|
[DPRG] Re: Allen Robots (Will's emotions)
Subject: [DPRG] Re: Allen Robots (Will's emotions)
From: John Swindle
swindle at compuserve.com
Date: Thu Mar 8 19:41:13 CST 2007
What difference does it make how the robot is programmed? It just
does exactly what you told it to do.
The escape behaviours are marvelous. But I think the emergent
behaviors are only the ones that the programmers allow to emerge.
Don't the programmers mess with the subsumption and the monitor
until the robot escapes from whatever the roboteer observered?
Here is what gets me about programming anything:
Processor designers and software designers spend their whole
job-life trying to optimize a set of parameters that evolved.
Parameters such as number of pipes, pipe depth, page size, TLB
size, cache architecture and size, coherency protocols, mutex
protocols, and system topology. They try to decide such ephemeral
things as whether the page table entries should be cached in the
L1, L2, L3, and L4 caches, as well as in the TLBs. They try to
decide whether the compiler should be allowed to discard the
PREFETCH and static branch hints that the programmer supplied.
They do on-the-fly recompilations while the code that is being
recompiled code is already running! The people who do these
things are fascinating people, but this all seems to be a total
waste of human existence to me. I am tainted by Asimov's fiction
of a compute engine (his positronic brain) designed by two
successive generations of compute engines, to do what we want,
but in a way that no person has any hope of understanding. I am
amazed that we do not already have logic compilers that can take
a complete program and its target processor and compile a state
machine to do the whole thing.
This forum strides between the practical and the theoretical. A
noble stance.
Later,
John Swindle
More information about the DPRG mailing list
|
