> If you could model a conscious agent in a computer, and they got hurt when
> they hit your (information only, pattern only) model of a wall, they'd think
> that the wall was solid. When in fact it'd just be information.
Would this necesarily mean, though, that the computer agent would not be
able to pass through the "wall"? It may recieve the feedback that there is
a "wall" there, but is it incentive enough to try a different direction,
or to perhaps prevent the agent passing through that space?
> Sorry. I broke my golden rule and drank espresso after 8pm. Please now allow
> me to sit here gibbering till 4am. I've already gnawed off one of my arms.
Was it tasty?
Drakir