[conspire] Wikipedia editing suckitude; the art of checklists

[Rick Moen]

Remember: a well-debugged checklist will save your life.

----- Forwarded message from Rick Moen <rick at linuxmafia.com> -----

Date: Sat, 15 Apr 2017 23:50:49 -0700
From: Rick Moen <rick at linuxmafia.com>
To: Kevin W Enns <kwenns at gmail.com>
Subject: Re: [skeptic] And see, this is why he makes the big bucks...
Organization: If you lived here, you'd be $HOME already.

I wrote:

> BTW, the things you learn from reading NTSB disaster reports!  I'd
> somehow assumed Dad had been a longtime captain, but the disaster report
> on the crash that killed him in December 1968 said he'd been rated for
> full captain status only in June 1967, in the middle of our stay in
> Victoria, Hong Kong.

Around the time I was writing that, I was doing a low-key corrective
edit of this Wikipedia passage:
https://en.wikipedia.org/wiki/Elmendorf_Air_Force_Base#Aviation_accidents

Over the years, there have been several mentions on Wikipedia of the Pan
Am flight 799 disaster that killed my father, and they've all had the
same appalling problem:  The clumsy editing and writing always ends up 
blaming the pilots:  The Wikipedians always boil it down to 'Crew
attempted a no-flaps takeoff', which makes them sound like a trio of
idiots who incompetently killed themselves and a $5M airframe.

And they do this while linking directly to the NTSB report, which means
they _don't fscking read it_.  The report goes out of its way to say it
wasn't the pilots' fault.[1]

This is a sore point with me and with anyone who grew up in an airline
family, because it's notorious that the airline, the airframe
manufacturer, and even the press tends to _immediately_ jump the gun
after almost any fatal air crash and _blame the pilots_.  'Must have been
pilot error', we hear.  Why _not_ blame them?  They're dead and can't sue,
so who cares if everyone casually defames them?

_So_ tired of that.

I have to be extremely low-key about my Wikipedia editing, though,
because if my personal connection is known or suspected, there could be
a huge backlash.  The justification for this is their mania for NPOV
(neutral point of view), but functionally (IMO) it's hauled out by
Wikipedia regulars mostly as a passive-aggressive way to fend off
non-regulars with inconvenient expertise.

So, I edit without logging in, and try hard not to get the regulars'
backs up.

Around late 2004, a Wikipedia article about my friend Eric S. Raymond,
a rather flamboyant (and, to be fair, self-promoting) member of the open
source community, had been turned into a rather appalling hatchet job
pretty much devoted to attacking him personally, citing as the 'neutral'
source an extremely dubious, scurrilous free-hosted Web page
(http://esr.1accesshost.com/) that for some years had been maintained
by anonymous parties who dislike Eric in order to ridicule
him.  Eric made the mistake of directly attempting to edit and correct
the Wikipedia page under his own name as the editing login -- and
immediately ran into a buzzsaw of Wikipedians saying he was not allowed
to make the article cease sucking, because he wasn't 'neutral'.  So, his
edits were reverted(!), even though he amply explained them.  At that
time, in particular, the Wikipedians would rather reject corrections
from an expert and keep a terrible page if the expert knew too much
because, y'know, nobody knows more about that particular subject, being
a primary source, but isn't 'neutral' so we mustn't listen.

I drew the appropriate lessons, and, not long afterwards when someone
created a page about me, rather than try to improve it, I invoked 
Wikipedia's non-urgent process for deleting the page, which got done
about a month later.

Ironically, what ended use of the http://esr.1accesshost.com/ page as a
primary source was it being cited by one of its proponents, one Jim
Thompson, on my 'Conspire' Linux user group mailing list in California,
in reply to which I debunked the factual claims cited from that page,
even to Thompson's satisfaction.  Afterwards, Wikipedians were able to
reference my mailing list discussion as a reliable external authority to
fix the page about Eric.  So, they were willing to accept better
information about Eric from me, but not from Eric.  Weird.

(Since that time, Wikipedia has mostly cleaned up its act, getting
serious about enforcing its Biography of Living Persons policy and
disallowing use of Wikipedia for personal hatchet jobs.)


There is also at least one other Wikipedia page that includes the Pan Am
crash, that I _also_ had to delicately fix some years ago -- same
serious problem -- but I can't find it at the moment.


Here's a weird thing:
https://en.wikipedia.org/wiki/List_of_accidents_and_incidents_involving_the_Boeing_707#1960s
and
https://en.wikipedia.org/wiki/List_of_accidents_and_incidents_involving_commercial_aircraft#1968
have nothing about the crash.  Can you guess why?

It's because _even though_ those purport to be comprehensive lists, no
additions are permitted (to flesh out the lists) unless each links to 
a full Wikipedia page about the disaster.  Because nobody has yet made a
page about Pan Am flight 799, it remains nearly absent.

Of course, I _could_ create a
https://en.wikipedia.org/wiki/Pan_Am_Flight_799 page from scratch and 
make it scrupulously match the NTSB report and several classic books 
that extensively cover it as a key case study.  I'm almost willing to try.
That would be a fairly high-profile edit, however, and might attract
attention.  (Articles that are too substantive also start getting
complaints about violations of the 'No original research' rule, even if
scrupulously referenced.)



[1] Among the many things you'll never learn from the lamentably bad
discussion of this crash most places on the Web (including Wikipedia) is
that the flight 799 disaster (eventually) brought about a revolution in
human-factors research for both the airline industry and others, and
forced rethinking about the way checklists are used.  In a real way, the
sad lessons learned from studying this crash have saved countless lives
(later).  (But it took more checklist-related deaths and 18 more years
before NTSB prompted FAA to convene a research group to improve
checklist design.) 
See:  https://ti.arc.nasa.gov/m/profile/adegani/Cockpit%20Checklists.pdf
http://lessonslearned.faa.gov/Northwest255/Review%20of%20Takeoff%20Configuration%20Warning%20Systems%20on%20Large%20Jet%20Transports.pdf

In its mid-1969 report on the crash, NTSB identified as critical
causative factors (a) a fatal flaw in the preflight checklists that
interacted catastrophically with (b) an unfixed hardware defect in a
crucial warning system.  The warning system was a horn designed to sound
if the pilots apply takeoff thrust and any of several control structures
including the flaps aren't set right for takeoff.

First Officer Johannes Markestein, following the taxi phase of the
checklist as they rolled away from the gate, lowered (set) the flaps to
the correct 14 degrees for takeoff.  (There are separate checklists for
each successive phase of takeoff:  'BEFORE START' at the closing of the
passenger doors, 'TAXI' after receiving the taxi clearance, 'BEFORE
TAKEOFF' to be completed by reaching the hold line before the runway,
etc.)  However, then Captain Arthur Moen, following verbatim requirements
of the cold weather operations section of the Pan Am aircraft operations
manual, which says flaps in cold weather conditions (like Anchorage, AK
in late December at 6am) should be left _up_ (retracted) until lineup
for takeoff, in order to reduce the chance of snow or ice wedging the
flap screws into extended position or lodge between the flap and the
wing edge, did so.

Captain Moen immediately discussed his having done so with Markestein.
Markestein replied 'OK, let's not forget them.'  But there was no
post-taxi checklist item to -recheck- flaps before takeoff.  The
checklists in use made the brittle, often-untrue assumption that there
would have been no reason to touch the flaps after setting them to 14
degrees after taxi phase.  In fact, as NTSB pointed out in a subsequent
special report, the checklists' assumptions were easily invalidated by
any number of unplanned events or interruptions as the crew runs through
them.

And this is where the poorly debugged checklists interacted
catastrophically with the jet's hardware defect.  This was the 'takeoff
warning system', which is intended to sound a loud horn if pilots apply
takeoff thrust with either flaps, speed brakes, or the stabilizer
(vertical tail) in the wrong position for takeoff.  As this was in
the primitive era of electronics, application of takeoff thrust was
detected by a mechanical linkage to the thrust lever.  If it were
advanced to 42 degrees of angle, then the checks of those three systems
would be triggered and a horn go off if they weren't set right.  

And this is where there's a subtle physics problem.  Boeing had done
testing two years before, and found that if the ambient air were really
cold -- like Anchorage, AK in late December at 6:15 am -- triggering
wouldn't occur, and sent out a bulletin recommending that the actuator
be changed from 42 degrees to 25 degrees.  

And why would cold weather cause this?  (My surmise:)  Because cold air
is denser, and provides more lift, hence you get takeoff thrust without
the thrust lever being pushed as far forward.  Boeing didn't indicate
any special urgency, and completely failed to define what it meant when
it said this would be a problem in 'cold weather operations'.

It was about 1 degree F, that day in December, 1968.

For its part, Pan Am decided to not bother implementing the recommended
fix even though it would have cost less than $50 per plane:  Some
unnamed engineer in Pan Am service engineering 'decided the modification
was not necessary'.  An equally unnamed supervisor reviewed this
decision and 'decided, after coordination with flight operations, that
the bulletin was not applicable to Pan Am aircraft, and no further
action was taken.  The reason for this decision was not fully
documented.'  Great job, guys!

So, First Officer Markestein had said 'OK, let's not forget them', but
they were following procedure by working scrupulously through the
checklists, and nothing there said anything about the flaps.  The flight
was cleared for takeoff, and Markestein applied takeoff thrust --
advancing the thrust lever, but not to 42 degrees on account of the cold
air.  They rose, but started having trouble maintaining attitude control
and remaining above stall speed.  Doubtless they started working through
the emergency checklists, which unfortunately were written with the
assumption that the takeoff warning system was functional.  And, problem
is, the crew didn't have enough time to uncover the broken-checklists
problem.  They had exactly 59.2 seconds before the plane hit the ground
at 187 knots (215 MPH).  The broken checklist and the broken warning
system conspired against them, and they didn't find the problem in time.

And all of that gets boiled down by Wikipedians to 'Crew attempted
a no-flaps takeoff'.

_So_ tired of that.

(I have a running gag with Deirdre every time we fly together, whereby
one of us needs to check out the window before takeoff to ensure 14
degrees flaps.)

I need a drink after writing that.


----- End forwarded message -----