESS
----------
THURSDAY, MAY 24, 2001
House of Representatives,
Committee on House Administration,
Washington, DC.
The committee met, pursuant to call, at 11:00 a.m., in Room
1310, Longworth House Office Building, Hon. Robert W. Ney
(chairman of the committee) presiding.
Present: Representatives Ney, Ehlers, Doolittle, Hoyer, and
Fattah.
Staff present: Jeff Janas, Professional Staff; Reynold
Schweickhardt, Technology Director; Roman Buhler, Counsel; Paul
Vinovich, Counsel; Chet Kalis, Professional Staff Member; Sara
Salupo, Staff Assistant; Robert Bean, Minority Staff Director;
Matt Pincus, Minority Professional Staff; and Keith Abouchar,
Minority Professional Staff.
The Chairman. The committee will come to order. I want to
thank the witnesses for coming to the committee. I would note
we have several things going on in a conference on the Majority
side, so the members will be coming in. Also the ranking member
has a previous commitment, but he will be here; and a couple of
the other members will also be coming in at different times. We
do have Mr. Fattah and myself, So we appreciate you coming.
Our witnesses are Christopher Baum, Vice President/Research
Director of the Gartner Group, Stamford, Connecticut; Thomas
Palfrey, Professor of Economics and Political Science, Co-
Director MIT Voting Technology Project, California Institute of
Technology, Pasadena, California; David Woods, Professor,
Institute for Ergonomics, Associate Director of the Midwest
Center for Inquiry on Patient Safety of the Veterans Health
Administration, from my alma mater, Ohio State University. I
attended there when we used to beat Michigan regularly, years
ago. And Ronald Rivest, Viterbi Professor of Computer Science,
Laboratory for Computer Science, Massachusetts Institute of
Technology, Cambridge, Massachusetts.
I want to welcome all the witnesses. As you are probably
aware, we have had different panels on election reform, and I
think they have been interesting panels as this has progressed.
This is the fourth hearing in a series of hearings. Your panel
brings a different, I think, perspective to the hearing
process. The first panel we had were secretaries of state,
State legislators, county commissioners, disabled community,
and American Legion.
Our second hearing had the local election officials, who
were on the front line of this, and then the third hearing
included the vendors. And now we have the technology
exposition. And now I think from the academic side and the
technological side, you will bring a different perspective for
us.
Mr. Fattah, do you have a statement?
Mr. Fattah. Thank you, Mr. Chairman. Let me thank you for
your continuing role in this to tackle this issue that is
critically important to the lifeblood of our democratic
process. I have an opening statement that I would like to
submit into the record on behalf of the ranking member, the
gentleman from Maryland, and I would like to do that if there
is no objection.
The Chairman. Without objection.
[The statement of Mr. Hoyer follows:]
[GRAPHIC] [TIFF OMITTED] 87478A.001
[GRAPHIC] [TIFF OMITTED] 87478A.002
Mr. Fattah. And I would like to thank our witnesses today
and I look forward to their testimony. I have a particular
interest in the Caltech-MIT project, but I am sure that each of
our witnesses will have testimony that will help us grapple
with this issue. Thank you, Mr. Chairman.
The Chairman. And thank you, Mr. Fattah.
And we will begin with the first witness, Christopher Baum.
STATEMENT OF CHRISTOPHER BAUM, VICE PRESIDENT/RESEARCH
DIRECTOR, GARTNER GROUP, STAMFORD, CONNECTICUT
Mr. Baum. Mr. Chairman, distinguished members of the
committee, Gartner is honored today to give testimony on the
trends in voting technology. The events of last November point
clearly to a crisis in confidence in the American electoral
system. However, there is no clear consensus on how to fix the
problem. Some believe that electronic voting is the only way to
go. After all, it is the same technology that worked so well in
the banking industry. However, there are other opinions. Last
week I was present at an election site in Boiling Point,
Pennsylvania, where voters were invited to try a new
computerized system in a ``shadow election.'' after they had
cast their official ballot on the regular system. I saw one
gentleman who declined to try the new system. I asked him why.
He said, if you aren't smart enough to vote on paper, you
shouldn't be allowed to vote.
However, Gartner believes that computer-based voting
technologies, with appropriate supporting processes and
infrastructure, can significantly reduce election problems and
restore the public trust. The public has already demonstrated
its faith in electronic systems using ATMs to make deposit and
filing taxes electronically. Computerized voting appliances can
significantly improve both the balloting and tallying
processes, increasing accuracy both in the votes cast and the
vote count.
Gartner recently completed a survey on the intentions of
the United States' largest voting districts in the aftermath of
the 2000 Presidential elections. The findings of the survey
have been submitted to this committee; but, in summary, we
would like those present to consider the following:
Forty percent of the districts surveyed have already either
implemented new technology or have begun major new voting
technology projects.
Those favoring Federal guidance on purchasing and
implementing new technologies are evenly matched with those
opposed to such guidance, about 45 percent on each side, with
10 percent abstaining.
One quarter of those opposing Federal guidance specifically
listed implementation problems with the Americans With
Disabilities Act, ADA, and the National Voter Registration Act,
``Motor Voter Act,'' as examples of why Federal intervention is
not warranted.
However, 50 percent of those in favor of Federal guidance
cited the need for uniformity in Federal elections. The most
cited example was guidance in establishing proper unit trails
that can support recounts, but do not impinge on privacy.
While desires for overall Federal involvement is mixed, 65
percent favored a specific Federal role in the valuation of new
voting technology.
And, not surprisingly, 75 percent reported that their
jurisdictions would require increased fundings to acquire new
technology.
As a result of this survey, Gartner believes that the local
voting jurisdictions are willing to purchase and implement new
voting technologies in time for the 2004 Presidential election,
if the Federal and State funds are made available by mid-2002.
Gartner predicts that if Federal and State funds are not
available by mid-2002, no more than 40 percent of the voting
precincts currently using antiquated systems will successfully
implement new voting technologies in time for the 2004
elections.
By comparing the survey results with buying trends, voting
system life cycles, voting appliance costs, and projected
funding, Gartner has created a projection of voting technology
for the next 3 Presidential elections. These projections have
been submitted to the committee.
In the short term, optical scan technology will experience
a quick growth, as it is the easiest alternative to implement,
and offers a single solution for both polling place and
absentee voting. Eventually, however, direct record electronic,
or DRE, and the closely-related i-voting technology also win
out.
There are a number of stumbling blocks in addition to
funding. First and foremost, is the vast number of different
procurement procedures used throughout the country. These
processes are at best cumbersome and lengthy. This is not just
another technological purchase. There are significant political
pressures at work as well. For example, Philadelphia recently
signed its contract for new voting technology in April, capping
off a process that began in 1995. Also daunting, is the fact
that no single standard exists to certify voting technologies.
There is an opportunity for the Federal Government to provide
guidance and promote innovation in this area.
While this hearing focuses on voting technology, it is
critical to realize that there is more to an election than
casting a ballot. Without rational improvements in voter
registration, authentication, Election Day planning and
operations, and the ballot tallying process, we run the risk of
fixing Election Day but not improving the voting process.
Thank you for your attention.
[The statement of Mr. Baum follows:]
[GRAPHIC] [TIFF OMITTED] 87478A.003
[GRAPHIC] [TIFF OMITTED] 87478A.004
[GRAPHIC] [TIFF OMITTED] 87478A.005
[GRAPHIC] [TIFF OMITTED] 87478A.006
The Chairman. Appreciate your testimony.
And next, Mr. Palfrey.
STATEMENT OF THOMAS R. PALFREY, PROFESSOR OF ECONOMICS AND
POLITICAL SCIENCE, CO-DIRECTOR/MIT VOTING TECHNOLOGY PROJECT,
CALIFORNIA INSTITUTE OF TECHNOLOGY, PASADENA, CALIFORNIA
Mr. Palfrey. Mr. Chairman, and members of the House
Administration Committee, thank you for letting me speak to you
today. I am Thomas Palfrey, professor of economics and
political science at Caltech, and codirector of the Caltech/MIT
voting technology project. My counterpart at MIT is Stephen
Ansolobehere who is a political science professor there.
A week after the 2000 Presidential election, David
Baltimore, the President of Caltech, called up Charles Vest,
the President of MIT, with an idea. The idea was that our two
institutions should collaborate to develop improved voting
technologies. The problems observed counting the vote in
Florida and elsewhere originated with technology.
Presidents Vest and Baltimore assembled a team of computer
scientists, mechanical engineers, process engineers, systems
specialists, and social scientists. The Carnegie Corporation
and our two institutes have funded our endeavors.
We are in the initial phase of our project, which is a
study phase and will culminate in a report this summer, which
will include a number of public policy recommendations. You
will hear some of those recommendations today.
First, what is voting technology? Voting technology is
usually and incorrectly equated with voting machines. In fact,
voting technology encompasses a wide range of issues related to
conducting an election, and the Caltech/MIT voting project has
been addressing all of these issues collectively. Voter
registration systems, ballot and voter-interface design,
security, absentee ballots, voter education, polling place
practices, and recounting procedures are some of the important
components of the overall technology of voting systems.
First, what are some of the problems that we have
identified? First of all, the high-risk rates of uncounted,
undermarked, and spoiled ballots, what our preliminary study
called residual votes. The average incidence of such votes is
on the order of 2 percent and has been there for about half a
century, as far as we can go back and measure. Counties using
punch cards average the highest residual vote rate of about 3
percent, but even if you exclude these, we are still over 1\1/
2\ percent in residual vote rate. That converts to 1\1/2\
million votes or more in the most recent Presidential election.
Our project identifies what we think is a realistic goal of
shooting at about a half a percent, because we must understand
that some of these uncounted ballots are intentionally
uncounted in the sense that the voters did not actually cast a
vote.
What are some of the other problems? There is a poor
monitoring in the system. There is very little systemwide data.
Something we noticed when we began our study is when we began
to look for numbers, they aren't all there. So, we basically
contacted thousands of counties directly to try to get
information about undervotes and overvotes, absentee ballots,
how much they spend on election polling place practices, and so
forth, and it is not collected in the systematic way.
This lack of monitoring makes it difficult to provide
feedback for industry and for voting administrators to improve
the system.
There is flawed and haphazard recount procedures. Some
States have nothing in place. Other States have rules that are
so vague that it is very difficult to implement in a systematic
and uniform way. We discovered that with Florida.
Lack of uniform standards for usability and accessibility.
Many of the systems that exist now do not conform to typical
standards that we have acquired under ADA. The user interfaces
on some of the newer machines--some of the older machines--make
it very difficult for the voter to figure out how to correctly
cast the vote. Examples like the butterfly ballot are the most
obvious ones, but one can find other examples in large numbers.
Finally, there are shortcomings in the voter registration
system. The system of cleaning or not cleaning, and purging or
not purging voter registration rolls, has created problems at
the polling places.
What sort of recommendations do we have in mind? Well, the
first and most obvious is probably to phase out at some pace--
we are not sure what the right pace is--what we call dominated
technologies. These are technologies that are underperforming
and do not have significant enough cost advantages to justify
them.
The second complementary recommendation is to phase in
undominated technologies. The current leading technologies
right now appear to be precinct-counted optical scanning
equipment and two forms of electronic voting. The first form
being what I would call integrated touch screen machines. That
is, a big box, so they are basically modern lever machines,
except they are electronic and have a touch screen interface,
and modular systems which have a touch screen, and then it is
connected to more off-the-shelf type of equipment. So the
architecture is actually different.
All three of these have advantages, however some of them
have disadvantages. The electronic technologies are really in
their infancy, so innovation is playing a big role in the
development of these. It is an ongoing process and it needs to
be encouraged.
Third, the transition from old machinery to newer machinery
has to be done at a measured pace. It is not something that can
be done overnight. We are talking about a relatively small
industry, that in a good year, might have $150 million to $200
million in revenue. We are talking about over 3,000 counties
who have strapped budgets. We are talking about complicated
ways of providing Federal funding through grants processes and
things like that.
This is not something that can be done overnight. It has to
be done in such a way that there is a lot of thought and study
that goes into it, and it should be done as an ongoing process.
Finally, along with Federal funding, we think it would be
advisable to have set up an independent Federal agency,
independent from the FEC, that oversees election administration
in the country, helps administer these grants, administers
research grants to research institutions in universities, and
organizes the data, serves as an information clearinghouse.
Thank you for the opportunity to speak today.
The Chairman. Thank you.
[The statement of Mr. Palfrey follows:]
[GRAPHIC] [TIFF OMITTED] 87478A.007
[GRAPHIC] [TIFF OMITTED] 87478A.008
[GRAPHIC] [TIFF OMITTED] 87478A.009
[GRAPHIC] [TIFF OMITTED] 87478A.010
[GRAPHIC] [TIFF OMITTED] 87478A.011
[GRAPHIC] [TIFF OMITTED] 87478A.012
[GRAPHIC] [TIFF OMITTED] 87478A.013
[GRAPHIC] [TIFF OMITTED] 87478A.014
The Chairman. Professor Woods.
Mr. Woods. Mr. Chairman, Congressman Fattah, and
distinguished members of the committee. November 2000 was a
vivid time in all of our lives, and in the heartland of America
as well as here within the Beltway. The intense debate
following the electoral surprise and crisis paralleled debates
I have participated in as a human factor psychologist and
researcher in industries such as nuclear power, aviation, and,
most recently, health care.
The first step in that debate always seems to be assigning
blame. Some people argued it was a voter error problem. Others
commented on the antiquated technology such as punch cards.
Many of the young people caught up in the debate were quite
intrigued by these devices, thinking them only relics of their
parents' ancient history.
My field of human factors studies the interaction of people
and devices, people and computers, and how these systems
sometimes fail to have both technological and human components.
What has our science learned from aviation and health care
work that we can apply to election technology? First, the
difficulties we witnessed are not simply voter error. Rather,
they are system issues in the interaction between people and
technology. And I want to point out that these interface issues
apply just as much to the election official interacting with
the equipment as well as to the voter interacting with the
equipment.
Second, the difficulties we witness cannot be solved simply
by replacing apparently antiquated equipment. Replacement
systems can exhibit poor user/device interface that results in
predictable risks of error.
Third, the good news. Many of these issues can be addressed
by basic bread-and-butter usability, engineering and testing
techniques, techniques that have been developed through our
work with the Department of Defense and aviation and aerospace
industries, and today have matured in the computer software
industry, and are readily available, quick, and economical to
apply.
Fourth, there are unique aspects to the voting context that
create potentially difficult design decisions and tradeoffs
that require careful consideration and longer-term investment
of our energy and innovation skills.
Let's go back through these four points a little bit more.
It is very important to recognize--and we constantly deal with
this problem after a crisis like the November election--we very
easily fall back into the blame game. Which is easier? A black
eye for human intelligence or a black eye for technology?
Instead, we have to look at the system and the interaction. The
failures are in the failure design for effective interaction
between people and technology.
Second, unfortunately, buying the vendor's latest model or
bringing in computer interfaces, in and of itself, will not
make issues and problems revealed by the Florida ballots and
electoral controversies go away. The kinds of problems that
were revealed can apply to the interaction of people and any
kind of technology.
I also commented that we have a mature research base that
is available, and a mature engineering base that is available
to apply. Techniques for usability-testing the prototype
designs have matured in the software industry and these can be
brought to bear very economically.
Examples of the kinds of principles and techniques we bring
to bear is the principle of good feedback. If you have to have
effective device design for interaction, for usability, give
people feedback so they can see the results of their actions,
recognize problems, and correct them. The same principle
applies to the interfaces with the election officials
tabulating the vote; provide a visible audit trail.
With computer technology you can design electronic voting
systems in many ways. You can even try to copy old paper or
lever technology over inside the computer system. The change to
computer technology brings potential benefits, but also new
pitfalls. And this imposes a responsibility on the designers to
think through how the functions you want to accomplish can
break down, and how trouble can arise. This requires usability
testing.
There is also the need in the long run for careful
consideration of the new issues that arise. Balancing security
and visibility feedback, providing wide access across a diverse
and aging population, handling large numbers of issues and
ballot choices in a timely fashion, supporting recovery for
mistakes, and doing it all in a low cost, are formidable design
challenges.
I want to point out that adopting new technology may reduce
our overall average in accuracy or imprecision rate, but create
the possibility of new forms of failure.
From past research we also find if vendors' claim for
failure-proof designs merit skepticism, as the humorist Douglas
Adam quipped, ``The major difference between a thing that might
go wrong and a thing that cannot possibly go wrong is when a
thing that cannot possibly go wrong in fact goes wrong'' it
usually turns out to be impossible to get at or repair.
So part of usability design is to take into account the
possibility for error and unanticipated situations.
Computerized voting and tabulation systems must support our
human ability to check and detect if new inaccuracies are
creeping in. It is easy to rationalize away the need for
action. Hyper-close elections are rare. My precinct didn't
really have a highly publicized problem. It was only the usual
error rate.
The Chicago Tribune, in a study, concluded that the error
rate in Cook County in the last Presidential election had
doubled to 6 percent. I and my colleagues in the human factors
profession are shocked that we seem so willing to tolerate even
that traditional 3 percent failure rate as a norm. And we ask,
where in business or transportation or medicine would we
tolerate such failure rates?
Voting is the centerpiece of democracy, and we need to
establish systems to monitor for the early warning signs that
inaccuracies or systematic errors are creeping into our voting
system.
In closing, I would like to remind you that technology
alone is not sufficient. Harnessing the system of people and
technology to fulfill the ideals of the democratic process
calls us all to make a commitment to excellence. Thank you.
The Chairman. Thank you.
[The statement of Mr. Woods follows:]
[GRAPHIC] [TIFF OMITTED] 87478A.015
[GRAPHIC] [TIFF OMITTED] 87478A.016
[GRAPHIC] [TIFF OMITTED] 87478A.017
[GRAPHIC] [TIFF OMITTED] 87478A.018
[GRAPHIC] [TIFF OMITTED] 87478A.019
[GRAPHIC] [TIFF OMITTED] 87478A.020
[GRAPHIC] [TIFF OMITTED] 87478A.021
[GRAPHIC] [TIFF OMITTED] 87478A.022
[GRAPHIC] [TIFF OMITTED] 87478A.023
The Chairman. Mr. Rivest.
STATEMENT OF RONALD L. RIVEST, VITERBI PROFESSOR OF COMPUTER
SCIENCE, LABORATORY FOR COMPUTER SCIENCE, MASSACHUSETTS
INSTITUTE OF TECHNOLOGY, CAMBRIDGE, MASSACHUSETTS
Mr. Rivest. Chairman Ney, Mr. Hoyer, and distinguished
members of the Committee on House Administration, I thank you
for the opportunity to testify to your committee on the issue
of security in voting technology. I have been involved in the
mathematical aspects of security for the last 25 years. I lead
the cryptography and information security within MIT's
laboratory for computer science. I am a founder of RSA Data
Security, a leading provider of security technology. Codes I
have developed are used daily to secure millions of on-line
Internet transactions.
For the past 5 years, I have investigated the security of
electronic voting. My students have implemented an electronic
voting system used for student elections at MIT. I am currently
participating in the Caltech MIT voting project just described
to you by Professor Palfrey. Our initial report will be out
this summer. The opinions expressed here are my own.
I find voting intriguing. It is not only important for
democratic society, but it is also technically challenging.
The challenge arises primarily from the need to remove
voters' identities from their cast ballots, in order to prevent
vote buying and the coercion of voters. This requirement for
anonymity makes electronic voting different than electronic
commerce or electronic banking, where well-labeled receipts and
well-labeled audit trails are standard. This requirement for
anonymity can also make fraud easier as the addition, deletion,
or modification of an anonymous ballot is harder to detect.
In 1869, inspired by the potential benefits of electricity,
Thomas Alva Edison was granted U.S. Patent 90646 for an
electric vote recorder. Congress declined to use it because it
reported votes too quickly. Today, inspired by the potential
benefits of computing and Internet technology, inventors and
election systems vendors are offering new technologies. We need
to carefully assess what these new technologies can offer, to
see if they really meet our needs and do so securely.
Given the short time available, I would like to offer some
personal opinions on the security of existing prospective
voting systems. I would be happy to expand further on any of
these points in response to your questions.
Number one, we are not ready for Internet voting from home.
I believe that voting equipment should be under the control of
election officials. At least a decade of further research and
development on the security of home computers is required
before Internet voting from home should be contemplated.
Number two, however, I believe that we should use the
Internet to post A lists of registered voters, B lists of
actual voters, and C lists of actual ballots cast. Not being
matched with the voters names, of course.
Number three, as far as getting the biggest bang for the
buck as far as security goes, I believe we should (A) improve
voter registration procedures and the computerization of voter
list registration lists; and (B) eliminate absentee balloting
except for cases of need. I am against voting by mail for
convenience. I prefer having a national voting holiday or
allowing voters to vote several weeks early at the town hall if
need be. Voters who vote absentee are simply not guaranteed the
same freedom from coercion and bribery that ordinary voters
have.
Number four, I believe voting systems should have a
physical audit trail. That audit trail should be directly
created by the voter, or at least directly verifiable by the
voter when he casts his vote. It need not be paper, but it
should be immutable and archival. Many have proposed electronic
systems fail this requirement. Electronic voting systems offer
improved ease of use and lots of flexibility, but they do not
intrinsically offer improved security. On the other hand, a
physical audit trail is not a security panacea, although it is
a big help.
Number five, we must ensure the highest degree of
confidence that our elections are free of manipulation and
fraud. The certification of voting systems should be an
important part of this process.
However, it is difficult to certify complex software-based
systems involving elaborate user interface and cryptographic
functionality. Experts in computer security and cryptography
need to be involved in the certification process. Requiring
that all security-critical portions of the source code be
``open source'' can greatly help to establish confidence in
such complex systems.
We are no more guaranteed protection against election fraud
by buying flashy electronic equipment, than we are guaranteed
protection against fire by buying a shiny new fire engine.
Security demands depends on the entire system, not just the
components, which need sound operational procedures managed by
training personnel. These operational procedures, which
themselves should be documented and certified, should primarily
ensure that no single person or vendor is ever in a position to
compromise the integrity of our democratic process.
Finally, I know that we are in the midst of a technological
revolution that provides both an enduring and improving set of
opportunities, and an increasing set of vulnerabilities. If
there is a chance to improve things now, then our focus should
not be on immediately spending money for new equipment but,
rather, on improving the higher-order processes of voting
system research, evolution certification, selection, financing,
staffing, and oversight as well as on improving voter
education.
I thank you for your attention.
The Chairman. Thank you very much.
[The statement of Mr. Rivest follows:]
[GRAPHIC] [TIFF OMITTED] 87478A.024
[GRAPHIC] [TIFF OMITTED] 87478A.025
[GRAPHIC] [TIFF OMITTED] 87478A.026
[GRAPHIC] [TIFF OMITTED] 87478A.027
The Chairman. I appreciate the panel's input. I am going to
go ahead and yield to Mr. Ehlers first.
Mr. Ehlers. Thank you, Mr. Chairman. I appreciate that,
since I have to leave shortly.
I really appreciate the testimony we have heard, and it
brings out a number of factors that have always struck me as
very important. My background is both having served in local
government where we deal directly with the election process,
and also with being a scientist by training and being appalled
at some of the things that I saw.
I think the very first step, of course, is good law. That
was a very major problem in Florida. The law was not well
written, well structured, it was not clear, and could not be
clearly interpreted. I think that problem is being dealt with
rather quickly by most legislatures throughout this land.
Secondly, we need good technology. I think we have done
reasonably well in that in terms of the technology itself,
ranging from the paper ballot up to the computers. What I find
missing in that, however, is the next factor we need, and that
is good human factors engineering. I think that has been a
major problem with the newer election devices that have been
developed and are on the market.
I also think something we don't pay enough attention to is
good security, good privacy, and good integrity. We pay a huge
amount of attention to that in our computer systems in
commerce. Even in the House, when we put that system in, that
is one thing that I insisted on very strongly. And I am very
proud, as is Ronald Schweickhardt who sits behind me, a staff
member involved in this, and a number of other members.
It is interesting to read in the paper about all the
hackers breaking into the Pentagon, the Senate, the White
House, but you have never read about them hacking into the
House. I don't want to say that too loudly because then we will
become the prime target.
I am very pleased. We put a lot of effort on security, and
I believe we have succeeded, but you have to have precisely the
same security in the voting booth that we installed here. And I
don't see that at all in the electronic systems, and even many
of the other systems are not as secure.
Just a quick comment on what was said earlier, I think it
is important for everyone to vote at once, to the greatest
extent possible. I am prejudiced on that because I run for
office and my campaign is planned so that all the information
is out there by Election Day. People vote 3 weeks ahead of
time, they miss most of the information that I am providing.
But also, absentee ballots lend themselves tremendously to
abuse and fraud. The one thing we haven't mentioned here, I
think the greatest opportunity for fraud, is in voter
registration; and we need to pay much more attention to voter
fraud there and ensuring that voting lists are good, that we
purge them regularly; that when someone moves, they can't keep
registration at their former address and so forth.
Now, given these facts and some suggestion of research
grants, what would you recommend the Federal Government spend
its money on for research grants if we should decide to go in
that direction? Where would you concentrate the efforts in
trying to get at these various problems I mentioned?
Mr. Woods. One of the resources that is available we have
already, through work with aerospace, the Department of
Defense, the government has already built up expertise on these
human factor engineering issues. For example, in the national
laboratories, they have expert groups who provide advice on
these interface design issues to many government agencies. We
could bring those in to, in the short run, provide a great deal
of guidance about how to evaluate potential new systems and how
to implement them and train the kinds of election officials in
operating these to achieve greater levels of success and avoid
some of these difficulties that could be done very quickly.
Some of these organizations could provide input to election
officials in the form of a guidance document, probably on the
order of months, and on the order of tens of thousands of
dollars in investment for a short-run benefit.
Mr. Ehlers. But we have talked in here about educating the
voters, training the poll workers and so forth. I really would
like to see a system where that is not a factor. In other
words, the human factors are so good that you don't need to
educate voters.
For example, in voting I think you should show the final
slate and have a Regis Philbin question at the end: Is this
your final vote? And, if not, you can go back and change it.
Poll workers, too, they are wonderful people, just the salt of
the Earth, really trying to serve their country in a very
difficult job, which they only do a couple of times a year for
long hours and hardly any pay, but they only do it a couple of
times a year. You can't train them every time. And again, I
think we need a system that doesn't require much training for
poll workers.
Let me ask a follow up question to that, and the others can
comment on any of these issues. What is the Federal
Government's appropriate role in this? We don't run the
elections. We have always trusted local governments and States
to handle that. How would we deal with the human factors issue?
Are we going to recommend certain systems? Are we going to set
Federal standards that systems have to meet in order to be used
to elect Federal candidates? How do you see us playing a role
in that?
Mr. Palfrey. I think the Federal Government could certainly
play a role in setting standards for certification of
processes. I don't think the Federal Government wants to insist
on everything being exactly one way. I think, for example, the
current testing procedures that are done to certify equipment
is machine testing. It is basically machines testing machines.
I think human testing is needed for the ballot interfaces and
for the various designs that are proposed. Currently there is
no human testing. I think that is one thing that a Federal
agency could do is to oversee the testing of these machines and
the development of appropriate standards.
Mr. Woods. There are many kinds of guidance documents
available to organizations that design computer devices for
human use. For example, several organizations have just put out
new guidance documents on access to computer and electronic
systems for the disabled and the visually handicapped. These
kinds of documents do not tie the hands of designers but,
rather, try to be a positive resource, to say there are some of
the effective techniques and ways that you can design your
electronic interfaces, as in this case, to make access for the
disabled more successful.
Mr. Ehlers. How do we deal with the security issue? You
know, if we have the modern equipment of Tamany Hall, any
college freshman can make some changes to the software on the
computers. How do we really ensure the security of the hardware
or the software?
Mr. Rivest. That is a very complex question. I think
continued monitoring of what is going on, making sure there is
separation of function, taking apart voting equipment after it
has been used to see if it has been tampered with, making sure
the code is ``open source,'' and looked at by lots of people.
It is a multifaceted problem. Making sure the poll workers know
what is appropriate to do and what is not, making sure the
equipment doesn't support modes of operation that would allow a
poll worker to reset the clock or whatever you can do through
various kinds of tampering. We are dealing with computer
systems now that are very much--voting systems that are
computer systems and have all the complexity and security
problems of computer systems. We need to keep the system simple
as possible to minimize the complexity. And security often
arises from simplicity. So, looking for simplicity in design is
an important criterion here, too.
Mr. Ehlers. Do you think it would be reasonable after each
election that all the systems are tested, that you run a quick
program through them to make sure there----
Mr. Rivest. It is reasonable to do some sampling, random
sampling, look at some of them to see if there has been any
tampering.
Mr. Baum. Also, sir, you could apply statistics to the
districts and find out where there are anomalies and test that
equipment stronger, and you can harden the equipment that is
being used. You can also move to internal memories on the
computer systems that are unalterable so they can't be changed
on Election Day. There are a number of technologies that you
can apply here to this equipment to make them almost military
grade in terms of hardness.
Mr. Ehlers. What I hear coming through, though, is that you
see the Federal role as one of doing the research, setting the
standards, and helping the States and localities to meet the
standards.
Mr. Baum. That is certainly what our survey shows would be
acceptable by State and local governments.
Mr. Ehlers. Thank you very much. Thank you, Mr. Chairman.
The Chairman. Thank you, Mr. Ehlers. Mr. Hoyer.
Mr. Hoyer. Thank you, Mr. Chairman. Gentlemen, I apologize
for my lateness. I had another meeting at 11 o'clock, and took
that as quickly as I could and got here.
Thank you, Mr. Chairman, for including my statement in the
record. I won't repeat it.
Mr. Baum in the end of your prepared statement, you
mention, and I quote, ``the daunting fact that no single
standard exists to certify voting technology.'' This is an
opportunity for the Federal Government to provide guidance and
promote innovation.
You have just been discussing that, obviously, in response
to Mr. Ehlers' question. Can you describe specifically how the
Gartner Group found--what it found when it looked at the voting
machine certification process.
Mr. Baum. Yes, sir. What we found was disarray. There is,
of course, the NASED who will approve voting machines, and
anyone can adopt NASED-approved machines. But that process,
while well documented, is still open to a certain amount of
interpretation.
For the States and local governments that do not accept
NASED-approved machines, there quite often is no set standard.
It is conceivable and I have seen instances where new
technologies are being introduced, for the companies to call on
local sales or local election officials. The response is, come
back when you have been certified by someone, not necessarily
who, but just someone.
That is where the difficulty comes in. There are over 3,000
touch points for election systems. One of the things that is
very interesting, is we don't even have consensus on the number
of counties there are in the United States currently. We went
out and called the secretaries of state of each State and got
the number of counties that they thought they had, and added
them up and got 3,066. According to the Federal U.S. Census,
there were 3,042. So there are some fundamental issues here
that we really need to address. And the lack of standards, you
simply cannot survive in a market where you have got a $20
million opportunity or $200 million opportunity, and you have
to deal with 3,000 separate standards for equipment.
Mr. Hoyer. Could I say this as an aside, not necessarily as
a question? In the State of Maryland, which is a relatively
simply organized State in that we have 23 counties and
Baltimore City, could the discrepancy be that for many purposes
Baltimore City is considered to be a county, NACo however would
presume it to be a city?
Mr. Baum. Right.
Mr. Hoyer. So that might--I don't know how many other
States have that same phenomenon--but that might count for that
20-plus discrepancy between the two.
Clearly, it would be useful from your standpoint, to have
an agency at some level that looked to assist not in a
mandatory way, but in an advisory way, to solve this chaos or
disarray--I think you used the world ``disarray,'' chaos may be
a harsher word--but disarray that you confronted?
Mr. Baum. Yes, sir. We do not necessarily believe it needs
to be a separate organization than the FEC, but we do think
that there needs to be a place where officials can go and where
vendors can do to say these are the requirements that we
recommend, these are the procedures that can bring you to those
requirements, and these are the sources of information.
Mr. Hoyer. I would say, as an aside, to all of you as you
do your work, it is my view that the time has come when the
election administration responsibilities at the Federal level
and the monitoring responsibilities of finances in elections
need to be separate, not because they are inherently
contradictory to one another as the OEA currently exists in the
FEC, but because there is so much immediate demand on the
financial oversight, that electoral reform at the Federal
level, as it has at the State and local levels, has taken a
second or third share.
Professor Rivest--how do you pronounce that?
Mr. Rivest. It is pronounced different ways within my own
family.
Mr. Hoyer. Well, I would like to pronounce it the way you
pronounce it.
Mr. Rivest. Rivest.
Mr. Hoyer. You offer some very sobering insights regarding
the fraud security of new voting technologies, specifically
Internet voting. Let me say I also agree with the chairman and
with yourself very strongly that voting is a communitarian
process--George Will wrote a column on this some months ago--
and that coming together in and of itself has a value, I think,
above and beyond the process value.
You express confidence in using the Internet for improving
voter registration, but please elaborate on how the Internet
can improve registration. Obviously, as a number of you have
referenced in your comments, that is a very key issue with
which we need to deal if we are going to have voter confidence;
that they come to the polls and they will be allowed to vote.
That, coupled with a very good provisional voting process. But
would you comment on that?
Mr. Rivest. I would be very happy to comment on that, Mr.
Hoyer. I think at the highest level, the voting system will be
improved by a greater degree of transparency, having more
information available to more people, more eyes looking at the
process. Applying that to the voting registration process, if
the voting registration lists are posted on the Internet in a
way that anybody can look at, you will see fewer dead people on
the rolls, you will see fewer people that are still on the
rolls that have moved. You know, somebody will call up their
official and say, did you know that so and so has left town?
Things like this can happen.
So just the process of keeping the accuracy of the voter
registration lists will be improved by having it open and
public, as I believe it should be.
I think the process of registering to vote per se, is not
something we should be attempting to do over the Internet. I
think it should require an in-person visit. But once you are on
the registration rolls, having your name listed as you are
registered to vote in this county, I think that would be a
help.
Mr. Hoyer. Do you have any thought as to how long it would
take to construct a statewide system in a State like
California, which is--well, there are no States like
California. That was a stupid thing to say. There are only a
few Nations like California. But a State--let's say a smaller
State, Florida, has just gone to mandating a system of central
registration. I don't know what cost that they attributed to
that. Maryland is going to a central registration system. I
think that is going to be critical, because in an era of very
mobile individuals and families, when you move from precinct to
precinct, house to house, and you really don't--there is
nothing that tells you this is a precinct line--that we need to
have that kind of system.
What is technically--how long would it take to create such
a system in a State like Maryland or Florida or Pennsylvania,
and what would its cost presumably be? Anybody have a guess on
that? Have you looked at it?
Mr. Rivest. Let me just respond. Part of it is different
States do it in different ways. I think it is Michigan that is
combining it with the DMV database. If you have a single voter
registration database and that is its only function, you might
be able to do it in 2 or 3 years. And I don't know what the
cost would be per voter. But if you are trying to organize your
State's citizen records along more lines than one, so you are
doing the DMV and the Social Security or whatever, everything
else altogether with within the State, I think that could be 6
or 7 years and lots more money.
Mr. Hoyer. Professor Rivest, one of the things, though, if
you try to combine it, your transparency issue becomes more
complicated, it seems to me, because there is information that
voters clearly do not want and should not want transparent to
the rest of us.
Mr. Rivest. Yes.
Mr. Hoyer. So that from my perspective, you are really
going to have very limited use of a document that is totally
transparent, and I agree with you it ought to be totally
transparent, because I think that will protect against fraud
and mistakes. I think my own view is that there are far more
mistakes than there are actual intentional fraudulent acts.
Mr. Baum. Sir, the National Voter Registration Act of 1993
pretty much requires a strong link with the Department of Motor
Vehicles in each of the States, so that has to be investigated.
Even so, there are commercial products available already. At
its most simple level, a statewide voter registration system is
simply a database.
Now, there are security issues that surround it, but they
are all known and solvable. And when you are taking a look at a
State with the population of New Mexico, or something like
that, it is not even a large database, so these are--again, it
is more of a political consideration than a technical
consideration.
Mr. Hoyer. Well, I am hoping that one of the things we do
in legislation, that we are considering, is to have dollars
available to assist States in the creation of a central
database. In my view, in large part, the States will have to
decide the individual problems and how they create that; that
we can assist them in doing that to provide greater voter
confidence when they go to the precinct, somebody will know in
this computer age, when everybody gets on the Internet, can get
access to gargantuan amounts of information, that they can get
information that Joe Dokes has in fact registered to vote in
this State.
So, Mr. Chairman, I have got other questions but I know
that time keeps running. Do you have questions?
The Chairman. I have got questions.
Mr. Hoyer. Why don't I yield to you?
The Chairman. And if you want more time. I want to ask a
specific question, Mr. Palfrey. Are you aware of any research
that has studied the voting patterns of persons who are
dyslexic or, let's say, illiterate?
Mr. Palfrey. With respect to the voting technology?
The Chairman. And how to facilitate and help those
individuals. Or is there technology that--for example,
technology comes forth--is it helpful, not helpful?
Mr. Palfrey. There is some informal research. I don't think
there has been very much that has systematically been done to
identify specific problems with technologies. Certainly, with
respect to user interfaces, maybe Professor Woods can say more
about this as far as dyslexic interfaces, but I think as far as
handicapped considerations, use of audio, use of Braille and
also overlays for the optical scanning equipment.
The Chairman. We know, if it is an issue of sight, I saw
the equipment here. So we know you have the earpiece. If it is
an issue of height, you can accommodate downwards. But if it is
an issue of dyslexia or an issue of illiteracy--
Now, an easy answer is, we take somebody in the voting
booth, a poll worker. For a lot of people, though, it is such a
private item about voting. So I just wondered--again, for
sight, there are certain devices, but what about dyslexia?
Mr. Palfrey. Well, there are examples----
Mr. Hoyer. Can I make a comment, because I would like to
hear--everything you mentioned, the hearing--that the audio
transmission may not only solve the sight problem but may solve
the inability--the dyslexic problem or another reading problem
that might occur. That might solve, most of us wouldn't think
of it in those terms, but it may do that as well. You might
want to comment on it.
The Chairman. And touchtone screen. It was an interesting
issue that we were discussing today.
Mr. Palfrey. Right. And that is a possibility in some
foreign countries, where illiteracy is a more serious problem
than it is here. In South America, for example, they actually
have pictures of the candidates, or icons to represent
different parties, for example, to assist voters who can't
read.
The Chairman. Also, Professor Palfrey, in your study over a
20-year period, 60 percent of the counties have adopted new
technologies of one sort or the other. Forty percent of the
counties have been using the same technology basically, I would
assume, for 20 years. Are there any similarities or common
characteristics about counties that have been using the same
technologies for a long period of time versus the ones that
have switched to some new technology? Are there any
outstanding----
Mr. Palfrey. Yes, there are two factors that seem to be
important with that. The larger counties, populationwise, we
have tended to switch to new technologies, largely to handle
problems of bandwidth and just dealing with large numbers of
voters. So, they are concerned about speed and cost. That
actually was one of the reasons for the transition even earlier
to punch card systems, but certainly transitions from paper
ballots to optical scanning equipment was done for cost and
speed considerations in larger counties. There is also a
correlation, with sort of average income levels or the sort of
county revenues controlling for size, that counties that don't
have as much to spend, available to spend, have not been
changing as fast.
The Chairman. Mr. Baum, I wanted to ask you a question. You
had stressed, and I think made a couple of good points about,
the fact of the different standard where somebody turns to an
election official and they say, ``as long as you are
certified.''
Mr. Baum. Right.
The Chairman. Let me just throw something out here, because
you mentioned the word ``agency'' a couple of times, as other
people indicated. Let's assume you are not going to create an
agency per se, as we would think an agency to be, because the
downside of that, frankly, in the election process, is that it
is not an environmental issue or a highway where you have a
daily ongoing situation. So therefore, the thought of agency
would tend to scare people across the board here, I think on
both sides of the aisle, and a majority of both, only because a
downside could be that there would be constant rules made by a
rulemaking body that would make people eventually shiver about
the agency saying whether you have an ID at the poll or not,
versus, you know, the local governments and the Congress.
So let's assume we don't have a full-blown agency. But what
would you--or could you speculate, on what you would want to
see if we created some type of body that was an advisory type
of body; who would be on that body, in your opinion?
Mr. Baum. Well, first I would like to respectfully suggest
that it is certainly more than a one-day project or a single-
day project. The example I like to give most is that of a
wedding. When you start planning a wedding, the events of the
day actually become very much smaller in proportion to what it
took you to get there. And running an election from an election
board's point of view is very much like planning a wedding year
after year after year.
The Chairman. I should clarify. I know it is year-round. I
am just saying, if you had an agency, though, and you set up a
full-blown agency, if it does have a daily function, I think we
would be scared of the product after about a year. If you have
an agency that creates a new staff daily, thinks of new rules
to create on local people, I should tell you where I am coming
from on that.
Mr. Baum. So in that setting, then, I think that what you
need is representative--I think what we are talking about here
is really a governance board, not a new kind of committee, but
people that are involved with the process. And that means that
you have people from State and local and Federal elections; you
have people that have other concerns, interest groups, people
who come in that are proponents for those who have
disabilities; making sure that people that come in, from the
Majority and Minority parties, to ensure that ballots are fair
in the way that they are set up, and people with other
interests coming in--I was going to use the term
``disinterested,'' but we are all voters, we should all be
interested. Like my fellow panel members here; people that can
come in and come together and build a consensus on these,
including the vendors themselves.
The Chairman. Thank you.
I had a question of Professor Rivest. For the machines you
are familiar with--because Mr. Swigert tells me he wouldn't let
you bring a laptop in here. That is how good you are. So those
voting systems which you are familiar with, how long would it
take to break into them while in a polling place? That could
change the results of an election, if that could be done? Is it
a long time process?
Mr. Rivest. I think it depends on who is doing it. I think
if you have a voter who is just walking in to try to, you know,
monkey with the equipment he is probably unlikely to have much
effect. A poll worker, similarly, has little background with
the equipment and wouldn't know quite what to do except to
operate it. A vendor's technician on-site might have intimate
knowledge of how the machine works and might have a
preprogrammed smart card to change a tally or something like
that.
So I think it depends on knowledge of the equipment. It
could be very quick if someone knowledgeable--changing
electronic systems, you know, can be done quickly. If there is
a card that is plugged in, as many machines have, that controls
the operation of the machine. If the card can be changed you
are changing the programming of the machine. If there is no
audit trail, if that card was removed and replaced, you may
have a successful hack.
The Chairman. Some observers have recommended an open
source process, so software can be used to control voting
systems that would be subject to impartial inspection that
could lead to increased quality. What do you think about that
idea?
Mr. Rivest. I am in favor of it, basically. I think that
this poll process is too important to be left wrapped in the
cloaks of the vendors' claims of proprietary need. I think that
having the software available for inspection by whomever will
increase our confidence that the voting is being tabulated
correctly.
It is not sufficient by itself, of course. You need to
assure that the software that you have looked at is actually
the software that is running in the machine and have procedures
installed to ensure things like that. But having the software
out for inspection I think will be to everyone's benefit.
Mr. Chairman. The last question I have is for Professor
Woods.
We had the voting machines in here, and I found them of
interest because that is the first time in my life I have not
used a punch card. I went originally to the X's we used to do
when I first voted down in Ohio in Belmont County. Then we went
to the punch cards, because that is what we used. So this is
the first time I ever actually touched some other type of
device.
The one thing I would ask of the vendors is, don't help me,
let me see if I can do that. I don't claim to have the ability
of Mr. Ehlers on scientific knowledge at all our computers, but
it was interesting just using the machines.
Now, I noticed an exhibit was there when, once you wanted
to make your choice, you had to retouch to cancel out and then
you touch back. But then there was another machine that you
pushed the next name, it automatically canceled back. Because
of what you do, dealing with human interface and ergonomics, is
there any machine, when you talk about the touch screens or
maybe a non-touch screen, that is a better type of machine or
the machine of the day?
Mr. Woods. It is difficult for us to recommend a particular
technology or system. But the example you use is I think a very
illustrative example where some of the principles we found
apply. One is feedback. So either of those systems can confuse
people depending on the degree to which they get clear feedback
with respect to their actions.
So, for example, take the second case you have mentioned
before, where you simply push the second candidate. Someone
could get confused about what the impact of that is. Are they
going to end up the equivalent on the punch card of having two
holes punched? If you have clear feedback--for example, when
you press the second one, the first one starts to blink in a
salient way and then goes dark--those kinds of systems are
likely to work better.
The second aspect of my response, this is the sign of
usability testing, is about that we have procedures for doing
tests so we don't rely just on opinion or quick walk-up, a
couple users trying it out. We have processes to do that that
are very economical to generate the best kind of information to
act on and get the best leverage when you have to make design
decisions or purchasing choices. So by using that technology of
user testing, technology that is used every day in the software
industry for the products you and your family buy on the local
scene, those kinds of tests can result in many of these kinds
of questions and result in more easy to understand and use,
especially for the diverse population that you pointed out is
so critical to the voting application.
The Chairman. One final thought--I will see if Mr. Hoyer
has some questions--basically indicated here and at least one
of the testimonies that, you know, it is the human error. You
know, we know that is part of it. But if it is just a human
error factor, should we just keep punch cards and spend a lot
of money? Well, if it is only human error, could we keep those
punch cards and spend a lot of money to educate people, or
should we go on to something else?
Mr. Palfrey. With respect to punch cards, I think we should
go on to something else.
The Chairman. Your testimony did indicate, though, that it
is human error. Well, if it is, do you just work with what you
have?
Mr. Palfrey. One way I like to think about it is in terms
of a technology being dominated. In other words, if you think
about spending the effort to educate the voters, if you want to
spend the same amount of effort educating voters using punch
card systems or you are educating voters using an optical
scanning system, what would the results be? And I think the
results of one would be better than the other.
The Chairman. Is there anyone that would like to keep punch
cards?
Mr. Baum. I think punch card manufacturers would like to
keep punch cards.
The Chairman. We know that.
Mr. Woods. The success we have achieved in cockpits and
other high criticality domains has come from dropping this red
herring of human error. When we want to label a problem as
human error, the scientific approach says you take that as a
symptom of an underlying problem in the system and the
interaction between people and technology. And there lies the
grounds for improvement so that we can achieve what we all
desire, in this case in the democratic process.
So, now, with respect to punch cards in particular, it
violates a very old rule for designing interfaces between
devices and people, all right? The punch card, you actually
make a hole, all right? The absence of something, the hole is
supposed to indicate the presence, the state that we are
interested in, the vote.
If I had designed--when I started our designing nuclear
power plant control rooms 20 years ago, if I had tried to code
the critical state stage of the nuclear reactor using that
principle that is done in punch cards, I would not be here
today. We have learned through bitter experience in aviation
interfaces and energy systems that is not a good way to set up
the interface.
The Chairman. You are making a good point. So I totally
grasp it, it is not a good way to set it up to read that card,
or it is not a good way to set up to have the human being, the
voter, reacting that way of the punch?
Mr. Woods. Both.
The Chairman. I understand, on the reading side of it, I
believe, of the machine reading it--it is probably not the
word--but why the voter?
Mr. Woods. We say that in all these issues of did they
punch through, was the card aligned, the basic rule of thumb is
you want people to make a positive indication so that you have
transparency and traceability, as my colleague to my right
indicated earlier.
The Chairman. Do you have any questions? That is
interesting.
Mr. Hoyer. Well I think that--we pretty much have a
consensus I think in the country at State, local and Federal
levels that we want to get--rid ourselves of the punch card,
save perhaps the manufacturers, as Mr. Baum put it. But it is
interesting that systemically there is a reason for that as
well. I was glad to hear that.
Let me ask Dr. Rivest, then I am going to go to some other
quick questions. You have reservations about the Internet
voting, and that is shared I think by a vast number of people,
including myself. However, I would like your comments on
whether it is worth testing in an area that we think does not
work as well, and we need to make it work well and may be an
area where we want to at least test this, and that is overseas
voting and military voting. Could you comment on applications
that may be perhaps more appropriate in that limited setting?
Mr. Rivest. I would be happy to.
I think that the military application is a good one to look
at. I think that you have the kinds of controls both over the
environment and the networks that you may not have in other
kinds of remote Internet voting, and you also have a very clear
and important need for our soldiers to be able to vote, and
also I am in favor of experimentation.
I think we need to plan to evolve our voting systems over
the next 20, 30--you know, as technology improves, we are going
to keep changing our minds as to what, as Professor Palfrey
said, what the dominating technology is. But in order to learn
what the dominating technology is, what the best voting systems
are, we need to have experience with them.
So, for all of those reasons, I think that experimenting on
a small basis at first, maybe expanding if the system seems to
work well, with remote Internet voting for military purposes, I
would be in favor of that. But it needs to be carefully
controlled and looked at. The Internet is a very fragile and
vulnerable entity, and it is vulnerable to attack by malicious
organizations from outside our country. So we need to make sure
that there is back-up systems in place should our soldiers not
be able to use the Internet.
Mr. Hoyer. Thank you.
Professor Palfrey, I was pleased to see in your statement
that you support the five principles that Chairman Ney and I
are essentially working around; and I take it from your
statement that our focus should not be exclusively on the
72,000 precincts, or whatever number precinct we have, the
72,000 precincts, but on a broader array of the technology use
in other technologies, used in other precincts, is that
correct?
Mr. Palfrey. That is correct, yes.
Mr. Hoyer. How expensive do you think we ought to be on
this in terms of assistance to improve an election
infrastructure?
Mr. Palfrey. Well, I think as far as coming up with a
single number I am not sure that is quite the way that I think
about it, partly because I view this as sort of an ongoing
process. As far as approximately, I don't know if you are
looking for approximate numbers.
Mr. Hoyer. No, I was not looking for a number. I apologize
if I implied that, obviously.
But the bill that I have introduced, along with a lot of
other people, sought first, as Florida has done, to eliminate
the punch card system. And then, as you know, it provides
ongoing resources for technology replacement as well as a lot
of other things in terms of education and technology
development.
What I am wondering is, if we limited our first step to
that, would you think that is an error, or do you think we
ought to make sure that this can be applied to other
technologies, i.e., lever machines initially? Obviously, we are
talking about limited resources that we will come up with in
the first traunch, if you will. That is what I am really
getting at.
Mr. Palfrey. I think there are other technologies that have
known problems besides punch card systems. Without identifying
specific vendors, there are examples of full face DRE machines
that have not been successful, that the administrators that use
them would probably be happy to replace. I think centrally
counted precinct scan equipment should be replaced by precinct
counting. If you are looking for a quick fix of that, that is
certainly a better technology having to do with the voter
feedback and also the ability for errors to be detected before
the vote is tallied.
Mr. Hoyer. so in other words--and I hadn't thought about
that--but clearly everybody believes that in order for a system
to be as accurate as it can be you need precinct where-a-voter-
is-still-present technology to let the voter know whether or
not they have made mistakes.
Mr. Palfrey. Yes. I think we are at a stage now where there
exists technologies to do that. We should take advantage of it.
Mr. Hoyer. Okay. That is helpful.
Dr. Woods, you addressed three key factors of voting--user
friendliness, voter feedback and balance the design. What is
the best feedback tool to let voters verify whether they
actually voted for the candidates that they want without
compromising voter privacy?
Mr. Woods. Well, you have hit the heart of the design
problem from a human factor's perspective.
Normally, as Professor Rivest indicated, we don't have
those simultaneous constraints to deal with. Obviously, the
feedback has to be something that can be terminated by the
voter so that they get a display, they get feedback that says
this is what the machine thinks you told it. Then, in
registering their vote, the voter must have a confident feeling
that that information has gone into the computer and the
traceability back to them has gone away; and that requires
careful design and testing with users, again meeting the
challenge of the diverse kinds of populations we have in the
voting. And that is the kind of usability testing that is
standard in our profession and that we can carry out on a very
rapid basis. But there is no way to give a global answer
without going through that indurative testing process.
Mr. Hoyer. Okay. Obviously, that, however, is going to be
one of the key issues we need to deal with; and I say that
perhaps not at our level. We are talking about a lot of things
that, frankly, States and locals are going to make the
decision--final decision on, perhaps with advice of counsel and
best practices and best standards advice, but----
Mr. Woods. This is where an independent technical group can
come in and demonstrate. Often, we can show you some clearly
undesirable ways to try to accomplish those goals, and we can
point people in these kinds of guidance documents I have
referred to to several different kinds of techniques that will
work and how to do quick tests to verify that your particular
choice as an election official or a State official will be
successful.
Mr. Hoyer. Thank you.
Dr. Woods, let me ask you about the DRE machines, because
I want to know what extent the flaws in DRE machines are
attributable to poor ballot design versus problems that are
inherent in the technology itself or perhaps a combination of
the two. I think Dr. Palfrey just made the observation that DRE
machines themselves--some technology applications have not
worked as well and administrators are not pleased with how well
they have worked. Could you comment on the issue?
Mr. Woods. I haven't looked directly and evaluated those
interfaces. Ohio State did run a study 2 years before the
recent controversy evaluating a variety of different voting
systems and anticipated many of the punch card difficulties and
identified a variety of other problems with the visual layout
of the ballot design, ways that people could miss certain
aspects of the ballot choices they had. For example people were
able to--we had some issues where people--shorter people missed
certain referenda, didn't even realize that they had those
options available to vote on them. That is why there was no
choice made on those issues facing the voter. So those layout
issues can apply.
Mr. Hoyer. Anybody else have any comment on that?
Mr. Palfrey. I think actually Mr. Ney identified one source
of some of these problems when he mentioned the machines that
he looked at and there was all sorts of variety of ways that
lights popped on and went off and whatever. And if there isn't
good, you know, clear instructions on the screen for what is
going on, voters could easily get the wrong idea.
Mr. Hoyer. Seems to me you need to take this technology and
expose it to large focus groups and make a scientific analysis
of the human error rate, user friendliness of the technology. I
think that is going to take a lot of--I imagine you folks are
the perfect people probably to do that on contract for some
group, perhaps the OEA, which would be able to give grants to
carry out such testing. I mean, ultimately, a human being is
going to use this, got to put a lot of human beings in the room
and see what happens.
Mr. Palfrey. Not only sort of human testing, sort of human
laboratory testing where you put them in a room and have them
bang around it, and also field testing could be useful as well.
Mr. Baum. In addition to in-field testing, we are not
really talking about voter education here. That is an oxymoron.
You don't go out and educate voters on the technology. They
come in and use it.
Mr. Ney, you indicated that was your first time seeing that
technology. For a majority of voters the first time they are
going to see that technology is when they go in to vote on it.
That is incorrect. They should have had a familiarity with it
before. You put it out in the shopping malls, you put it out in
the libraries, you put it out in the high schools and let them
go out and play with it some. So by the time they go in and see
it, it is not an intimidation of the equipment.
The Chairman. Especially for certain age years now. My
children, if I buy a VCR today, they install it, my 12-year-
old. It used to be my 17 year old when he was 12. If I need
something done on the computer, they come over, and they help
me. I am the problem. They aren't.
So I know you know the school age children through
computers through school are not the--that they are not--that
they don't need the education. They will have a much easier
time. They don't fear things as I do. I am not a user of
technology. I try and, you know, I do e-mail and things like
that. So there is also age groups.
That is why you have got a great idea, the malls. If you
have a mobile traveling unit that goes to senior centers, it is
you know, open to the public. The schools are great, too. But
school children are going to adapt quicker because they are
used to computers.
Mr. Baum. It depends. I think my kids would be very
intimidated by a punch card machine----
The Chairman. Yeah, that is right.
Mr. Baum [continuing]. And have a whole different
experience with what the marked sensor cards are, too. The
optical scan, you look at one of those, and you are taking a
test.
Mr. Woods. We refer to this problem as the walk-up
interface problem, because people won't have practice. Even if
it is you are voting in the same area, you only do it once a
year. So it is always this kind of walk-up interface, and you
have to remember again how to do it.
I would point you, without even the latest technology, to
the L.A. Olympics in 1984. One major telecommunications company
in the U.S. Human factors group was--volunteered to help design
the information system--the electronic information system for
the athletes, the participants in those games, how to deal with
people with different languages and backgrounds and compute
awareness. And they needed information about the times of their
various practices and preliminary heats and also how to contact
and schedule meetings with that tremendous opportunity to get
to know people from around the world.
It was a tremendous success because they applied the
science of human factors to those walk-up interfaces and
experienced very few problems. People didn't think about it as
a problem or a challenge, how to learn how to operate the
device. They just walked up and found the information or
communication to the people they wanted to contact.
The Chairman. I think Mr. Hoyer made a good point about
focus groups. You have the scientific community that watches
the group's interaction, you have your different type of people
that are watching, but have you the people that participate in
focus groups? I think that in fact, you know, would help a lot.
Mr. Woods. Those usability tasks can be run.
Again, I would point out there are plenty of interfaces and
potential for problems to arise when the election officials
interact in the various stages of processing the votes. We
shouldn't forget about those interfaces as well, because that
is where we would inadvertently introduce large inaccuracies.
Mr. Hoyer. Thank you.
Let me sort of ask a global question. If we could do only
one thing this year, do you have any thought as to what that
ought to be? In other words, the first step. I have mentioned
some first steps personally, but do you have any views?
Mr. Woods. Well, from my point of view, it would be very
easy to put out a kind of initial guide to the human factors of
voting technology that would give some examples, some basic
principles and some testing focus group techniques, how to run
the focus groups, to provide an initial resource to local
election officials as they contemplate what decisions they
should make now or put off as more information and options
become available. And I think there are independent bodies,
national laboratories such as Brookhaven National Lab or
universities, who could respond quickly to provide that
guidance.
Mr. Hoyer. Any other Comment?
Mr. Baum. I think if you were only going to do one thing--
and that is my sincere hope that that is not the case--it would
be to publish some information on how to live with what you
have. I mean, yes, there are challenges in each one of these
technologies. But with proper planning, with good processes in
place and good training in place, we can certainly reduce the
error rate of even the equipment that is there by handling it
better.
Mr. Rivest. I think I would respond by saying to create a
organization whose mission is to assist the States, the local
officials, by creating a fund, and organization to generate and
distribute and share information about voting systems,
supporting research, supporting focus groups on usability,
supporting hacker attacks on systems to see what can be done.
You know, publishing a source code, when that could be made
available. An organization whose goal is to help the State
officials know everything that needs to be known about these
systems so they can choose well.
Mr. Palfrey. I would second that. Because I think one of
the biggest problems is lack of information, okay? You have
election administrators who are in a situation, whether the
money is coming from the Federal Government or not, they are
basically under pressure. They are going to have to replace
some equipment. They don't have good guidance for how to do it,
how to implement it, what kind of pitfalls there are. And it
goes to security, it goes to all these other different issues.
I think you really do need an agency or an organization who is
responsible for collecting and providing an information
clearinghouse of that sort.
Mr. Hoyer. As I say and as you know, we do have an
organization, small and incorporated within the FEC, the OEA.
But, I was stricken by the fact that all of the Secretaries of
State that testified and all of the election administrators
that testified were looking to the OEA for better information,
better--best practices advice and counsel, better standards
advice. And you have echoed that, and I hope we do that.
I want to say very quickly, I hope we don't do just one
thing. I agree with that, and the question was, to that degree,
rhetorical.
Last question I will ask, Mr. Chairman, and I have some
others here but, if we might, I hope that our staffs can feel
free to contact you from time to time and get the value of your
expertise.
What would the best way for Congress to facilitate research
and development of new voting technology, and how can Congress
make it easier to develop and implement new technology? I think
that is going to be, I hope, one of the aspects of what we will
do. Because such a limited market, such a relatively--
``sterile'' is the wrong word but stable market, not a very
volatile or vibrant market here, so the technology has not
turned over as quickly as it otherwise might have. Florida has
spurred that, and we will have a short window, in my opinion,
to take advantage of that sort of awareness. So how can we best
do that--that is, to spur our DT&E in this area?
Mr. Baum. I would like to point out that, although the
market for public elections may be set and stable, there are
other kinds of elections where this technology applies. There
are union elections, professional organization elections, and
all kinds of places where these other factors are also a part.
That allows the opportunity for the private sector and the
public sector to cooperate here in building up standards that
become then, generally accepted election standards. So that is
one area of cooperation with the private sector where Congress
can make a huge contribution. Also, as we have all stated
before, coming up with standards that can be applied in these
areas would make a significant contribution.
Mr. Palfrey. I think one of the barriers to innovation, is
that election administrators have to be cautious when they make
an acquisition of a replacement for whatever technology they
had. And what has happened is that they just switch wholesale
to a new technology.
I think one thing that would help innovation--and there are
start-up companies that are out there trying to work on
innovation in this area--is to provide some sort of funding for
pilot testing, pilot experiments, and field testing so that you
don't have to jump into it full force. You can try it out in a
couple precincts, see how it is working, and monitor it. I
think that is one thing that might help.
Mr. Rivest. I think perhaps some guidance to either NSF or
DARPA, or both, to support research in this area could be
helpful. DARPA may seem strange, but in fact there is national
security involved with these elections. NSF has traditionally
funded research in computer security and cryptography, for
example. As my colleagues have pointed out, there are also
businesses that need to be involved. I am not sure what the
best way to encourage them to innovate is.
Mr. Woods. Another aspect of the activity of the
independent resource in that area, we need a mechanism for
people to monitor potential or emerging sources of inaccuracy
and problems. This is almost a classic advice that we give to
almost every agency that has safety, for the potential for a
crisis to arise. We don't want to react after the fact to a
crisis, like Florida, and try to repair and intervene. High
reliability organizations are out there testing and monitoring
their systems to notice early warning signs or even dress
rehearsals. Instead of rationalizing away the dress rehearsal
of inaccuracies in this, our system, which we have been too
tolerant of, we need these bodies to be out there saying,
whoops, look at what we are starting to see as problems,
sharing that information so people can change the technology,
change the education, and change the procedures they are using
to prevent crises from happening.
Mr. Hoyer. Thank you very much to the panelists. You have
been very helpful, I think, and very thoughtful. Hopefully, as
I said earlier, we will be able to work with you so you can
help us. Again, there's a much longer term--States and
localities and whatever agency, whether it is the existing OEA
within FEC or some other organizational structure like that,
work with them to accomplish objectives which I think clearly
all of us want to accomplish.
And they are not partisan in nature. Everybody wants to
make sure that, not only does every American have the right to
vote, but every American's vote is made easier to cast, more
accurately, and counted correctly. And although this is not
solely a technology problem, clearly it manifested itself in
many ways as a technology problem, as well as a human behavior
interface with technology problem.
Thank you very much; and thank you, Mr. Chairman.
The Chairman. We do appreciate your testimony today. You
are an important component from the technology side.
Also one thing, I want to assure you too, and I feel very
good about, the way we have proceeded and also about the ideas
that Mr. Hoyer has been able to develop and we have been able
to develop together and, you know, to reach some basic
consensus among members when you start to talk about it.
I think the statement you made, Professor Woods, is what we
are doing--you know, this isn't debating Florida, but Florida
caused the debate. And I can barely remember anyone coming to
me over a period of 20-some years, whatever office I have held,
of saying, gee, this certain technology needs to be looked at
or it needs work. It wasn't drawn out until you had some
national view. And that is what the Presidential did.
So I think what we are doing is trying to work with the
desires of the locals to do something about the existing
systems that are out there, but also I think what we are doing,
is we are not in the middle of the crisis right now, we are
personally not letting this situation go. We are driving a
piece of legislation. I think it is going to be good. But I
think we are reacting the right way to look ahead.
So, I think your statements were well taken. It is the way
to proceed on this, and I think that is what we are doing.
Mr. Hoyer. Mr. Chairman, on a nonelectoral reform issue, if
I might, I want to say to you and to Dr. Woods that--how highly
resentful we in Maryland are that you have taken our president
of the University of Maryland away. I don't know whether you
have had an opportunity to get to know Brit Kirwan very well,
but he is an extraordinary asset. He is a wonderful human
being. Ohio State is very fortunate to have him. We miss him a
lot in College Park.
I am on the board of regents of the university systems, and
I was a graduate of the University of Maryland and worked with
Brit very, very closely for all the time that he was at the
university which, as you know, was over 2 decades. I trust that
Ohio appreciates him as much as they should. You certainly
compensated him better than we did; and he appreciates that, I
know. But I hope you enjoy working with Brit. He is a terrific
fellow.
The Chairman. We like him, and we in Ohio will not mention
what Baltimore took from us in Cleveland a few years ago. We
will just leave it at that.
On that note, I ask unanimous consent that witnesses be
allowed to submit their statements for the record. Members have
7 legislative days to insert extraneous material into the
record, and for those statements and materials to be entered
into the appropriate place within the record. Without
objection, so ordered.
I ask unanimous consent the staff be permitted to make
technical and conforming changes on all matters considered, by
the committee at today's hearing. Without objection, so
ordered.
Having completed our business of the day for this hearing
on election reform, the committee is hereby adjourned. Thank
you.
[Whereupon, at 12:35 p.m., the committee was adjourned.]