Chronology Current Month Current Thread Current Date
[Year List] [Month List (current year)] [Date Index] [Thread Index] [Thread Prev] [Thread Next] [Date Prev] [Date Next]

[Phys-l] lab safety rules +- time on task +- symbolism



On 09/11/2011 02:31 PM, David Marx wrote:
No, the accident that triggered a major surge in safety inspections
and regulation occurred at another national lab around 1990 when a
graduate student was blinded by a laser on her(?) first day on the
job. The story goes that the superviser of that student went to
Washington to demand massive new safety inspections, training, etc.
Subsequently, safety inspections took place at all of the national
labs and many things were found and subsequently fixed, but it didn't
stop there. People have simply gone crazy over the years pointing at
things in labs and saying they are unsafe to the point that no one
can do anythng.

I don't buy it.

I have personally worked at NIST under the post-1990 safety
rules, which are almost certainly the same rules that apply
at Argonne.

To say that safety has gotten "to the point that no one can
do anything" is ludicrous.
-- With one tiny exception, I hold myself to standards that
are *higher* than what the federal rules require.
-- I will mention the exception not to complain, but to show
just how tiny it is: The requirement as to "no unlabeled
reagents" was interpreted so strictly as to forbid an
unlabeled beaker of water, even though I knew what it was
and was sitting only a few feet away. Left to my own
devices, I wouldn't have labeled such a beaker ... but on
the other hand, I could see the goal of the regulation,
and I support the goal, and most importantly, the burden
of labeling the beaker wasn't going to make me or anyone
else unable to "do anything".
-- While I was there, a researcher suffered serious permanent
eye injury ... so you really can't tell me that the rules
are too strict or too zealously enforced. Sorry.

===========

Now let me return to the "time on task" discussion from last
week. The only thing at NIST that I would complain about was
that I had to attend a certain amount of safety training every
so often. This was measured in terms of "time on task" i.e.
time sitting in the training room, not measured by how well I
understood the issues. I had to put in the hours, even though
everybody in the place knew my personal standards were tighter
than the official standards.

This is a blatant violation of the principle that says you
should measure the thing you care about.

Closely parallel remarks apply in another situation: To keep
my flight instructor certificate, every two years I have to
sit through an FAA-approved "Flight Instructor Refresher
Course" (FIRC). In order to be approved, an FIRC has to
involve 16 hours of "time on task". Cue the sound of
fingernails scraping on a chalkboard. The fact that I've
attended a lot more than 16 hours of seminars over the last
two years, and indeed /led/ more than 16 hours of seminars
covering the same material (more broadly and deeply) doesn't
count; they want to see 16 hours of time on task. At the
end of the FIRC there is a test. Why not put more emphasis
on measuring the thing you care about? Why not start by
giving a test, and if the student gets less than 100% then
we can talk about which ones were missed.

The only good thing about this FIRC requirement is that
there is such a thing as an online FIRC, which means I can
sit at home and read a book and actually learn something
while also paying enough attention to the FIRC to make
sure I don't miss anything.

============

One more aviation story to illustrate the point I am making:
Suppose a student pilot moves the landing gear switch to
the Down position and says "Gear down." We are going to
have a conversation about that. That switch is related
to the landing gear, but it is not the landing gear. The
switch lever is even decorated with a miniature wheel,
symbolizing the landing gear, but still it is not the
landing gear. The proverb is:
"Do not mistake the symbol for the thing symbolized."
In particular, there are dozens of reasons why the actual
landing gear might not be down, even though the switch
is down. There could be a hydraulic leak, or the circuit
breaker could be blown. At some point the wily flight
instructor is going to help you out by pulling the circuit
breaker and making sure you notice. The conversation
also always covers the various indications you can use to
close the loop, i.e. to make sure the actual landing gear
is actually down-and-locked.

Sometimes in the physics lab, and in life in general, we
cannot directly measure the thing we care about ... in
which case we have to measure some proxy variables and do
the conversion. However, the same logic applies:
"Do not mistake the proxy for the thing
you actually care about."

This is relevant because "time on task" is not the thing
you care about. You ought to care about knowing the course
material, knowing how to think, and knowing how to do things.
"Time on task" may be weakly predictive of something useful,
but only very weakly, and there are even some situations
where it is _anti_correlated with the things you care about.

As another example, grades and GPAs are not the thing that
anybody ought to care about. You ought to care about knowing
the course material, knowing how to think, and knowing how
to do things. Under some conditions, a GPA might serve as a
proxy for those things, but it is not a very good proxy, and
there are plennnty of situations where it is _anti_correlated
with the things you care about.

-- People will optimize the thing that gets measured, so
be careful what you measure.
-- Measure the thing you care about.
-- Do not mistake the proxy for the thing you actually
care about.
-- Do not mistake the symbol for the thing symbolized.