Re: testing psi

From: Damien Broderick (
Date: Mon Jan 02 2006 - 19:55:32 MST

At 06:06 PM 1/2/2006 -0800, Yedidya Weil wrote:

>I actually submitted an experimental protocol to the email address on the
>site for [Randi] to comment on and he (or the person writing in his name)
>didn't even consider it, just sent back a short, rude reply.

Even if he and his associates were as sweet as pie, the thing to notice is
that to win the challenge, you need to do at least the following (and bear
in mind that you don't just stroll in, shake hands with the judges,
levitate across the room in front of the video camera, and pick up the
million-dollar check; first you pass a screening test with a probability of
success of one in a million, then you do it again sometime later):

To beat these odds, the psychic has to do the equivalent of calling
correctly an unbroken sequence of 20 random binary options -- white/black,
male/female, whatever -- out of... 20. Very few subjects in parapsychology
experiments have come remotely near that level of accuracy, nor could they
be expected to, given that we know psi under laboratory conditions has an
effect size ( of 0.2 or much worse.

Of course, anyone who could reliably guess 20 spins of a roulette wheel out
of 20 would not need to bother Mr Randi in the first place. Since no
serious parapsychologists ever claims this sort of success rate, it's time
to move smartly on and start considering more reasonable tests of the
existence of such an effect -- as parapsychologists have done for years.

Damien Broderick

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:55 MDT