|
![](/i/fill.gif) |
Warp wrote:
> There was an article somewhere on the subject of why "clever" questions
> which expect "clever" answers are actually counterproductive.
>
> A typical "clever" question is like "how would you weigh a Boeing 747?",
> and it's expected that you come up with some convoluted solution to the
> problem, while pragmatic answers like "I would consult the specification
> manual" or "I would ask an engineer" are discarded.
That would, indeed, be a rather stupid question.
> However, from a practical point of view those pragmatic answers are
> much, much better than any convoluted answer you could come up with.
> Why? Because if you come up with a convoluted answer it shows that you
> tend to come up with convoluted solutions to simple problems, rather than
> doing the practical thing and just consult a manual or other source which
> *already* knows the answer, which would be the best thing to do in a
> practical situation in the job.
Some people evidently feel that doing things the hard way makes them
look somehow "superior".
Tangentally related: Some scientists used genetic algorithms to try to
"evolve" a simple electronic oscilator. Like, they hooked a computer up
to some kind of programmable IC, set it off with randomly initialised
ICs, and let it mutate and cross the wiring sequences and apply a
fitness function - you know, the whole genetic algorithms thing.
Eventually they came up with a device which did indeed produce the
desired sinewave. But the engineers looked at the wiring diagram and it
looked absolutely nothing like an oscilator. In fact, they couldn't
figure out how the hell it works. (This is apparently not uncommon with
genetic algorithms.)
More surprisingly, they tried to do a demo for somebody else and
discovered that the device no longer worked. The rather puzzled
experimenters did some experiments, and eventually discovered that the
device only actually works in the room where they originally evolved it.
At this point, they figured out how the device actually works. It isn't
an oscilator, it's a radio receiver. And it was receiving interferrence
from one of the other items of kit in the lab. This simultaneously
explains why the circuit looks nothing like an oscilator (i.e., IT
ISN'T), and why it stopped working outside the lab.
It also demonstrates that genetic algorithms will solve a given problem
by any and all means possible, not necessarily the one you were
expecting. If you say "generate this signal please", it will generate
that signal - even if that means receiving it as radio waves!
I'm sure biological evolution is prevaded by similarly counter-intuitive
adaptations...
Post a reply to this message
|
![](/i/fill.gif) |