Re: A320 cockpit visit)

From:         Robert Dorsett <rdd@cactus.org>
Date:         08 Jul 93 01:27:46 PDT
References:   1 2 3
View raw article
  or MIME structure

In article  <airliners.1993.493@ohare.Chicago.COM> pab@po.CWRU.Edu (Pete Babic) once wrote:

>I'm a total layman when it comes to piloting but I've been following the A320
>views with great interest. If the cocoon approach continues to be used would
>it make sense to setup some kind of logging mechanism in the aircraft to 
>keep track of the times when the system "saves the pilots bacon"? If a pilot
>uses the automation to let him get away with sloppy flying this would show up
>in the log resulting in disiplinary action from his supervisors. This would
>serve as incentive to not push the envelope too much.

The technology already exists to do this, and other interesting things,
like videotaping the crews and recording the voice-track of entire flights
(not just the last 30 minutes).  

The major problem is the ramification this has on the pilot's decision-making 
authority.  By legislative fiat, the pilot is the final authority as to
the safe and responsible operation of the aircraft.  Thus, once the doors
are closed, the pilot is King.  If he crashes, he's liable.  If he accepts 
an airplane with a known squawk, he's liable.  In fact, the notion of 
responsibility is so severe that he's theoretically liable for regulatory 
omissions that could be buried deep within a modern airline's chain of 
responsibility.  It's therefore a position of absolute responsibility, 
both for the airplanes and the lives on board (in fact, one rhetorical
line against Airbus' FBW design is the extent to which it limits pilot 
*authority*).

Despite a captain's demi-god status, airlines (and the feds) regulate and 
provide guidelines for a truly awesome number of operational issues.  
Everything from which hand one should use to set a heading selector to how 
the pilot should be dressed when he goes to the bathroom (e.g., hat and
coat vs. "merely" shirt and tie).

Like all "rules," these are best viewed as guidelines.  It is occasionally 
desirable and necessary to deviate from the guidelines.  For example, a 
pilot may not wish to depart with one generator inoperative, even if that's 
kosher with the minimum equipment list, if he has doubts about a remaining 
generator: this could, for instance, be based in a large number of "squawks" 
in the maintenance logs for the remaining generator.  The airline, however, 
may hold to the letter of the rules (cf. Eastern and Continental under 
Lorenzo) and discipline the pilot for not accepting the airplane.  In such a 
case, the airline would be within its rights, but the pilot, under FAR 91.3, 
is also legally and morally obligated to act.  What results is a conflict of 
priorities which usually place the pilot at a disadvantage when it comes to 
arguing his case.  Most airlines are run professionally, but the regulatory 
framework can support some startling excesses, when conditions are ripe 
(again, cf. the reign of terror of Lorenzo).

With full-time monitoring, the pilot, knowing that every action is being 
followed and is subject to review, might be more likely to "toe the company 
line," or follow the rules to the letter--even if that raises the probability 
of flying into cumulus granite.

Take, for example, the early FMS experience.  The airlines had bought into
the technology, and required the crews to use the FMS.  This resulted in 
generally unsafe conditions under 18,000': as time went by, more and more
pilots opted to "click it off" in the terminal environment.  Initially, these
people were characterized as not being proficient or as "fogies" unable to 
cope with the high tech.  But research and experience validated the click-
it-off philosophy, which eventually got absorbed as part of most airlines' 
operational policies.  These days, FMS use is generally discouraged under 
10,000' (often under 18,000').  But at the time, a pilot flying an airplane 
the "old fashioned way," selected-autopilot, would have been censured.

Or consider in-flight deviations: if the dispatcher's instantaneous satellite 
picture from ten minutes ago shows moderate weather, but the pilot sees a 
monster thunderstorm evolving ahead of of him, it becomes problematic 
defending the decision to adopt a potentially costly re-routing, rather than 
continuing with the straight-line course.  Especially if the next satellite
pict shows clear weather, after the cells break down. :-)

So a major issue is that there's an awesome spectrum of second-guessing
that will go on, both after and during the flight (with telemetry).  A 
conscious decision has to be made by all parties (regulators, airlines,
manufacturers, pilots, ATC) as to *how* this will work: so far, nobody has 
adjusted their operating policies significantly--but there's already a 
spectre of the possibility that with real-time monitoring, "unpopular" 
decisions can have serious consequences to one's livelihood.  

Yet, despite the most "blue sky" theories, and the encroachment on pilot
authority, the regulatory authorities maintain the sanctity of 91.3, out
of the simple realization that "shit happens," and that carefully analyzed
*systems* don't always work as planned in the real world.  My point is 
simply that we *cannot* plan for "business as usual" and expect to maintain
a halfway decent safety culture.


Other, related problems with monitoring are that the pilots are only human: 
should *every* utterance in-flight go on the record?  More likely, they'd 
learn to keep their mouths shut, which would create a tense, stressful 
atmosphere, which isn't conducive to safety, either.  In high-workload 
environments, it's best to keep to the job at hand (the "sterile cockpit" 
concept), but 12 hours at cruise is hardly high-workload, and having Big 
Brother listen in isn't fun.  And what if, in these politically correct 
times, "inappropriate" views are held against someone?  There isn't a single 
profession where the potential for such an extreme invasion of privacy 
exists-- even IF it's on the company's time.


ALPA's drawn some very firm lines on these issues: as satellite commun-
ications evolves and the potential for in-flight monitoring increases, 
ALPA's been very aggressive about preserving the pilot's command prerogatives.
Access to cockpit voice and data recorders are also tightly regulated, as are 
tapes made during training (CRM training, for example, is videotaped and 
analyzed by the students in a classroom session--then the footage is promptly 
destroyed).  There is no doubt but that more sophisticated monitoring 
capabilities will come, and rightfully so, as they will be critical in
crash investigation--but they shouldn't be used as a management tool.


The airlines are paying pilots enough and training them to the degree that
they should let them do their job.  The deepest bureaurcratic instinct,
however, is to distribute responsibility as widely as possible.  The word 
"supervisor" is a hallmark of this philosophy (pilot-populations within an 
airline tend to operate on a peer basis, and are subject to the oversight of 
fleet captains and chief pilots).  It's something to be aware of, and fought, 
IMHO.  Some passengers don't like the idea of their lives being in the
hands of two or three individuals in the front, and would prefer that people
on the ground be involved: I, however can think of nothing more harrowing.
At least I know that if I buy it, the crew is toast, too, and will do its
best to avoid that fate.  A ground-based dispatcher doesn't face death if
he makes a boo-boo.


On the OTHER hand, it would certainly be interesting to run a study using
the avionics now available, to determine just how many "marginal" situations
pilots (as a population) manage to get into.  Airbus' rationale for its
protections on the A320, for instance, claimed support from the "accident"
record.  But I'm really unable to duplicate this reasoning: what kills
most airplanes is controlled flight into terrain--not g-loaded stalls,
overspeed break-ups, etc.


More rambling, from...

---
Robert Dorsett
rdd@cactus.org
...cs.utexas.edu!cactus.org!rdd