15may2002

Biologic computing


"natural systems escape legacy, because you can't escape them anyway"
"be smart about searching a complicated fitness landscape", "make big jumps",
"not to follow a specific mechanism"

[can you have a strategy that isn't emcompassed by an algorithm?]

"software today is basically dead" -- it doesn't change, doesn't breathe

need to make software like a living organism [why? what problem is this solving?
yes, there are all the standard problems, but is this a big leap thing: an
aesthetic leap that may or may not bear fruit?]

reflection: smalltalk is written in smalltalk [nobody uses "metaclasses" <--
what are they?], so can redefine methods or even what it means to invoke a
method inside the program.. but apparently nobody uses this feature. hey, what
is it?

[why not? is it because writing programs from inside the program is really hard
to do? i mean, if you write something generic, how is writing the program inside
itself going to achieve the goal in a different way? programming is *artistic*
[or is it?], is that the problem?]

MIT: amorphous healing. check this. going to run on bacteria genetically
engineered to send 1s and 0s to other bacteria. what?!
http://www.swiss.ai.mit.edu/projects/amorphous/
from that page: "How do we obtain coherent behavior from the cooperation of
large numbers of unreliable parts that are interconnected in unknown, irregular,
and time-varying ways?"

loose coupling is an idea which is out there, but we're still building tightly
coupled -- what?

[THIS IS A REALLY GOOD BIT]
aha, that's what: he's building a language where you can't label anything else.
all you can do is emit and accept, so you emit stuff, and if something roughly
can deal with it, it accepts it. wow!

there's encapsulation in biology: inside cells lots of complexity, but only a
few proteins which go around carrying things.

idea of programmer might have to go away
trying to find a way for anyone to create programs

ah, the phone operator analogy [where is that from?]: we're all operators now.
and that's why having many programmers is a good thing. but programmers with a
different definition.

benjamim j whorf 1897-1941: the language you use determines the way you think.
he's a linguist. man says: linguists have generally decided this isn't true. but
it *is* true for programming languages.

so he's saying we don't necessarily need a language to do biologic computing,
but it might be useful if we had one to channel our thoughts in that direction.
and maybe the big revolution will happen then.

what we have to give up to get there:
- the idea that performance is what we're after
- that software is repeatable/predictable; it isn't anyway
- human readability, we won't debug natural systems in the same way
- unambiguity, the computer says "yeah maybe"

he says we're entering this world *anyway*, so we might as well be aware of it.

finally:
people are part of nature. we should learn from their social interactions.
computers can be more human, not just more natural.

[again: what problem is this solving? um. i think he is taking an aesthetic
judgement about where the next big leap is]


====

rumblings:
- some people gather domain knowledge for the purpose of collapsing that
knowledge into a small information gestalt.. kind of what science is. what
guiding principles are there for what extra pieces of knowledge to go after?
that's the scientific aesthetic.
- interconnected anecdote, jesus, king of kings, star trek, bombs.