Recently I completed reading of the second book in the Golem series, the complete being The Golem at Large: What you should know about technology by Harry Collins and Trevor Pinch. The book discusses cases from technology field in which there is a ‘regress’, in even expert people are not able to decide objectively what to make out of results of experiment, which at first sight seem to be so objective.
Some of the examples that they choose are well known, some are not. For example the much famed demonstration by Richard Feynman on O-Rings is brought out from its almost cult status. The demonstration by Feynman when looked at with all the background seems to be very naive. Similarly many other examples de-mythify different examples from different technologies.
Some of the quotes that I have liked are as under.
+ 4 It would, of course, be foolish to suggest that technology and
science are identical. Typically, technologies are more directly
linked to the worlds of political and military power and business
influence than are sciences.
+ 6 But disputes are representative and illustrative of the roots of
knowledge; they show us knowledge in the making.
+ 10 It would be wrong to draw any conclusions for science and
technology in general from wartime statements; wartime claims
about the success of the missile reflect the demands of war rather
than the demands of truth.
+ 28 As always, if only we could fight the last war again we would
do it so much better.
+ 28 Just as military men dream of fighting a war in which there is
never any shortage of information or supplies, while the enemy
always does the expected, so experts have their dreams of
scientific measurement in which signal is signal and noise follows
the model given in the statistical textbooks. As the generals
dream of man- oeuvres, so the experts dream of the mythical model
+ 28 Even when we have unlimited access to laboratory conditions, the
process of measurement does not fit the dream; that was the point
of our earlier book ¡V the first volume of the Golem series.
+ 32 Skimp, save and cut corners, give too much decision-making
power to reckless managers and uncaring bureaucrats, ignore the
pleas of your best scientists and engineers, and you will be
+ 38 Whether two things are similar or different, Wittgenstein
noted, always involves a human judgement.
+ 40 The `correct’ outcome can only be achieved if the experiments or
tests in question have been performed competently, but a competent
experiment can only be judged by its outcome.
+ 62 The treatment of the controversial aspects must be different to
the uncontroversial aspects. The same is true of what we loosely
refer to as experiments: one does not do experiments on the
uncontroversial, one engages in demonstrations.
+ 64 In an experiment, that would be cheating, but in a display, no
one would complain. A demonstration lies somewhere in the middle
of this scale. Classroom demonstrations, the first bits of science
we see, are a good case. Teachers often know that this or that
`experiment’ will only work if the conditions are `just so’, but
this information is not vouchsafed to the students.
+ 64 A demonstration or display is something that is properly set
before the lay public precisely because its appearance is meant
to convey an unambiguous message to the senses, the message that
we are told to take from it. But the significance of an experiment
can be assessed only be experts.
+ 71 Anything seen on television is controlled by the lens, the
director, the editor and the commentators. It is they who control
the conclusions that seem to follow from the `direct evidence of
+ 74 The public were not served well, not because they necessarily
drew false conclusions, but because they did not have access to
evidence needed to draw conclusions with the proper degree of
provisionality. There is no short cut through the contested
terrain which the golem must negotiate.
+ 77 A vast industry supported by national governments makes sure it
understands how oil is found, where it is found and who has the
rights to find it.
+ 82 In some ways it is easier to delve into the first few
nanoseconds of the universe than to reconstruct something buried
deep in the core of the earth.
+ 86 This is the `experimenter’s regress’. If you believe that
microbiological activity exists at great depths then this is
evidence that a competently performed experiment has been carried
out. If you believe that microbiological activity is impossible or
extremely unlikely then the evidence of biological activity is
evidence for doubting the experiment. Experiment alone cannot
settle the matter.
+ 91 In short, Gold’s non-biological theory and its assessment are
intertwined with the politics and commerce of oil
exploration. There is no neutral place where a `pure’ assessment
of the validity of his claims can be made.
+ 96 With several hundred equations to play with, this is an area
where `theory’ and `guesswork’ are not as far apart as
conventional ideas about science would encourage us to think.
+ 102 I think there are really two different approaches. One is to
say that this is a branch of science and that everything must be
based on objective criteria which people can understand. The other
is to say that is just too inflexible, and that there’s something
called judgement – intuition if you like – which has its place in
the sciences and that it’s the people who are intuitive who are
+ 104 It is also possible to argue that modellers who did not suffer from big
mistakes were lucky while some others were unlucky to have been wrong.
+ 106 Even if you believe that large errors are bound to prove you
wrong, you may still argue about the meaning of `large’ and you
may still think that the difference between accuracy and
inaccuracy was not clever economics but luck. Finally, you may
always say that the economy changed radically.
+ 106 … it was not the model but the economy that was wrong.
+ 107 The experimenter’s regress occurs when scientists cannot
decide what the outcome of an experiment should be and therefore
cannot use the outcome as a criterion of whether the experiment
worked or not.
+ 107 Oh absolutely, that’s why it’s absolutely pointless to publish
these forecast error bands because they are extremely
large. . . . I’m all for publishing full and frank statements but
you see the difficulty [with] these standards errors is that
+ … In fact, we could have done this at the National Institute in
the mid 70s, but we suppressed it on the grounds that the standard
errors were so large, that it would have been difficult for
non-specialists, you know people using the models, using the
forecasts, to appreciate. It would have discredited them.
+ 108 Science is often used as a way of avoiding responsibility;
some kinds of fascism can be seen as the substitution of
calculation for moral responsibility.
+ 110 That is, it selected those who were `. . . willing to
subordinate their education to their careers’.
+ 111 The economists who build the models deserve credibility, but
their models do not; one should not use the same criteria to judge
expert advice as one uses to judge the coherence of a model.
+ 124 Flipping to and fro between science being all about certainty
and science being a political conspiracy is an undesirable state
+ 149 In effect, a group of lay people had managed to reframe the
scientific conduct of clinical research: they changed the way it
was conceived and practised.
+ 151 Feynman gives the impression that doubts can always be simply
resolved by a scientist who is smart enough.
+ 151 The danger is always that enchantment is the precursor of
+ 153 Golem science and technology is a body of expertise, and
expertise must be respected. But we should not give unconditional
respect before we understand just what the expertise comprises and
whether it is relevant. To give unconditional respect is to make
science and technology a fetish.