[From Bill Powers (2008.09.15.1627 MDT)]
Ted Cloak 2008.09.15
10:27 MDT �
Ancestors
schmancestors. The only way one can be Sarah Palin is to be Sarah
Palin. (But who would want to?)
That’s the essence of what I was trying to say, only it took me more
words. All this arose from the question (from Gary Cziko) about whether
if you had
Sarah Palin’s genome and her upbringing, you could do anything different
from what she does. That question begs another: it assumes that this
proposition could be tested. It can’t be tested. Even if you cloned Sarah
Palin as an embryo, so she had exactly the same genome as the original,
you could not arrange the environment in the same way – for example, you
couldn’t supply another 911 attack, or the same President as the one we
had when Sarah-I was a baby. You couldn’t arrange for the same sharp
elbow to Mommy’ abdomen on a crowded bus at the same stage of cell
division. And (to argue like a lawyer) even if you could somehow
magically recreate the same environment, you could not predict the
outcome of all the reorganizations that Baby Sarah I underwent, because
repeating a random process (like Brownian movements) does not repeat the
outcome of that process.
The example I like to think of is rolling a ball bearing lengthwise down
a piece of railroad rail. The rail has a rounded top forming a piece of a
cylinder. If you roll the sphere down a longitudinal line an inch to the
left or right of the high point of the rail and parallel to the highest
line, the sphere will very predictably fall off the rail to the left or
right after traveling only a short distance. But as you roll the ball
along a line closer and closer to the exact high point of the rail, it
goes farther and farther before falling off, and your predictions of
which side it will fall on become less and less reliable. In the limit,
the more precisely you try to place the ball on the exact centerline, the
more closely your prediction will approach 50% success – pure
chance.
Any process containing one or more time integrations has this property,
because integration errors are cumulative. The longer the integration
runs, the less certain the value of the integral becomes, and the more
precisely you measure that value, the less repeatable it becomes. This
has to be related to the Butterfly Effect of chaos theory. Richard
Kennaway, am I talking nonsense?
Anyway, if the effects of a long series of reorganizations is cumulative,
we would expect the two Sarahs to diverge farther and farther as time
goes on. So the question can only be answered YES – the second Sarah
would almost certainly act differently from the first one, the
differences increasing with time and number of reorganizations.
Best,
Bill P.