From: David Heddle <heddle@gmail.com>

Date: Fri Aug 17 2007 - 06:34:03 EDT

Date: Fri Aug 17 2007 - 06:34:03 EDT

Iain,

*"I'm not sure I agree here. The sensitivity of the constants to

suitability for life is the correct form of fine-tuning. However, when you

say we can't a priori estimate the probabilities, I think you're overlooking

the fact that from a Bayesian perspective, one always has a prior estimate

of the probability distribution, and if one can't estimate it at all, then

one chooses an "uninformative" (ie containing no information) prior

distribution, which would usually be a uniform distribution. So if a

parameter lies between (1-eps)*x and (1+eps)*x, and the physically possible

values of x are between 0 and 2 with a uniform distribution, then one can

argue fine-tuning if eps << 1 because of the probability. However, if (by

some science) we can find (for reasons other than the suitability for life)

that it can lie only in the region (1-eps)*x ... (1+eps)*x, then clearly you

can't claim fine tuning because any of the physically allowable values would

be suitable for life.

So I would say it has to be linked to a probabilitic argument, otherwise to

talk about "sensitivity" has little meaning."

*I respectfully disagree with your disagreement.

First of all, it makes little sense to me to talk about the physically

possible values of the physical constants. For us they seem to be free

parameters. What is the range of physical values for G? How would you pick

that range to perform a Bayesian analysis? It seems to me you can't. All we

can approximate is the range that life requires. Is that range a modest

fraction of the measured value? If so, we have fine tuning. But how that

range fits inside a larger range of physically possible values is impossible

to say. My point is that fine tuning is, counter intuitively, independent of

that knowledge.

I think this can be seen by concentrating on just one example, the

cosmological constant. For those who do not know, the cosmological constant

cannot be much smaller or bigger than its current value or the expansion of

the universe would be affected to the point where any kind of life would be

impossible. (E.g., no stars, therefore no heavy elements, therefore no

life.)

Scenario One: (the present scenario).

The theoretical value of the cc is 100 or more orders of magnitude larger

than its actual value—the value that has the right magnitude to make the

universe habitable. It is reasonable, for the sake of argument, to assume

the actual value and the theoretical value are in the "physically possible"

range. So here we have a case that without question the physically possible

range is many orders of magnitude larger than the habitability range. So we

say agree there is fine tuning. We are surprised that the value required for

life was chosen from this huge sea of possible values. (Here we also have

circumstantial evidence for a multiverse.)

Scenario Two: We develop a fundamental theory that predicts the value of the

cc. Now we are surprised, not that the cc was chosen from such a large set

to have a value in the habitable range, but by just the opposite: that the

fundamental theory spit the value out—habitability was built into the

theory. Fine tuning survives. (And here we have effectively ruled out the

multiverse as an explanation for fine tuning.)

So in either case we have fine tuning, even though the underlying

distributions ranged from one that was effectively infinitely broad compared

to the habitability range to one that was narrowly peaked in the

habitability range.

In my opinion, the fine tuning only is based on the idea that if a constant

changed by a modest amount from its known value, life would be impossible.

It does not depend on underlying distributions.

David Heddle

Associate Professor of Physics

Christopher Newport University &

The Thomas Jefferson National Accelerator Facility

On 8/17/07, Iain Strachan <igd.strachan@gmail.com> wrote:

*>
*

*>
*

*>
*

*> On 8/16/07, David Heddle <heddle@gmail.com> wrote:
*

*> >
*

*> > Moorad,
*

*> >
*

*> > That is *exactly* right and yet a point often missed. The stength of the
*

*> > fine tuning argument is so-often incorrectly linked to the probability of
*

*> > the constants. That, in turn, leads to obfusacting and irrelevant
*

*> > discussions about how we possibly could, a priori, estimate those
*

*> > probabilities. Well, we can't.
*

*> >
*

*>
*

*>
*

*> I'm not sure I agree here. The sensitivity of the constants to
*

*> suitability for life is the correct form of fine-tuning. However, when you
*

*> say we can't a priori estimate the probabilities, I think you're overlooking
*

*> the fact that from a Bayesian perspective, one always has a prior estimate
*

*> of the probability distribution, and if one can't estimate it at all, then
*

*> one chooses an "uninformative" (ie containing no information) prior
*

*> distribution, which would usually be a uniform distribution. So if a
*

*> parameter lies between (1-eps)*x and (1+eps)*x, and the physically possible
*

*> values of x are between 0 and 2 with a uniform distribution, then one can
*

*> argue fine-tuning if eps << 1 because of the probability. However, if (by
*

*> some science) we can find (for reasons other than the suitability for life)
*

*> that it can lie only in the region (1-eps)*x ... (1+eps)*x, then clearly you
*

*> can't claim fine tuning because any of the physically allowable values would
*

*> be suitable for life.
*

*>
*

*> So I would say it has to be linked to a probabilitic argument, otherwise
*

*> to talk about "sensitivity" has little meaning.
*

*>
*

*> I think probabilistic arguments are perhaps one of the motivations behind
*

*> "multiverse" theories - by invoking astronomical numbers of universes, all
*

*> with different parameter sets, one greatly increases the probability of at
*

*> least one having just the right values. Buy one lottery ticket and you are
*

*> extremely unlikely to win the jackpot - but every week millions are bought,
*

*> and someone usually wins.
*

*> [Incidentally this kind of multiverse is very different from the "many
*

*> worlds" type with which I started this thread. In the MWI, all the
*

*> universes obey the same physical laws and have the same constants].
*

*>
*

*> Iain
*

*>
*

*> -----------
*

To unsubscribe, send a message to majordomo@calvin.edu with

"unsubscribe asa" (no quotes) as the body of the message.

Received on Fri Aug 17 06:34:56 2007

*
This archive was generated by hypermail 2.1.8
: Fri Aug 17 2007 - 06:34:56 EDT
*