One thing that I have been writing about, perhaps with less clarity than I
would like, is the difference between something being reducible to
observations and being consistent with observations.  Another is whether
one can define values like right and wrong or good and evil in terms of
phenomenon.

Illustrations may be helpful in explaining this idea.  I'll grab examples
from physics and meteorology because I'm familiar with those fields.  Sorry
to the folks who aren't, but you're welcome to your own examples.

Let us first consider quantum electrodynamics (QED) and chemistry.  As the
name implies, QED is the quantum theory of electromagnetic interactions.
It does not cover strong, weak, or gravitational interactions.  But, it
should explain, not only things like the energy bands of the atoms, but
interactions between the atoms too.  With enough computational horsepower,
one should be able to start with electrons and nuclei and deduce the
chemical reactions from them.  Last I have heard, progress has been made in
this area, with simple chemical reactions being derived from QED.

My cousin, who has his PhD in chemistry, has no doubt that, in principal,
all chemical reactions can be derived from QED.  Further, while the
computational requirements for complex chemical reactions are significant,
the complications are more linear than exponential.  That is to say that
that we don't have the type of compounding that would require the
theoretical limit of computational power run for the life of the universe
to solve these problems.  It's more on the order of what will be
increasingly possible as computing speeds increase.

Compare this to deducing a given pattern of light on a screen from a one
slit experiment.  There is no way to deduce why a given pattern exists.
There is no particular reason why the first photon hit in the lower left
quadrant, the second in the lower right, the third in the lower left, the
fourth in the upper right, etc.  So, even if we break the probability
distribution into only four buckets, we have an inherently unpredictable
pattern. There is no way to find out why a particular pattern was seen, it
just happened.  These patterns are consistent with, but not deducible from
QM.

Let's go to meteorology next.  Take the pattern of heavy tropical showers.
In Houston, we can get 3 inches of rain one place, while a couple of miles
away its bone dry.  In one case, we got half an inch of rain at our house,
drove to the store, and the pavement was dry less than a quarter mile away
from the place where it was pouring for the last 20 minutes.

There is no way to explain why our house was rained on and the nearby store
was not.  There is, of course, a way to explain why a location at X
receives a lot of rain and a location .25 miles from X does not.  But,
there is not a way to explain why it was X that got the rain and .25 miles
away that didn't instead of .25 miles away that got the rain and X that
didn't.

So, the rain pattern is consistent with everything we know about
meteorology, but it not deducible..even if we knew the starting conditions
to with Planck's constant. There is no reason for one pattern happening
instead of another.  To relate to another thread, the fact that Florida has
been hit 2+ times and Louisiana has been spared is just coincidence, its
not that Mother Nature prefers Democrats.

Let us now try to obtain some general rules from this.

There are some empirical patterns that can be seen to be derived from more
fundamental concepts.  Chemistry is derivable, in principal.and bit by bit
in practice, from QED.  If asked why chemical reactions have the properties
they do, one can point with confidence to QED.

There are some supposed patterns that can be practically ruled out because
they violate known laws.  Perpetual motion machines provide one example.
It is true that one cannot prove a negative; it is always possible that the
next perpetual motion machine will work.  But, I will take as an axiom that
proposed explanations which require a secret hidden violation of well
verified scientific theories can be rejected unless they can show such a
violation actually exists (and they wouldn't be hidden then.)

Some patterns are consistent with known laws, but not reducible.  The
location of scattered thunderstorms is one example.  Heavy rain over my
house and no rain at the local store is consistent with all we know about
science.  So is heavy rain at the store, and no rain at my house.  If we
call heavy rain at my house H, and heavy rain at the store S, we have the
logical propositions

(H & ~S)  and
(~H & S)

as statements that are consistent with known science.  Defining the
statement A by

A= (H & ~S)

We have

A & ~A as valid expressions.  One of the rules of logic is that a set of
axioms is considered invalid if both A and ~A can be deduced from them.
Thus, we cannot consider A deduced from axioms, its just consistent.

This also gives us insight into when an observed pattern cannot be
determined by reference to a simpler known phenomenon.  If, using the
techniques to determine A, one can also arrive at ~A, the technique is
shown not to provide the supposed linkage.  For example, if one has a
technique that shows how good can be expressed in terms of evolutionary
principals, using the same technique to obtain evil shows the that the
technique does not work as promised.  If that happens, we have the same
case as the rain case given above.  Both are consistent with evolution, but
neither is derivable from evolution.

Finally, lets bring in baseball.  Gautam has argued that there is no such
thing as clutch hitting.  In one sense, of course, he is absolutely wrong.
We can point to people who were very productive in key situations, even
though their career numbers were not stellar.  We can see .200 hitters with
3 home runs a year getting a key home run in a World Series game.  What
more proof do you need?

But, in another sense (I'm sure the one Gautam intends) he is right. His
argument, as I understand it, is that this clutch hitting is consistent
with normal statistical variations.  You can find the same type of "clutch"
hitting if you select people who have superior performances on August
13ths, for example.  The Houston Rockets' management are clutch coin flip
callers.  They won two coin flips in a row for the first draft choice, and
have won the draft lottery, when they only had the 6th (or so) worst record
in the league.

No-one would suggest that this is an example of their superior management
skills.  Some things just happen that way.  The same is true for clutch
hitting.  Someone will hit above and someone will hit below their average.
The real question is if it is statistically significant.

Having walked through the general rules, let us state them.

There are three categories we consider when we consider the relationship
between already established scientific theories and higher level
descriptions of phenomenon.

1) The higher level phenomenon are derivable from the lower level theory.
2) The higher level phenomenon are inconsistent with lower level theory.
3) The higher level phenomenon are consistent with but not derivable from
lower level theory.

Establishing the first requires an A->B proof.  Establishing the second
requires an A->~B proof.  Establishing the third requires that ~(A->~B).
The third is established exclusively if  ~(A->~B) &&  ~(A->B).

Dan M.




_______________________________________________
http://www.mccmedia.com/mailman/listinfo/brin-l

Reply via email to