Thank you for that kind
introduction. I am honored to be your luncheon
speaker this month, especially during Black History Month, because I have a
historic family connection in aviation.
My great uncle, James Herman Banning, was the first African American to
receive a pilot’s license from the U.S. government. That occurred in 1926, which was the first year
the Commerce Department began issuing pilot’s licenses. Two other African Americans received their
pilot’s licenses prior to that – Emory Malick in 1912, and Bessie Coleman in 1921
– but their licenses were issued by the Federation Aeronautique.
Today I would like to discuss two
things. First, some process improvements
that I intend to pursue at the NTSB. The
NTSB is an amazing place to be – not because of anything I did, but because it
is populated with world-class experts who love what they do, who are passionate
about it, and who do it very well. The
investigation process that they follow every day has been working well for many
years – but there is always room for improvement.
Second, I would like briefly to touch
upon two areas of concern regarding future aviation safety – fatigue and increasing
First, some NTSB Basics. The NTSB was
created by Congress to be an independent federal agency that investigates
accidents in all modes of transportation.
The purpose of our investigations is to determine the probable causes of
the accidents and then make recommendations to prevent those accidents from
occurring again. Our primary product is recommendations;
we are not a regulator, and we cannot require anyone to do anything. Notwithstanding our inability to require,
however, the recipients of our recommendations do what we recommend, or some
facsimile thereof, more than 80% of the time.
That is why it is such an honor and a privilege for me to be Acting
Chairman because this high recommendation acceptance rate is a reflection of
the high quality of our accident investigations, as well as the high quality of
the analysis from those investigations that leads to our recommendations.
I am the Acting Chairman of the NTSB,
but I have told my staff that while I am Acting, I do not intend simply to
maintain the status quo. Instead, I will
be expecting continuous improvement, and I will be expecting improved
collaboration, both internal and external, to drive that continuous improvement. External collaboration has historically been
a challenge for us because our independence is so important to the quality of
our investigations that we sometimes isolate ourselves too much from the
industries we investigate. That can
cause us to “lose touch” with how these industries operate, which in turn undercuts
the accuracy and validity of our investigations.
I am a strong believer in the power
of collaboration to produce continuous improvement because of the amazing safety
improvement that this industry has enjoyed because of its collaborative process
known as CAST, the Commercial Aviation Safety Team. The moral of the CAST process is very
simple: Everyone who is involved in the
problem – in the case, further improving a safety record that is already very
good – should be involved in developing the solution.
Regarding our investigative process
improvement, I have tasked my staff to look at how they would design our
investigation process if they invented it from scratch – not a tweaking of the
existing process, but the creation of a new process. I am looking for that new process to reflect
two realities that greatly affect how we do business – the amazing advances in information
technology, and the increasing internationalization of accident investigations.
Reflecting the external collaboration
that I mentioned, my staff has been reaching out to many of you for your
suggestions regarding our process improvement, and I would like to thank those
of you who have volunteered your time and effort to help us by providing your
thoughts, ideas, and suggestions.
Two major IT improvements have significantly
changed the way we do business. First,
many video sources provide very helpful information that is often not otherwise
obtainable, which helps our investigations be both more efficient and more
accurate. In the Asiana crash in San
Francisco in 2013, for example, a video of the accident sequence was seen by
the public on YouTube even before we were formally notified that the accident
had occurred. In the Reno Air Races in
2011, many spectators provided us video that enabled us to calculate that the
P-51 Mustang that crashed underwent a 17-g pull-up, which rendered the pilot
unconscious. Very few g-meters go beyond
6-8 g’s, so this video provided information that was otherwise
unobtainable. In the mid-air collision
over the Husdon River in 2009, video provided by viewers on the ground who
happened to see the accident helped us determine the angle of convergence of
the two aircraft – also information that might otherwise be unobtainable. Last, but not least, at the spacecraft
accident in Mojave, CA, in 2014, in-cockpit video enabled us to see immediately
that one of the pilots moved a critical lever from lock to unlock at a time
other than that specified in the checklist.
Second, increased ability to collect
and analyze data has resulted in large quantities of data from flight data
recorders. The ASIAS database, for
example, contains data from millions of flights. This source of information greatly increases
our ability to conduct investigations that are more focused and more
accurate. Conversely, our accident
experience can inform the collection and analysis of data by ASIAS to make it more
responsive to the mishaps that are occurring.
We certainly understand the
reluctance of industry to give us information from ASIAS, given how important
transparency is to our process, but we are very careful not to disclose data we
receive from ASIAS for two reasons.
First, because it’s the law; and second, even if it weren’t the law, we
know how valuable ASIAS information is, and we’re the last ones to want to
“kill the goose that laid the golden egg.”
Kudos to ASIAS for their persistence in working with us to develop procedures
that respond to the need not to disclose in conjunction with the importance to
us of transparency.
Finally, our investigations are only
as useful as our ability to get our analysis and conclusions out to the
industry, for their use in improving safety, and information technology
advances have also helped with the process of dissemination. That starts with our ability to get on-scene
photos out to the public immediately, continues with our placing our on-site
briefings on YouTube and our website, and concludes with our exploring the use
of video alerts and even video reports of our analysis and conclusions.
Increasing internationalization has
become an issue because very few future accidents will be entirely domestic,
yet our domestic investigation processes are not the same as the investigation
processes prescribed by ICAO for international accidents. One difference relates to who sees what
when. The ICAO process includes giving
the report to the parties for comment before it is final, while our domestic
process does not allow anyone to see our analysis, conclusions, or probable
cause before they are final. Examples of
the disconnect that this process difference creates are:
landing on the Hudson River, in 2009, in which Airbus saw the report before it
was final, because Airbus was a technical advisor to the BEA, which was an
accredited representative under ICAO protocol, but under our domestic
procedures, US Airways did not see the report before we issued it as final;
Birmingham UPS accident, in which Airbus saw the report before it was final but
UPS did not; and
Boeing 787 battery investigation, in which Boeing’s battery contractors, based
in France and Japan, saw the report before it was final, but Boeing did not.
I am confident in the very high
quality of our reports, but I would like to explore the ICAO process, which is
roughly comparable to a peer review process, because I believe that peer review
can make a high-quality report even better.
The concern is that the perception of our independence would be
undermined if we changed the report because of input from the parties, so our
challenge is how to have peer review while maintaining public confidence that
we are independent.
I look forward to results of this investigative
process improvement effort, and again I would like to thank those of you who
are helping in response to our outreach.
Now I would like to turn to two areas
of concern that I have regarding future aviation safety – fatigue and increasing
Fatigue is a problem in all of the transportation
modes we investigate, for the simple reason that there is a basic
disconnect: Commercial transportation is
24/7, but humans are not. The problem is
that, while sleep science is well developed, fatigue science is not so well
developed. Sleep, to be sure, is a major
part of fatigue, but what about other factors, such as diet, exercise, drinking
habits, smoking habits, age, and other factors that might affect fatigue? What
about the effects of shift changes, or in this industry, of crossing several
time zones in a short period of time? In
the instances in which the pilots know that they are in danger before they
crash, what effects, if any, does the “adrenaline rush” have on undoing decrements
in performance caused by fatigue?
Add to all of this that fatigue is
difficult to measure. How can a person’s
fatigue level be measured when they arrive at work? What about the decrements in performance that
result from increasing fatigue while they are at work? And how can we, as accident investigators,
measure post-accident fatigue?
Unfortunately, not only is it difficult to measure someone’s fatigue,
but experience has demonstrated that self-diagnosis of fatigue is also not reliable.
Last but not least, rest and duty
time rules cannot control what employees do on their own time. In the recent Walmart accident in New Jersey,
the media reported Walmart as saying that the truck driver was in compliance
with the rest and duty time rules, while the driver was saying that he had been
awake for the previous 24 hours. Our
investigation will determine the veracity of those assertions, but what if both
Hence, the challenge is how to ensure
employees arrive ready to work, and how to ensure that they remain ready for
the duration of their duty time.
Added to this is another emerging
issue – Americans are gaining weight.
Research has demonstrated a strong association of Obstructive Sleep
Apnea, or OSA, with weight, and if a person has a Body Mass Index of 40, they
are 50% likely to have OSA.
Unfortunately, OSA, which can result in daytime sleepiness and thus be a
major transportation safety issue, often goes undiagnosed. If someone sleeps 8 hours and believes that
should be adequate, but is still sleepy at work, their OSA may not be diagnosed
unless they ask their doctor or, absent that, the doctor explores it
further. We just investigated a
passenger train derailment in the Bronx in which the train approached a 30mph
curve at 82mph, and our investigators concluded that the train operator had
undiagnosed OSA and had fallen asleep.
We hope our accident investigation
experience will help inform this debate about fatigue.
Another concern is the loss of skills
and proficiency that can result from increasing automation. A classic accident of this type was the crash
of Air France Flight 447 into the Atlantic Ocean in 2009 while it was en route
from Rio de Janeiro to Paris, when ice on the pitot tubes caused a loss of
airspeed information. The automation was
very complicated, and not fully understood by the pilots, and they had never
encountered a loss of airspeed information during cruise flight before, even in
training. In addition, the loss of
airspeed information rendered both the autopilot and the autothrottle inoperative,
among other systems, so the pilots had to fly the airplane manually, but they
had never flown it manually before at that altitude, even in training. Given that it was night, in clouds, in
turbulence, this situation overwhelmed them and they crashed. Unfortunately, this manifests a problem that
is industry-wide, and not just limited to these pilots having a bad night.
That is an example of the problems
that can occur when something goes wrong with the automation. The other side of the coin is that there can
be too much of a good thing – can too much automation, when it is working
properly, undermine professionalism? An
example of this is many subway systems, in which the objective is to make the
system as automatic as possible. Hence,
the automation takes the train from the station, maintains appropriate speeds,
maintains adequate distance from other trains, stops at the right place in the
station, and then opens the doors. The
operator performs only one function – closing the doors.
When the operator’s only function is
to close the doors because everything else is automatic, would that operator
love his or her work and enjoy the pride of accomplishment, or would he or she just
be there to get a paycheck? If the
paycheck is the primary objective, what does that do to professionalism? This problem is particularly difficult to
remedy because, as noted above, it occurs when the automation is performing correctly.
The bottom line is that automation is
very beneficial, and it has a demonstrated history of improving safety,
Unfortunately, however, the industry still has a way to go to achieve a better
understanding of the human/automation interface.
We also hope to inform that safety
improvement issue with our accident investigation experience.
In conclusion, it’s been said that safety
is a continuing journey, rather than a destination. Likewise, as good as the NTSB is, there is always
room for improvement, and I am very excited to have the opportunity to help
foster that improvement.
Thanks again for inviting me as your
speaker, and I look forward to working with all of you in our common goal of
further improving a very safe aviation system.