The Structure of Scientific Revolutions
Seminal work on the history of science which gave us such concepts as ParadigmShift
and that new ideas come from the outside and can't take hold until the old folks die. In the introduction, ThomasKuhn
cites an older and less well-known book, GenesisAndDevelopmentOfaScientificFact
, that actually describes a lot of his ideas already. Some people argue he didn't give the author appropriate credit.
A few major points:
- Scientific work is done in the context of a specific worldview.
- ParadigmShifts ( KuhnParadigmShift ) happen when that world view is challenged by anomalous data and a new theory proposed by someone outside the existing worldview challenges it.
- Paradigms are extremely resistant to shifts, and most potential shifts fail or take a very long time.
- Knowledge is lost in the shift. New paradigms are not just additions to knowledge, but changes in the set of knowledge.
- Discounted worldviews were not wrong. They were self-consistent and reasonable.
See "The Revolution That Didn't Happen" http://www.astro.uni-bonn.de/~willerd/weinberg.html
for a riposte by Steve Weinberg.
Knowledge getting lost in the shift ?
If we consider the shift from structured to OO techniques a ParadigmShift
(which can probably be defended), then there were good things about the former that got lost. I suspect this is typical, because the students of the new paradigm don't bother learning much about the former, and so have to rediscover the subtleties the experts of the former paradigm had come to understand.
I've noticed that a lot of OO (good, experienced) programmers have no notion of CouplingAndCohesion
I think that "high cohesion and loose coupling" is a good example, especially when this was one of the causes of the paradigm shift in the first place. This also points to the fact that OO methodology is not
contained in a tool or language. -- JeremyCromwell
A possible few examples : "goto", how to know the location of islands by patterns in the waves, how to build the great pyramids. -- ErikMeade
It is certainly true that the new paradigm tends to reject
older ones rather than build on them. I'm not so sure that the old paradigm is lost
. It goes into suspended animation -- nobody works that way or talks that way after a generation or so -- but you can still use the old paradigm if you choose to. Thus I can use IsaacNewton
's Laws if I prefer them the AlbertEinstein
's relativity or Quantum Theory. Perhaps I might go back to MrAristotle
and all those epicycles. -- DickBotting
Kuhn is reported (I'm only just getting around to reading the book, for now I have to qualify what I got from secondary sources) as essentially saying that you
can't use Aristotle's or Newton's paradigm - they are no longer understandable to us on their own terms. Rather, we have access to a "Whiggish" reinterpretation of the old paradigms under the terms of the new. A lot of the careful analyses of old scientific notions that StephenJayGould writes so brilliantly seem to bear out this argument; his treatment of Lyell's "uniformitarian" doctrine is a nice example. -- LaurentBossavit