Liars, Damned Liars,
and Polygraphers
In the mid-1990s, I became interested in the science (or lack thereof)
behind polygraphs as one of
my closest colleagues at Sandia (and a very senior member of the
technical staff) lost his opportunity for a key appointment to a
high-level advisory position in the intelligence community as a result
of flunking his polygraph. Because of my background in medical
decision analysis (indeed, the polygraph is a "diagnostic" tool,
although almost completely without diagnostic value), he asked me to
look into the actual utility of polygraphs in identifying spies within
the national security context.
At that time, there were very few published, peer-reviewed scientific
summaries of the polygraph, and it rapidly became clear that
polygraphers were -- as a group -- nothing more than interrogators who
used the polygraph machine as a ruse to convince the unsuspecting
interviewee that their machine could devine deception from
truth-telling. This is almost completely false; there is some
modest evidence that in specific
incident investigations (such as the investigation of a murder,
robbery, or other highly unique event) that the polygraph has
diagnosticity greater than chance. However, in the setting of
national security screening, the polygraph is probably worse than
worthless. That is: the polygraph is much more likely to falsely
"diagnose" deception than it is to identify true deception.
In 1999, in the wake of the Wen Ho Lee scandal at Los Alamos National
Laboratories, then Department of Energy Secretary Bill Richardson
(along with several members of the Senate and House Intelligence
Committees) decided to institute a wide-ranging polygraph program at
the National Laboratories involved in classified or nuclear-weapons
related work. Some 15 - 20,000 employees at Sandia, Los Alamos
and Lawrence Livermore National Laboratories were suddenly subject to
polygraphs on a regular basis (initially intended as every 5
years). It was evident to just about all of the technical staff
members at these labs who thought the proposal through that polygraphs
would do more harm than good. In 1999, Sandia's president C. Paul
Robinson asked his senior scientists (the highest-ranking 40 scientists
out of the 6,000 members of the technical staff) to review the
polygraph literature. The report
concluded
that polygraphs were likely to do much more harm than good, not only
because of the high false positive (and high false negative!) rate of
the test, but because investing tens of millions of dollars in a
polygraph program would squander resources that could be better used
elsewhere to bolster flagging security at the Labs. The work was
ignored, including by Robinson who ordered it in the first place.
The DOE held hearings at each of the Labs in 1999 to get comments from
technical staff and the public. Most laboratory staff believe
that their substantive suggestions -- for example, to have independent
oversight to avoid abuse -- were completely ignored. Indeed, Paul
Robinson himself ignored his own senior scientific staff's request for
oversight.
Unsurprisingly, numerous abuses were documented shortly after the
polygraph. As the only Senior Scientist at Sandia who had also
practiced medicine, staff members came to me with their stories,
fearing retribution from Sandia and DOE management. These
testimonials were submitted on a half-dozen occasions to Paul Robinson,
the Secretaries of Energy (Richardson and Abraham) with requests for
simple independent oversight of the polygraph session, since each was
videotaped (allegedly for "quality assuarance" purposes). These
requests were completely ignored.
In 2001, one Sandia staff member complained to me that he was being
asked medical questions as part of his polygraph examination -- clearly
a violation of DOE policy and promises made directly by Secretary
Richardson to staff members at each of the Labs when he visited in 1999
and 2000. I forwarded the complaints to Sandia's Chief Medical
Officer, Dr. Larry Clevenger. Dr. Clevenger investigated the
allegations, and found them to be true. On March 6, 2001 he
advised
in a letter
to Sandia's Executive Vice-President Joan Woodard that there were other
abuses as well, and he recommended that for the health and safety of
individuals that the polygraph program be temporarily suspended while
carrying out a full review. Ms. Woodward tasked a Sandia
Vice-President to meet with the polygraphers regarding the medical
questions, and I attended that meeting. DOE agreed
to stop asking medical questions; but, when I asked for independent
oversight so that there might be some assurance that this promise would
be kept, I was turned down yet again.
When I inquired of Dr. Clevenger if he had followed-up with Paul
Robinson on any of his other recommendations, he informed me
that indeed he had. However, Paul Robinson never acted any
further. It was at this point in time that it became clear that
Sandia's president was failing to take even the most basic steps to
protect the safety and careers of employees -- his highest charge as a
DOE National Laboratory President. Ironically, Sandia’s own
newsletter has, from
the time then Department of Energy Secretary Bill Richardson
proposed them in 1999 until the present carried many articles
decrying the foolishness of polygraphs as a counterintelligence or
security tool. Even though Robinson is a trained scientist, he
never acted on the recommendations of the National
Academy of Science to abandon – or at least monitor – the use of
polygraphs at Sandia.
It was clear that reason and logic had failed, and in the one place in
government where that isn't supposed to happen: the "independent"
National Laboratories. Without putting too fine of a point on it,
as a physician I took my Hippocratic oath seriously. Since
scientific studies, documented abuse of Sandia (and other
laboratories') staff and illegal inquiry by polygraphers failed to
motivate Robinson, his staff, or anyone at the DOE to impose oversight
on the polygraphers, I had only one choice left: to go to the
media. Here are some of the pieces I wrote until my forced
resignation from Sandia in July of 2003.
- "False
Detector" -- Washington Post and Albuquerque Tribune, April 2001
(also published in The Skeptical Inquirer in July 2001)
- "Polygraph
Hypocrisy" -- Washington Post, August 2002. I wrote this this
OpEd after discovering that Sen. Richard Shelby (then chairman of the
Senate Intelligence Committee) had refused to take a polygraph as part
of the FBI's investigation of security breaches within the Intel
Committee. Sen. Shelby's committee was suspected as the source of
a leak of a National Security Administration (NSA) intercept of a
telephone conversation pre-saging the 9/11 disaster. Ironically,
Sen. Shelby was one of the architects behind the the DOE laboratory
polygraph program. It is hard to find a clearer example of
hypocrisy and self-serving political behavior than this.
- "A
Lie Detector Failure" -- Albuquerque Journal, Oct. 2002, commenting
on Bill Richardson's own lying with regard to the DOE polygraph program
and his refusal to accept the results of the National Academy of
Science's study on polygraphs (see below). Richardson is now
governor of New Mexico, and is widely touted as a Vice-Presidential or
even Presidential candidate. Not a good thing.
- "Polygraphs:
Worth than Worthless" -- Washington Post, May 2003, reviewing the
findings of the National Academy of Sciences and the failure of
polygraphs in yet another spy scandal, this time involving an alleged
Chinese agent operating in the United States.
The Albuquerque media, the national media and prestigious scientific
journals as well also opined on the DOE polygraph program (but to no
avail). Here is a selection of their editorials:
A recent study by the National Academy of Science (http://www.nas.edu)
--
summarized here by the Washington Post -- concluded that "the
polygraph has not place in any
Federal Government agency". This, too, has been ignored.
Thus, polygraphs remain as a major (and perhaps the major) counter-intelligence
tool in the Federal Government. In so doing, our national
government implicitly gives credence to this now discredited security
firewall, and perhaps encourages members of the legal community --
attorneys and defendants -- to rely on its results. I believe it
is instructive to read the comments of Aldrich Ames -- formerly
employees by the CIA where he became the most damaging spy in post-WWII
US history -- when he commented on my article in the Washington Post
and Skeptical Inquirer:
"Dr.
Zelicof’s (sic) discussion of polygraph junk science was a very useful
one, and its highlight is the apparent determination he and others have
to resist the destructive spread of that superstition. His
uncontroversial point that “protecting secrets is a challenging task”
suggests to me another point worth making. In my experience with
the polygraph, as user and subject, its junk science does provide an
important but discreditable service for lazy and timid national
security managers (also known as a species of bureaucrat).
Decisions about personnel suitability for sensitive positions can be
not only difficult for the unusual and proper reasons, but also quite
risky for the careers of decision-makers. There’s a lot at stake
for the bureaucrat. Faced with the prospect of excruciating hard
work, considerable expense and agonizingly difficult choices, the box
offers an attractive refuge from responsibility. Like handing
fate to the starts, entrails or the rack, bureaucrats can abandon their
duties and responsibilities to junk scientist and interrogators
masquerading as technicians.
...the new Rule 702 nails down the in admissibility of polygraphic
superstition even more firmly. Perhaps Dr. Zelicof or others (the
ACLU among them) might take the issue further. When people’s
livelihoods, reputations and personal lives are injured or even
destroyed by junk science, cannot the law protect them?"
It doesn't get any clearer than than, coming as it does from a high
level CIA officer who passed his polygraph every five years during his
2 decade career.
I am available as an expert witness for individuals who believe they
have been incorrectly categorized as "deceptive" by a polygrapher and
the unreliable box. Feel free to contact me for more information.
Addendum
#1:
The final demise of the polygraph at the DOE Labs?
After approximately seven years, the DOE
quietly slipped in a
new polygraph policy into the Code of Federal Regulations at the end of
October 2006 (10 CFR
Parts 709 and 710). The rule eliminated routine screening
polygraphs, but replaces it with an ill-defined set of "random"
polygraphs based on a loose interpretation of the National Academy of
Sciences report of 2001 on the possible
(but unproved) detterrence value of polygraphs (the Academy recommended
more research to support or refute this contention). Nonetheless,
it may be the case that a small step forward (which is to say a step
backward for polygraphs at the Labs by rescinding their use on a
regular basis for most scientists with high-level security clearances)
has taken place. As with so much of DOE policy (see commentaries
from the Editors of leading New Mexico newspapers above on this page),
sloppy execution, inattention to detail and -- most significantly --
the complete absence of independent oversight to prevent documented
previous abuses remain substantive concerns that will doubtless
re-emerge.
There are at least two important lessons to be drawn from the sad
history of polygraphs at the Labs which, aside from undermining
national security (as demonstrated by the yawning gaps in cybersecurity
still present -- ones that Sandia's Senior Scientists advised be
addressed first and foremost in 1999 during the start of the most
recent security problems at the Labs). First, it is clear that
when scientists (particularly those in government) fail to speak out,
bad policy inevitably follows. I believe that the history of
polygraphs at the national labs shows that laboratory management is
unwilling to do much to protect the health and safety of their
employees. All of the directors of the three main DOE labs
(Sandia, Los Alamos and Lawrence Livermore) simply waited to see which
way the political winds would blow. To this day, we have no idea
how many careers were destroyed as a result of a thoroughly disproved
“deception” test. Scientists might well wonder if the labs remain
institutions of serious science, integrity and scientific
freedom.
Second, the elected official in
whom we place our trust, our tax dollars, and occasionally our lives
never admit to error (nor do they even have the fundamental decency to
apologize to those loyal scientists who have been harmed, some
severely). On the grand scale, they are both smart and cynical
enough to realize that the average voter won't notice or care - which
in my view is but one more argument for scientists to carefully educate
the public (and perhaps those few political leaders who are open to
learning) based on solid science, explained in a straightforward way at
invited talks, in OpEd pieces, and public policy forums.
Two
recent OpEd piece of mine follow in the hope of inspiring others to do
the same.
Albuquerque
Tribune, Nov. 2006: The difficult, but important responsibilities
of scientists in public policy
Albuquerque
Journal, Oct. 2006:
DOE finally quiety abandons routine polygraphs without explanation (or
apology)
Addendum
#2: Hold the presses: The Polygraph is Back at the DOE labs
The DOE's "screening" polygraph program has
been re-incarnated as of May 2007 into what is now referred to as
"random" polygraphs, meaning the following:
- Approximately 9,000 employees at Los Alamos and
Sandia National Laboratories (plus
an unknown number at Lawrence Livermore National Labs as of this
writing) with high-level security clearances (generally associated with
direct nuclear weapons work) are subject, at any time, to a
polygraph. Since the program is "random", it is theoretically
possible that one individual could have many polygraphs in any given
time period whereas the old polygraph program (reveiewed in addendum #1
above) called for administration of a polygraph every 5 years to anyone
with a security clearance (who may or may not have work assignments
involving nuclear weapons or other highly classified programs.)
- It is unclear if the format of the "random"
polygraph is the same as that applied in the old "screening"
program. Will the questions be limited to national security or
will lifestyle (e.g. sexual practices) be queried as well?
- In the "screening" polygraph program, it was
policy (perhaps carried out, perhaps not) that failure to pass a
polygraph did not mean immediate loss of one's "Q"
clearance (so-called "Top Secret" clearance which provided access on a
need-to-know basis to a wide variety of classified documents.) In
the new "random" polygraph program, failure to pass a polygraph does lead
to immediate loss of high-level clearances (but not the
"Q"-clearance). Since, for most people with higher level or
special clearances loss of those clearances is loss of access to
docments needed for work, there is a probability of actually losing
employment as well (although none of the contractor managers will, as
of this writing (5/17/07) say what that probability is.)
- Further, and perhaps most terrifying is the
statement of the Security Director at LANL, who writes:
"If informed
to
present themselves for a polygraph, they are to understand
that
participation is mandatory. Should they fail to pass the test,
they will be removed from the pertinent category list and will
not be
allowed to participate in any of the other categories. If
they choose to object now to the
implied participation in the
polygraph program, they will immediately
become ineligible for
any of the accesses or programs listed
(note added: these would be the higher level clearances). They will be
informed
that failure to participate in the polygraph program by
itself
will not affect their basic security clearance.
" (note
added: "basic" security clearance is the "Q"-clearance.)
Put
another way, even speaking about
the polygraph program in critical terms may lead to loss of high-level
clearances.
It is hard to imagine that this kind of culture of fear will encourage
new, talented people to enter the labs. It also would seem that
egress from LANL and other labs may be further encouraged.
But perhaps as sad as the Orwellian intonations from the
counter-intelligence director at LANL is the abject ignorance of the
science – now widely published – on not merely the lack of efficacy of
the polygraph, but in addition its erosive effect on national
security. A detailed analysis is available here
which shows, in summary:
• When the
inconclusive results are ignored, the negative predictive value (NPV)
of the polygraph is 97% (92% - 100%). Although inconclusive results are
a fact of life (and thus difficult to in fact ignore), within this data
set an individual who passes a polygraph is almost certainly not being
deceptive.
• When inconclusive results are considered to be errors of ambiguity,
the NPV of the polygraph falls to 73% (62.5% - 78.1%). Thus, the
polygraph is better than flipping a coin, but would juries or agencies
that grant security clearances want to rely on it when a deceptive
individual is able to pass 27% (21.9% - 37.5%) of the time? I’m glad I
don’t have to make that decision .
• The positive predictive value (PPV) of the polygraph is 88% (82% -
87%) when inconclusive results are ignored.
• The PPV of the polygraph falls dramatically to 55.5% (45% - 60%) when
inconclusive results are accounted for.
In summary, if a subject fails a polygraph, the probability that she
is, in fact, being deceptive is little more than chance alone; that is,
one could flip a coin and get virtually the same.
One can only conclude that the US government's premier science labs
are, in fact, anti-science.
[Webpage last updated
5/17/07]
|
