Re: More on appeal to authority.

[John Denker <jsd@AV8N.COM>]

RAUBER, JOEL wrote:
> There is a grey area.  One person's hard evidence is everybody else's soft
> evidence.

At least initially, yes.

> E.g. John has hard evidence that he can design a voltmeter with
> input noise less than "root hbar".  Presumably because he did it.

_We_ did it.  Joint work.  Bernie and me.  He was the brains
of the operation.

> However, for me; the way I read John's piece, this can never be any better
> than soft evidence, because it is merely some claim I've read over the
> internet and indeed according to John it is an extraordinary claim as it
> flouts authority and should require extraordinary evidence; which of course
> I don't see.

Well, the general point has merit, but this isn't a good
illustration of the point, because in this case it's
a theoretical result, so you can harden the evidence by
replicating the calculation yourself.  It's slightly laborious,
but not insanely complicated or tricky, especially if you have
a reference that tells you what the desired result is, and how
to obtain it:
  Yurke and Denker, Phys. Rev. A 29, 1419–1437 (1984).
  http://prola.aps.org/abstract/PRA/v29/i3/p1419_1

You have to know what a creation operator is, and how to do
calculus-of-variations.


> There is a regime between hard evidence and soft evidence that deserves some
> sort of classification.  Namely something along the lines of the soft
> evidence, logically and coherently presented, provided by trusted
> authorities.

Let me add:  multiple independent authorities, or shall we say multiple
independent reputable sources.

As an illustration:  suppose you want to know the distance from
the earth to the moon with 10% accuracy.  There are umpteen different
ways of measuring that, and if one way has a systematic error it is
unlikely that all the others will have the same error.  The measurement
has been done many times by independent groups, so you would think
that due to natural human competitiveness, if the results were
inconsistent somebody would have made a big fuss about it before now.


> I think I'm questioning how far apart the top of the soft evidence is from
> the bottom of the hard evidence.

In principle there must be a gray area, but it seems remarkably
uncommon to actually land in the gray area.

Perhaps this is because there are some strong nonlinearities
involved.  As I like to say:
  One result with 10-sigma confidence will outweigh
  ten results with 1-sigma confidence.

Outweigh by a lot.