Is science "predictive"?

John Mcreery (JLM@TWICS.COM)
Thu, 7 Sep 1995 14:00:09 +0900

We need, I think, to look more closely at what it means to say
that science should be predictive. The prototype of this idea is
experimental research where manipulation of independent
variables (the ones we control) leads automatically to
predicted effects. Statistical research predicts that repeated
performance of experiments will lead to a certain
distribution of results--though not any particular one
automatically. Research on chaotic systems deals with
situations in which inputs beyond certain thresholds can, even
if apparently miniscule, have dramatic effects. Scientists who
study "complexity" have been coming around to the view that
"scientific" understanding of many natural processes is
inevitably "historical," since we could not predict the
particular outcomes that provide the data with which they
work. Stephen Gould is one example. In _Wonderful Life_ , for
example, he argues that fossils preserved in the Burgess shale
show that life could have evolved in quite different forms than
those we see today. He is arguing against those who see
adaptation to specified environments as inevitably producing
the same forms. By this point, we tremble on the verge of that
fine German distinction between "Natural" and "Spiritual"
sciences, where the former confine themselves to "outside-in"
explanations while the latter require "inside-out"
interpretations, though arguably both attempt to predict
behavior: the difference being that between an input implies
output system and one in which output is variable depending
on how an intervening "black box" processes the input. When
this distinction is crossed with a choice of linear, statistical,
chaotic, and complex processes, we have a lot to think about.
Add theories of legal and clinical evidence, hermeneutical and
stylistic considerations. We do have a lot to talk about.

John McCreery