[House Hearing, 110 Congress]
[From the U.S. Government Publishing Office]
MEETING TO DISCUSS THE STATUS OF THE INVESTIGATION INTO THE FL-13
CONGRESSIONAL DISTRICT ELECTION
=======================================================================
MEETING
before the
COMMITTEE ON HOUSE ADMINISTRATION
TASK FORCE FOR THE CONTESTED ELECTION IN THE 13TH CONGRESSIONAL
DISTRICT OF FLORIDA
HOUSE OF REPRESENTATIVES
ONE HUNDRED TENTH CONGRESS
FIRST SESSION
__________
MEETING HELD IN WASHINGTON, DC, AUGUST 3, 2007
__________
Printed for the use of the Committee on House Administration
Available on the Internet:
http://www.gpoaccess.gov/congress/house/administration/index.html
U.S. GOVERNMENT PRINTING OFFICE
38-707 WASHINGTON : 2007
_____________________________________________________________________________
For Sale by the Superintendent of Documents, U.S. Government Printing Office
Internet: bookstore.gpo.gov Phone: toll free (866) 512-1800; (202) 512�091800
Fax: (202) 512�092104 Mail: Stop IDCC, Washington, DC 20402�090001
COMMITTEE ON HOUSE ADMINISTRATION
ROBERT A. BRADY, Pennsylvania, Chairman
ZOE LOFGREN, California, VERNON J. EHLERS, Michigan,
Vice-Chairwoman Ranking Minority Member
MICHAEL E. CAPUANO, Massachusetts DANIEL E. LUNGREN, California
CHARLES A. GONZALEZ, Texas KEVIN McCARTHY, California
SUSAN DAVIS, California
ARTUR DAVIS, Alabama
S. Elizabeth Birnbaum, Staff Director
William Plaster, Minority Staff Director
Task Force for the Contested Election in the 13th Congressional
District of Florida
CHARLES A. GONZALEZ, Texas, Chairman
ZOE LOFGREN, California KEVIN McCARTHY, California
STATUS OF THE INVESTIGATION INTO THE FL-13 CONGRESSIONAL DISTRICT
CONTESTED ELECTION
----------
FRIDAY, AUGUST 3, 2007
House of Representatives,
Election Task Force,
Committee on House Administration,
Washington, DC.
The task force met, pursuant to call, at 10:00 a.m., in
Room 1310, Longworth House Office Building, Hon. Charles A.
Gonzalez (chairman of the task force) presiding.
Present: Representatives Gonzalez, Lofgren, Ehlers, and
McCarthy.
Staff Present: Liz Birnbaum, Staff Director; Thomas Hicks,
Senior Election Counsel; Charles Howell, Chief Counsel; Janelle
Hu, Election Counsel; Jennifer Daehn, Election Counsel; Matt
Pinkus, Professional Staff/Parliamentarian; Kyle Anderson,
Press Director; Kristin McCowan, Chief Legislative Clerk;
Daniel Favarulo, Staff Assistant, Elections; Fred Hay, Minority
General Counsel; Gineen Beach, Minority Election Counsel; and
Bryan T. Dorsey, Minority Professional Staff Member.
Mr. Gonzalez. I call the meeting of the task force to
order. I thought I would briefly go over a chronology to give
us a proper background as to where we are today in order to
receive this report from the representatives from GAO. On the
23rd of March, 2007 the late and wonderful Chairwoman
Millender-McDonald, established the three-member ad hoc
election panel, two members of the majority and one member of
the minority, to oversee matters relating to Florida's 13th
congressional district election contest and recommend final
disposition to the committee of that contest.
I was appointed, as well as Representative Zoe Lofgren, to
be the majority members, with myself serving as the Chair. On
the 16th of April, Ranking Member Ehlers recommended
Representative Kevin McCarthy as the minority member. On the
17th of April counsel to the parties informally briefed the
panel on the status of Florida election contest proceedings.
Also present at that briefing was Congressman Dan Lungren from
the minority. On the 2nd of May 2007 there was a public task
force meeting to initiate an investigation of Florida's 13th
congressional district election, and we authorized myself to
secure the assistance of the Government Accountability Office,
which was requested to design and propose testing protocols to
determine the reliability of the equipment used in the Florida
13 election taking into account recommendations by the
contestant and contestee.
On the 7th of June, there was an internal briefing and
planning meeting of the task force and GAO to discuss GAO
working plans. On the 14th of June, there was a public task
force meeting to approve the GAO work plan. On the 27th of July
there was an internal briefing and planning meeting of the task
force members and staff with GAO to work on the plan and
ascertain its progress. Today, the 3rd of August is our follow-
up meeting, our public meeting, to ascertain the status of the
work being conducted by the GAO. The purpose of today's meeting
from the Chair's perspective, of course, is basically a status
report from GAO, any determinations that can be made at this
time, what remains to be done, and also to establish a process
for the task force to operate during the August recess.
Anticipating that the work has not been completed by GAO
they will require more time and there may be requirements or
needs that will arise during the August recess during which
obviously members must somehow establish a process where we can
come together, even if it is by teleconference, in order to
make some determinations and pass on those requests. Appearing
before the task force today will be Naba Barkakati, senior
level technologist; Gloria Jarmon, managing director,
congressional relations; and Jan Montgomery, assistant general
counsel.
I believe that we will have a statement that is going to be
read by Mr. Barkakati, but at this time, I would yield to the
other members of the task force if they wish to make any
opening remarks. And I would recognize the minority member, Mr.
McCarthy.
Mr. McCarthy. Well, I appreciate the chairman for
recognizing me. I am excited about moving forward as we are. I
appreciate the update you had given us last week. I will tell
you from that update that I was very pleased to hear the
relationship and the working relationship with everybody down
in Florida from Sarasota to the State. It seems like they had
open arms. Everybody wants to get to a conclusion here. And so
I look forward to today's hearing and I yield back the balance
of my time.
Mr. Gonzalez. Thank you very much. And the Chair will
recognize Ms. Lofgren.
Ms. Lofgren. I would just like to get to the report so I
will yield back to the chairman.
STATEMENTS OF NABA BARKAKATI, SENIOR LEVEL TECHNOLOGIST; GLORIA
JARMON, MANAGING DIRECTOR, CONGRESSIONAL RELATIONS; AND JAN
MONTGOMERY, ASSISTANT GENERAL COUNSEL
Mr. Gonzalez. All right. At this time, the Chair will
recognize Dr. Naba Barkakati.
STATEMENT OF NABA BARKAKATI
Mr. Barkakati. Chairman Gonzalez, Mr. McCarthy, Ms.
Lofgren, I am here today to update you on our progress
regarding the Florida 13 voting system review. I want to begin
by thanking the task force for its continued support of our
efforts, and I also acknowledge the cooperation of everyone
else involved in helping us get the document and information
that we have needed to conduct the work so far. We have
accomplished a lot in the past few weeks from the time when we
agreed on the engagement plan on June 14, but we still have
some work to complete before we can formally conclude any
conclusions. To conduct our work, we have visited Sarasota
County, Tallahassee, had discussions with Florida's Secretary
of State, Division of Elections, and then we had talked to
Sarasota County Supervisor of Elections, ES&S, the manufacturer
of the system, and the team leader of the Florida system in
Florida State University's source code review team, the one who
looked at the iVotronic source code.
By analyzing the detailed ballot data that we have received
from Sarasota County we have verified that 1,499 iVotronic
machines had recorded 119,919 ballots in the 2006 general
election, and of these there were 17,846 undervotes in the
Florida 13 race, which corresponded to a 14.88 percent
undervote rate on the iVotronic machines. We examined the
undervote by machine, precinct and ballot style, looking for
any patterns. We have not yet noticed any apparent patterns,
but are continuing with our analysis.
While we have not completed reviewing all the testing that
is done so far, there are some observations we can make at this
time. Of the many different types of testing that one can
perform in a voting system, we have focused our efforts on two
types of tests that are called ballot testing and load testing.
Ballot testing refers to the functional testing of the voting
system focusing on the different ways in which the voter may
make selections on the ballot and then cast a ballot with the
iVotronic machine.
Load testing subjects the voting machine to a large number
of votes in order to verify that the system can properly handle
the expected volume of ballots. Considering all the different
ways that the iVotronic machine enables a voter to make
selections and cast a ballot, we counted 112 different ways
that the voter could make such selections, by navigating
through the pages and casting the ballot in the Florida 13
race, assuming that is the only race on the ballot. We found
that the Florida certification tests, the Sarasota County's
logic and accuracy tests, and the parallel testing that was
done under the state audit of the Sarasota County election
covered 13 of the 112 ways of the ones we enumerated. We have
not yet assessed whether this is significant.
Load testing was done as part of Florida's certification
testing, but the ballots were machine generated using a testing
feature built into the iVotronic machine, which means that
users did not touch the screen to make selections and cast
ballots. Again, we have not fully assessed whether this is
significant.
Last week we obtained access to the iVotronic source code
under a nondisclosure agreement, and then were able to review
the source code ourselves and verify some of the items that are
reported in the Florida State University's review of the source
code that was also done under Florida's audit of Sarasota
County's elections. We plan to discuss the source code further
with the audit team, as well as ES&S in the near future.
One item to note from the code review report is that the
review team did not convert the human readable source code into
machine code, which we call object code, and compare the
resulting object code to the certified version of firmware for
the iVotronic machines that were used in Sarasota County. These
steps would be necessary to assure that the reviewed source
code corresponds to the firmware in the iVotronics. As part of
our review of the State audit, we also examined the selection
of sample machines for the parallel testing and the firmware
verification. Ten machines were used for parallel testing, and
the State audit had examined the firmware from six randomly
selected machines. Our preliminary analysis indicates that
these sample sizes were too small to apply the results across
the board to all iVotronic machines that were used in the 2006
general election in Sarasota County. For example, checking the
firmware in six randomly selected machines is inadequate to
conclude that the firmware was not compromised on any of the
machines that were used in Sarasota County. I want to stress
that these are just our preliminary observations. It is not
clear to us yet whether these are items we think will need to
be tested. For any further testing that we identify we plan to
determine the relevance and the significance of the test, the
test procedures, and estimate the resources needed and the time
required to conduct the test.
Besides analyzing potential tests, we are planning to
examine the testing conducted by ES&S and identify other tests
besides ballot testing and load testing that I had mentioned
earlier that could be used to determine whether the voting
system contributed to the undervote.
Mr. Chairman, this concludes a summary of my written
statement. I would be happy to respond to further questions
that you and other members of the task force may have at this
time. Thank you.
Mr. Gonzalez. Thank you very much, Doctor.
[The statement of Mr. Barkakati follows:]
[GRAPHIC] [TIFF OMITTED] T8707A.001
[GRAPHIC] [TIFF OMITTED] T8707A.002
[GRAPHIC] [TIFF OMITTED] T8707A.003
[GRAPHIC] [TIFF OMITTED] T8707A.004
[GRAPHIC] [TIFF OMITTED] T8707A.005
[GRAPHIC] [TIFF OMITTED] T8707A.006
[GRAPHIC] [TIFF OMITTED] T8707A.007
[GRAPHIC] [TIFF OMITTED] T8707A.008
[GRAPHIC] [TIFF OMITTED] T8707A.009
Mr. Gonzalez. Do either of the other representatives from
GAO wish to address the task force? Ms. Jarmon.
Ms. Jarmon. No.
Mr. Gonzalez. And Ms. Montgomery.
Ms. Montgomery. No.
Mr. Gonzalez. Very good. I think I will start off with a
couple of questions. And we have the luxury right now, I
believe that the House is still in recess, so we are going to
try to accomplish as much as possible. I do have some
questions. Some of it is so technical, Doctor, I think at your
briefing last week we tried to get into it. At the end of the
discussion, I still have such a layman's understanding of what
you really do in many different ways, but I want to start off
with ballot testing.
One concern that you have, and I want to preface everything
that we go into today that you have not made a determination
whether this is significant or insignificant. But let us start
off with some of the areas of concern. Ballot testing, since
only 13 of 112 different ways the voter could have made a
selection in the Florida 13 race specifically is an area of
concern, and I want to know why would it be a concern.
Mr. Barkakati. I should probably say that we are trying to
stay away from saying it is a concern. Mainly because it is
just an enumeration of tests that you do. And in the end, we
may decide or we may conclude from our deliberations that it
really doesn't mean much and the functionality of the machine
tested fine by doing the cases that were done. So in other
words, the other possibility also remains that even though we
have enumerated 112, and in reality it could make combinations
that can grow even farther by considering other steps, so we
may say that, well, that is extraneous and doesn't require to
be done. So that is why we haven't said that it is a concern.
And also we have kept in mind that between 13 and 112, all we
are saying is different ways of selecting in a two-way
situation in the ballot, and it probably won't be too time-
consuming to do, even if we were to raise it as an item.
Mr. Gonzalez. I appreciate what you are saying. And the
description that it is a concern is the Chair's description.
And we could say it has--somehow it has drawn your attention. I
don't know how else to say it. For a layman, if you say you
have 112 variations of something, you think, well, you better
test as many of those as you can. But that is not necessarily
true when you get into the technical sense, and we will just
wait for your expertise to be applied. Load testing, one thing
that you indicated is that on the load testing, during the
certification testing that it was machine generated, which
means users do not touch the screen to make a selection.
Neither the Florida audit nor Sarasota County's logic and
accuracy testing performed load testing. We have not yet
assessed whether this is significant. So again, this is
something that has drawn your attention as to how you test,
whether there is the human element, the actual touching and so
on.
Mr. Barkakati. Yes. The machine has a built in feature. It
can say test for us, you know, casting 9,000 ballots. And it
will do it for you by almost exercising as much of the software
as it can, but excluding the part where you have touched the
screen. So we were raising it as an issue that to truly test a
machine for its ability to handle, you know, including the
human element of entering the vote, it wasn't there. Basically,
that is what the observation is. And it may turn out that it
might be good to do so or maybe it is not, again, not a matter
that needs to be pursued further.
Mr. Gonzalez. The other item that you, and we are reading
from your draft of course, is that one of the items noted in
the report was that the review team did not, one, convert the
source code to object code; and, two, compare the resulting
object code to the object code that was used to run the voting
machines in Sarasota County. Can you translate that?
Mr. Barkakati. Yes. Essentially the software is in a
programming language. They call it C, C programming language,
primarily. And you can read it. A human can read it. I mean,
you will be able to read it. There will be words that are like
English like words, but of course they have a special meaning
in the computer sense. And so programmers should write using
that language, and that is what we would be reviewing if we
were reviewing source code. And those programs are going to lay
out the intent of the machine, like, you know, cast the ballot,
do this or that. So we will be able to read through the logic
and say, oh, this looks good.
But of course, the machine doesn't run that. You would be
using it though like something called Compiler or Linker or
things that enable you to convert source code into binary, you
know, ones and zeros that are not readable by us, but that is
what the machine is going to execute. And that is the version
that is going to be loaded into a chip.
You know, they call it EPROM. That is like a flash memory.
Where it can stay for--even if the power is turned off, it will
stay on. And that is what the iVotronic always runs. So one of
the problems we have is even when I review the source code I
would like to know that what I reviewed was producing exactly
the same binary, which it probably does for all I know, as what
is in the machines. And all we are saying is that that step was
not done.
And we have talked to the team and they were saying that
they were relying on other certifications of that. In other
words, like independent testing already have done it. And we
have not verified. I mean, it hasn't given us complete
assurance that we have seen the steps performed and confirmed.
Especially if you know how this is done. It is not going to be
a lengthy, long-drawn process. If you had the tool, you could
do it very fairly reasonably well. Within a day maybe you can
complete it. So that is where we are wanting to see this done
basically. And it wasn't done. But then again, if we find that
the certifications that are provided by ITA, an independent
testing authority, are maybe in our judgment it is okay, then
perhaps it is all right. But I must say that even the code that
I reviewed I was always wondering if that is--I mean, I would
have liked to perform that step myself, except that you need
those tools from the manufacturer. And we didn't have that
available to do it immediately.
And that is the only point we are making in that they are
saying it wasn't done by the team and it was taken as a given,
as an assumption. So there is nothing worth faulting the team
for that. But it would be a critical step to complete to really
have an assurance that, okay, the code was reviewed and it was
really the same thing that drove the machine. That is the idea.
Mr. Gonzalez. It was one of the more contentious points,
and that is access to the source code. But my understanding is
you have not really had any real objection gaining access. But
we are working on, I believe, and Ms. Montgomery might be able
to answer it, on the finalization of the confidentiality
agreement, which will allow complete access and use of the
source code as you have indicated you wish to do is that
correct?
Mr. Barkakati. Yes. I would start, and Ms. Montgomery can
add also, yes, we were given complete access. We were not
allowed to take the source code with us, which is
understandable because it is a proprietary item. But we can
look at it as long as we want, as much as we want, take notes
on it. We have been able to do all of that. And for further
discussions and further technical information, we do have work
going on on another agreement, and perhaps you can add.
Mr. Montgomery. I believe we have an agreement with ES&S at
this point. In theory the team will be going down very soon in
the next several weeks to Nebraska to access the internal
testing information from the ES&S. So I would say we have had
very good cooperation and we have access to the information we
need to do this job.
Mr. Gonzalez. Excellent. Thank you. One of the last points,
and one that I sort of understand and I think, again, a
layperson may understand, and this is going to be on page 5 of
your draft report: As a part of our review of the State audit,
we examined the selection of samples for the parallel testing
in the review of the Sarasota County election practices. Our
preliminary analysis has found that these sample sizes are too
small to support generalization of the results to the overall
population. Now, to a layperson that simply means you have
1,500 machines that were used here in the election and that a
certain number were actually utilized for testing and so on.
And further you say: Our discussions with Florida officials
indicate that such limitations resulted from court imposed
restrictions on machine access and resource considerations are
performed in the testing. So is it really the number of
machines that were tested? How many out of the 1,500 at this
point?
Mr. Barkakati. Well, as it mentions here, 10 machines were
used for parallel testing. But if you remember what happened is
that they tested five machines that were not used in the
election, and five from the set that was used in the election.
Now, to pick the machines that were not used in the election
they only had a very few left over. They didn't have the choice
to pick from a large population.
So what I am trying to say is that there was such a small
set of machines that were not used in the election that they
could only pick five out of that. And then, of course, that
forced them to do five on the other side to keep it the same.
So they were not thinking from the point of I guess
generalizing using statistical methods and all to the whole
population.
Now, to say one thing in a clarification is that if you
want to make a statement about the whole population of
machines, then you have to kind of explain how much confidence
you want, you know, say 99 percent confidence, and how much
error you are willing to tolerate, meaning how many of those
machines, what percentage might still not conform to your test
and you are okay with that.
Say, if you want to say with 99 percent confidence out of
1,499 machines, I want to be able to say that only 5 percent at
most have an error. You know, whatever the condition is; error
in testing or error in not having the right version, it doesn't
matter. Then the selection would be something like our
statisticians say, 89 machines have to be tested.
So all we are saying is that if you wish to kind of make a
conclusion that, okay, we tested the machines and they are all,
say, don't have any problem, then you have to pick 89 and
perform the test. And at the end all you are able to say is
that with 99 percent confidence, I know that only 5 percent at
the most may have an error, so 95 percent is okay, you know, of
the population. And based on if things seem normal, the
machines may be different levels of the confidence and
different levels of error, but you have to set those things and
pick a sample and then your conclusions can be essentially
statistically significant and everybody can agree that, yes,
they have tested thoroughly. That is the reason for raising the
issue.
Mr. Gonzalez. Thank you. And the last thing I want to point
out on page 5, you said there may be some other causes, and we
covered this in our briefing last week. One would be the
effects of provisional ballots. And we understand provisional
ballots. The other thing that you said was environmental
conditions. And in last week's meeting, I didn't know what you
meant by environmental, and you pointed out temperature. And
that is, for instance, the machines can operate according to
certain tolerances, such as temperature and such. When you say
environment to a layperson, sometimes we think, you know, real
life conditions. In other words, where they were placed, how
long they were sitting there, the angle that people had to
press their selections and such. But that is not what you mean.
Environment is something entirely different?
Mr. Barkakati. I think either of those factors can come in
the primary environment of things or more like to be able to
handle shock, vibration, humidity and temperature. Those are
the kind of things that come into play there. And those things
are done particularly by the independent testing authority. So
we may very well conclude that that part is taken care of and
we don't need to worry about it. But that is what we meant in
that instance.
Mr. Gonzalez. All right. Thank you. And that concludes my
very long number of questions. But I know that the other
members of the task force at this time will have questions. And
I do want to welcome the ranking member of the full committee,
Mr. Ehlers. Good morning.
Mr. Ehlers. Thank you.
Mr. Gonzalez. And at this time the Chair will recognize Mr.
McCarthy.
Mr. McCarthy. Thank you, Mr. Chairman. Just a couple
questions. Do you have any determination currently about how
far along you are in doing the report. I mean, are you 20
percent along. Could you put a percentage to it?
Mr. Barkakati. I probably couldn't put a percentage, but I
should say in terms of our four objectives that we laid out,
the preliminary ones like looking at of course the first part,
the first two are basically figuring out what systems are used
and how they are used. I would say that part is done. We have
gotten all the information from everywhere to figure what
machines were used, how and all that. The second, including
the--we are still analyzing, though, the level of problem, like
undervote pattern, et cetera, is not completely done. We have a
little left there.
Looking at the previous testing was one of the things. You
know, like previous testing by manufacturer, by Florida State
and then Sarasota County. We have not looked at the
manufacturer testing, but we will be looking at that by August
15 or so. Which means that could get done soon. And that will
settle questions like whether they have maybe internally tested
certain things that we haven't seen elsewhere and we could
check them off. But I would say the most important thing that
is remaining are the things that we say here that we have not
determined the significance of, to have an internal discussion
and deliberation to figure out if it is important or not, and
whether we should raise them to the level of tests to be done.
And then, of course, the figuring out whether, you know,
how much time is needed, how much resource, so we have some
information to decide whether even to go forward; if it is cost
prohibitive or resource prohibitive, then you cannot do it.
So in other words, mostly the fourth item where we say
figure out what tests might be needed and come back to us with
detailed information about them is the ultimate biggest thing
remaining. So I am not putting a percentage, but we are
devoting our time to basically checking out the testing at ES&S
and doing our own internal deliberations and discussions to
figure out the testing we should be, minimum or more testing
that we will be proposing to you or listing to you with
information that might help you decide whether you do it or
not. And again with the caveat that maybe this is something to
say, all checked out there is nothing much you can propose.
That is a possibility.
Mr. McCarthy. Now, you have also studied the audit that the
State did and you have spoken to some of the PhDs that did the
study as well. Is that correct?
Mr. Barkakati. Yes. And we did say very briefly that we
were able to verify certain, like things that they talk about,
like the user touches the screen and this is how the vote gets
cast. We could check a few sequences of code and check what
they say about the code. And they, of course, match what they
say.
Mr. McCarthy. So you have a check and balance of that?
Mr. Barkakati. Yes. We didn't check all the details of it,
and we are going to do a little more. And as a matter of fact,
they also have been very responsive and he has, the leader has
offered to help us out if we need to; in other words, because
they have already gone through. Although we were trying
initially somewhat more to do on our own to be independent and
see on our own what is there.
Mr. McCarthy. Now, you are so far along and you have done
the analysis. Is there anything out there that makes it drive
you in a different direction? Is there, for no better word, is
there a smoking gun? Is there something out there that says,
hey, this is the problem already?
Mr. Barkakati. So far, I would say no, other than the
things that we raise as far as sample of the machines, you
know, like are not adequate and all. Those are more along the
lines of being able to say the same thing that there is no
smoking gun. Sure.
Mr. McCarthy. You just take them away.
Mr. Barkakati. Right.
Mr. McCarthy. And one thing, and I just draw on what the
chairman said, you have these 112 ways and only 13 of them had
been tested. And that kind of raised a question to me last
week. And when I probed further, I was wondering why the State
did these 13. And it was described to me the reason you did
these 13 is because the contestee, Ms. Jennings, argued that
there was something wrong. These 13 are the top analysis if
something went wrong and denied Ms. Jennings a vote, these 13
would show it. That is why they didn't test 112 because these
others that could be left would mean if something went wrong in
these other tests, it would give Mr. Buchanan more votes, and
he was already in the lead, so why would we test for the sake
of saving money, I guess, is that correct?
Mr. Barkakati. Basically, the reason for our number being
higher is because we think of like, well, the voter could go
forward and press the page back key and change it again or
change it from the review screen. There are multiple options of
doing things. When you take all the combinations it becomes
larger. Now, I would agree that they were primarily making sure
that Ms. Jennings could get votes, no matter how you went. And
that is what the rationale was given to us. And in that
scenario, there could still have been a few more to check the
other side, which is like you know, maybe Mr. Buchanan could
also be given votes by changing and that would raise it to, I
think, 18 or something like that instead of 10.
Mr. McCarthy. But these were the top 13?
Mr. Barkakati. Yes.
Mr. McCarthy. And you are testing everything, so it could
come back that maybe Mr. Buchanan got more votes?
Mr. Barkakati. I should really kind of emphasize that those
tests are going to only confirm that each party is able to get
votes and no problem. And the logic and accuracy test that the
counties do is also geared towards making sure that every
candidate on the list in any race is able to get a vote by the
machine. They cast it and confirm that it happens.
Mr. McCarthy. Now, you say you have looked at the source
code. And then when we talked last week, you have your own
expertise when it comes to programming and a few others. Was
there anything looking at this source code, I know you haven't
been able to run it because you are getting the agreement,
anything shoot at you just understanding programming and
others?
Mr. Barkakati. No. As a matter of fact just as the other
team, the Florida State University team that had looked at the
source code, their conclusion was it is a simple program that
is a very sequential processing of what happens, like a user
touches something it keeps on checking, and it determines what
the vote is, marks it and saves it when you press the vote cast
button.
So in that sense, I found the same exact thing. It was
readable. It wasn't hard to read or understand. And it was
simple enough to follow that logic essentially. So nothing
jumped out when I looked at the source code.
Mr. McCarthy. So when you looked at the source code you got
by looking at it the same thing that the PhDs that looked at it
from the State and the court did.
Mr. Barkakati. Yes. Because they did point out anything
that they saw in terms of like other flaws and all in a
proprietary appendix which wasn't released to the public, and
we could see those also. Partially we looked at some of those
too. Because they pointed out certain things that are not
pertinent to undervote. But we will explore that a little more.
If you recall, there were three appendices that were not
released to the public. And they included more information, but
they were saying that is not pertinent to an undervote problem,
and we have been looking at that as well.
Mr. McCarthy. Normally, and you may not be able to answer
this, but for our scheduling I know there is a motion that is
going to come forward, and we are going to a district work
period for the month of August. Do you believe you may have the
final report during August when we are out? Do you foresee that
happening?
Mr. Barkakati. I really think that we would require--I
mean, since it is only a month we will require this time to
really do our--there is a little bit of work remaining to
figure out testing done by ES&S and talk to some of the code
review team again. So it looks like that most likely it don't
happen that the report will be done.
Mr. McCarthy. The only reason why I ask is mainly for
scheduling. I want to make sure we have our public meeting with
them when that comes back. I am actually very pleased with how
everybody is working together. And it sounds like everybody
wants to get to a conclusion down there. And that is nice to
know from the manufacturer down to the election officer. So I
appreciate your work. I yield back, if I had any time left.
Mr. Gonzalez. Thank you very much. The Chair will recognize
Ms. Lofgren.
Ms. Lofgren. Thank you, Mr. Chairman. I think you have
asked all of the substantive questions I had. The only, I
guess, remaining question I have, is there anything we could do
to help you conclude this, any effort that we could make that
would assist you in getting this job done efficiently and as
soon as possible.
Mr. Barkakati. I would say that the task force was very
helpful when they initially told everyone they needed
everybody's cooperation. And we have not really had any problem
with anyone. So it seems that we are in touch with the staff,
and nothing is there right now that is holding us back, other
than the mere fact of like looking through everything and
deciding on our own. So it is a deliberative process a little
bit. And that is kind of expected that you would have to do
that. So I would say no. If I am wrong and we turn out to be
needing some help, I think we know how to--
Ms. Lofgren. You know how to get ahold of us.
Mr. Barkakati. Yes.
Ms. Lofgren. That was the only question I had. I think
obviously we don't know the answers yet, but we are on our way
to getting whatever answers can be obtained. And that is all we
hope to do in an objective, fair and expeditious manner. So I
think that this is a good meeting and a good report. We all
wish it were over, but it isn't yet. I don't know if Mr. Ehlers
has comments. I have a motion I would like to offer at the
appropriate time.
Mr. Gonzalez. Thank you, Ms. Lofgren. The Chair recognizes
the ranking member.
Mr. Ehlers. Thank you very much, Mr. Chairman. I am sorry I
was late, but I skimmed through the report. There is just one
point I want to raise, and it looks like you are on track and I
appreciate your good work on this. I have been involved in
elections since 1974, which shows how ancient I am. And I was
also involved in writing HAVA. In fact, I wrote the very first
part, which was the standards part. And I emphasized to NIST
when we did this, and we did require this is in the
requirements, and that was the human factors.
Because I had observed during my many years working with
elections, the majority of errors are human errors, they are
not machine errors, they are not equipment errors, they are
human errors. And that we try to accommodate that in HAVA. I
don't think we did it nearly as well as we should have, and
certainly not as well as I wanted to. But we did do some of
that. And running through your draft here and the discussion I
have heard the short time I have been here, it is almost
entirely about the machines; was the source code okay, was the
object code okay, was the machine operating properly and so
forth. I didn't hear a word about the human factors, which
based on my experience I would think would have to rank higher
in probability than either the machine or the source code
errors.
What have you done on the human factors part? Do you have
expertise in GAO or are you using experts in GAO to analyze the
possible human factors in this case?
Mr. Barkakati. It is true that we have not addressed the
human factors issue yet here. And of course, we are aware of
quite a few other studies where they have looked at the human
factor and kind of tried to draw attention to the fact that
there are similarities in other counties of similar--layout of
ballot and where they might have had undervotes, higher rates
like this one. We were trying very hard to eliminate, you know,
stay away from the human side because it is so--you know, we
couldn't really figure out how to completely quantify that
basically. But I would agree completely that there are--I mean,
it is very much possible that the whole reason was human
factors. And right now I should say that we are trying to--we
have not addressed it since a lot--there have been a lot of
people who have raised the issue.
We consider--we would probably have to consider this as we
go forward and before we issue a final report as an element,
whether either explicitly to say that we are unable to or we
could not or did not or something else. You know, that we
considered it. And maybe you can propose something that can be
done in that phase too. But we have not done much other than
reading the reports that are already out there.
Mr. Ehlers. I would strongly encourage you to. Because as I
said, based on my many years of experience that is the most
frequent problem. It is not the least likely, it is the most
likely. And I would consider the report terribly incomplete if
you don't get into that as much as you are able to or if we
need other experts to look at that. I myself, when I saw the
ballot for the first time, I missed this slot. I went right to
the ballot and I missed it. Now, I know there are safeguards,
you can go back and it catches you and so forth. I am not
dismissing that. But I have been around this business a long
time and I totally missed that line. I am not saying that is
the cause. I am just saying that has to be examined along with
all the physical parameters. You do have to look at the human
factors, whether you have to get outside help or not. But that
could be a crucial part.
Now, if you find a lot of other errors, then you have the
real problem of deciding which it might be. But if you don't
find other problems, that is certainly one to look at. And I
would very much appreciate a detailed study of that. Again, I
don't know to what extent NIST has expertise on that either and
whether they could be of any help. But it is certainly an
issue.
Mr. Barkakati. And we probably have to--we at least can try
to find out expertise wise who we might be able to consult on
this. I would say that obviously we cannot probably complete
looking at the human factors within the remaining time. But as
a factor to be considered, that is certainly a possible way to
handle it.
Mr. Ehlers. I recognize full well some of this would be
almost impossible to determine. For example, one newspaper
article I read speculated that the undervote was because the
primary was so nasty and some people were angry and said I am
not going to vote for either one. Well, there is no way you
could even measure that. But I think there are a number of
factors that you can determine actually by experiment with
groups of people to see if they were factors or not. Thank you
very much. I yield back.
Mr. Gonzalez. Thank you very much, Mr. Ehlers. A couple of
real quick points. We had already approved pretty much what the
plan was going to be. And I know that GAO felt it was necessary
to go into certain areas to get the background information on
how the election was conducted in the preparation of the
machines, the instructions and so on. To the extent that human
error may come and play into it, the only thing I would caution
again is that we stay with the operating plan, number 1. Number
2, not to go beyond the scope of it, because the resources,
time and energy, of course, are precious to all of us.
And lastly, that the burden really does rest with the
contestant in this case. And on June 22, the representatives,
the attorneys for the contestant and the attorneys for the
contestee provided this task force with information they felt
was necessary for GAO to take into consideration. These are
recommendations, suggestions, some sort of guidance. To the
extent that you believe they are relevant, to the extent that
you would believe they needed to be addressed, that was going
to be basically left up to you. I think the whole focus from
the very beginning, whether it is the actual notice of contest
to the June 22 submissions by the attorneys, really truly focus
on the machines and the malfunctioning of the machines.
So that is the only caveat. And I understand what Mr.
Ehlers is saying. The only thing at this point, you know, how
relevant it is going to be for us to resolve this particular
issue predicated on the grounds as asserted by the contestant.
But I want to make sure, again, for both sides, the contestant
and the contestee, you have had for your review and such the
submissions by the attorneys representing the parties, is that
correct?
Mr. Barkakati. Yes. And I should quickly add that you are
right, our plan actually has been very specific to say that we
are focusing on whether the machine contributed. And you know
as much as in that scope of course it doesn't appear, the human
factors doesn't appear as an item. On the other question of
contestant and contestee submissions, yes, we do have access to
those and we have reviewed those. And in our deliberation of
deciding what is going to be done or not done, all those are
factors to take into account. That whatever the contestant or
contestee is suggesting, that testing be done or not done are
factors that come into play in deciding what the final answers
are for what it is that could be done to figure out if the
machines contributed to the undervote.
Mr. McCarthy. Will the gentleman yield?
Mr. Gonzalez. Thank you very much, Doctor. Sure, I will
recognize Mr. McCarthy.
Mr. McCarthy. The only thing I would say, I have a
transcript here from our first discussion when we were going
through what the GAO was going to do. And I raised the question
because I think I actually had a motion to put in ballot
design. And the response back was, from the GAO, the ballot
design is part of the machine. Obviously at this point, we are
going to look at that. So, I mean, I think that was just going
in regards to what Mr. Ehlers was saying, that that goes into
the whole GAO report as well. Just for clarification.
Mr. Barkakati. Yeah.
Mr. Gonzalez. All right. Anything further?
Ms. Lofgren. I have a motion, if we are ready.
Mr. Gonzalez. The Chair will recognize Ms. Lofgren for the
purpose of making a motion.
Ms. Lofgren. After consultation with Mr. McCarthy, I
believe this is acceptable to all of us. I move that the
chairman be authorized and directed to consult the task force
by teleconference or other appropriate means to consider any
GAO request received during the district work period and
determined by the chairman to require task force concurrence.
For the purpose of consultation as described in this motion all
members of the task force must be in simultaneous contact.
To preserve our open process, any consultation under this
motion will be made open to the public and press through
teleconference or web technology in the House Administration
Hearing Room. No final disposition of the underlying Florida
District 13 election will be made pursuant to this procedure.
Mr. Gonzalez. The motion has been made.
[The information follows:]
[GRAPHIC] [TIFF OMITTED] T8707A.010
Mr. Gonzalez. Any discussion? All right. All in favor
signify by saying aye. Aye. Any opposed? Around here you never
know. Some people vote yes before they vote no and so on. The
task force meeting stands adjourned. Thank you.
[Whereupon, at 10:42 a.m., the task force was adjourned.]