[lit-ideas] Good enough and not knowing why

Over on anthro-L Jacob Lee offers the following thoughts about the role of
theory in science and engineering.

In engineering, biologically inspired techniques often sacrifice
comprehension in favor of robustness. We might not understand why an
algorithm or a structure generated using a genetic algorithm works, because
it is too complex to figure out without a lot of effort, but may be
satisfied with knowing that it does and with the methods for designing such
systems. Adrian Thompson's tone discriminator is a good example. Thompson is
a pioneer in the field of evolutionary hardware. One of his early projects
was to evolve a tone discriminator in hardware:
http://www.cogs.susx.ac.uk/users/adrianth/ascot/paper/paper.html. One of the
interesting features of this tone discriminator is that it shouldn't work
but it does. It shouldn't work because the design (as abstracted into its
logical components) wouldn't do the job. There are indispensable parts to
the tone discriminator which are completely disconnected from it in its
conventional wiring, "*possible mechanisms include electromagnetic coupling,
or interaction through the power-supply or substrate. Evolution was able to
exploit this physical behaviour, even though it would be difficult to
analyse.*"

So, while I don't entirely concur with the article's prediction of the
demise of science, there is a sense in which we have far less control and
far less understanding (and require it far less) in engineering and the
sciences than ever before, and there is a sense that in some cases, humans
are, well, a little obsolete, even if wouldn't want to overstate the
abilities of computers. We are starting to run projects for which we are not
well equipped.


That last paragraph is certainly thought-provoking.
John

-- 
John McCreery
The Word Works, Ltd., Yokohama, JAPAN
Tel. +81-45-314-9324
http://www.wordworks.jp/

Other related posts: