From: justin corwin (email@example.com)
Date: Sun Jan 01 2006 - 13:18:04 MST
On 1/1/06, Damien Broderick <firstname.lastname@example.org> wrote:
> It seems completely
> reasonable to some people to expect that if humans can do Y, and you know
> how to do Y, you should be able to do it every time you're asked to. That
> is largely true of such tasks as tying your shoe laces, but it breaks down
> badly when an artist is challenged to "Go ahead, then, if you're so
> `creative', write a book/paint a painting/compose a symphony, and make a
> million dollars with it!" Even so routine a human experience as falling in
> love has proved to be unpredictable, even zany. Those humans! So much more
> irritating than simple machines.
You keep reframing the problem as if we're asking for fantastic,
elaborate evidence, at the limits of performance, with perfect
reliability. All it would take to satisfy critics is ANY effect that
is robust with regards to experimentation, at whatever success rate.
We do in fact grade creativity in art colleges, we have artistic exams
and criticism and reams of data on success on the US Advanced
Placement Art classes and exams in high schools, for example. The
correlations and relationships that we find there are useful enough
that we can teach people to be better artists, musicians, composers,
writers. Mushy, unpredictable, of course, but no one denies that they
exist, or that the educators and scientists studying it have not
identified useful trends and factors.
You don't need to win the lottery, although if your claims were true,
you should have no problem doing so, albeit at significant effort,
which I explored in a previous post.
You should be able to produce much more modest results very robustly.
An error-correcting code should be able to pick a specified digit
perfectly, with just a few thousand redundant guesses at the levels
that are claimed. That's opening the door for as ridiculously
uncertain a process as you need, iterated across as many retries as
you need, but still giving a certain answer.
I admit that I don't know whether this has been tried or not. My
impression of the statistical juggling that occurs in psi research to
prove that something is happening *above chance* is that no one has
attempted to use cumulative statistics to extract more certain effects
from the 'above chance' results.
My cynical take is that this is because the effects, however slight,
cannot actually be *grasped* to do anything, because they do not
exist, in a practical sense.
-- Justin Corwin email@example.com http://outlawpoet.blogspot.com http://www.adaptiveai.com
This archive was generated by hypermail 2.1.5 : Wed Jun 19 2013 - 04:01:10 MDT