[Martin Taylor 960324 17:30]
Earlier today I claimed that the path of e-coli that was disturbed at every
moment would be unpredictable, and a controller would have no reason
to choose when to hit the spacebar/mouse. That was wrong, I think. The
situation would be one of Brownian motion, which was discussed by Hans
Blom a couple of weeks ago. The disturbed e-coli moves from its position
in such a way that after t seconds it is a distance proportional to
sqrt(t) away from its prior position, in some unpredictable direction.
This is predictability of a kind, even if that sqrt(t) is only a most
probable distance. And as Hans showed, this allows for the possibility of
control. The controller need only hit the spacebar faster, the further
the e-coli from the target. It will not sit on the target, but at least it
can be kept resonable close that way. In spite of the fact that the
predictability is very low, some control is still possible.
Martin