Re: [asa] The Multiverse - Physics or Metaphysics?

>
>
> If I understood, you are referring to string theory as "the most
> successful cosmology." I would caution care here, because string theory has
> not made any testable predictions and so it is dubious that it can claim any
> success at all. It has been formulated to be consistent with what we know
> in nature, but in trying to make it consistent, so many unobservable
> dimensions and extra degrees of freedom had to be put into the theory *ad
> hoc* that it ended up predicting this "string landscape" that is so large
> that now you can get any answer imaginable and not just what we observe. If
> the theory has to give you every possible answer in order to also give you
> the one right answer, then it is questionable that it is really a useful
> theory at all. This is why it has been called into question by many leading
> scientists. You could have produced the same result using any other
> framework other than strings. I hope you are aware of this. There are many
> ways to formulate a framework that can include everything we know plus a lot
> more. Simply keep adding unobservable degrees of freedom until the
> framework allows the right answer. The key to science is Occam's razor in
> which the fewest number of *ad hoc* assumptions gives you the correct
> answer. So the critique on string theory is exactly correct, as many
> scientists have formulated it.

Again, this is very much along the lines I've been thinking. I see the same
line of reasoning in my field of statistical pattern recognition, and
data-fitting. Given any data set of any size, one can fit a model that
agrees exactly with all of the data points by having sufficiently many
degrees of freedom (adjustable free parameters) in the model. N data points
may be exactly fitted by a degree N-1 polynomial. e.g. one can always fit a
parabola (degree-2) through three points. But this observation can never
tell us anything interesting about the data - for it to be interesting in
any way (and to have any predictive value) the number of degrees of freedom
must be very much less than the number of data points. Specifically, with
the exact fit, one "overfits" the data, meaning that the function defined
between the data points will oscillate rapidly and be highly sensitive to
the measurement noise. By the same token, Koonin's paper invokes an
unlimited number (as many as necessary) of parallel worlds (in the MWI
sense) to get over the apparent extreme improbability of life.

I just discovered the following quote on the Wikipedia entry for Lee Smolin,
one of the strongest critics of string theory:

*The scenario of many unobserved universes plays the same logical role as
the scenario of an intelligent
designer<http://en.wikipedia.org/wiki/Intelligent_design>.
Each provides an untestable hypothesis that, if true, makes something
improbable seem quite
probable.*[3]<http://en.wikipedia.org/wiki/Lee_Smolin#_note-1>This
again is exactly what I have been saying from the start (though Smolin is
talking about string-theory based parallel universes rather than in the MWI
sense, the argument is the same). Koonin's paper contains at the end a
complete dismissal of Intelligent design. However, as Smolin's quote makes
evident, what Koonin replaced it with was something that fails in just the
same way.

Quite pleased I had independently arrived at the same idea as Smolin. [Smug
mode off]

Iain

To unsubscribe, send a message to majordomo@calvin.edu with
"unsubscribe asa" (no quotes) as the body of the message.
Received on Mon Sep 10 16:38:55 2007