Double Helixes and Double Crosses (was: Favorite NON-fiction)

Sat Feb 1 10:52:45 PST 2003

On Tue, Jan 28, 2003 at 08:35:50PM -0500, Michael Barr wrote:

> Many (ok, nearly all) actual scientists would reverse 1 and 2 below.  In
> fact, they claim that it is impossible to gather data except in support of
> some--generally ill-formed--theory.  The reason is that there is simply

As has been said, this is too simplistic.  It is possible to look at the world
and go "that's odd".  "Why do those stars move?"  A philosopher could talk
about an assumed prior theory "stars don't move" being contradicted by the
planets, but I doubt most scientists always work at such a self-aware level.

> example, if you look at the Darwin finches, the relevant data are the
> sizes and shapes of their beaks.  Not the details of markings or length or
> shape of toenails or a myriad of other features that they differ in.  It
> is only when you formulate a theory that they evolved in response to the
> dietary niche that you realize the bills are the relevant data.

Yet Darwin was a creationist when he set out on the Beagle.  Something got his
attention as he went around the world.  And it wasn't a simple contradiction
since you can't contradict "god made every species".

> then gather less relevant data.  In fact, data gathering and theory making
> really go on in parallel.  


I've always been baffled by these debates, it seems like people take whatever
list of "scientific method" they see as a rigid formula.  I always thought the
essence was the iterated generation and testing of hypotheses, and revision of
the model which generated the hypothesis.  I.e. 3-6 in the list.  Arguing
about the order of 1 and 2 seems odd, they're not the important part.  Sure, I
just did it above, but only in kneejerk response. :)

> > 1. Gather data
> > 2. Form hypothesis
> > 3. Devise experiments to test hypothesis
> > 4. Compare results
> > 5. Revise hypothesis
> > 6. Repeat previous three steps until confident you're right
> > 7. Publish so other scientists can try to reproduce your
> > results and find holes in your assumptions and chain of 
> > logic, etc.

-xx- Damien X-)