::scr Re: Cognitive Friction

simon wistow scr@thegestalt.org
Wed, 2 Jan 2002 11:09:21 +0000


On Wed, Jan 02, 2002 at 08:42:25AM +0000, Andy Wardley said:

> Unfortunately, many people simply apply a methodology expected it to
> design them a nice house or functional piece of software without
> having to engage any grey matter of their own.


But isn't that what programmers do all the time? From my lectures on
Turing machines I seem to remember them being defined as something that
could run an algorithm (as well as the lengthier explanation with
infinite tapes, heads, tuples and alphabets) i.e a methodology. 

We write programs that follow a methodology - there's nothing else a
program can do - to perform a task. As the task gets more complicated
we add more and more gotchas and exceptions - we try and distill our
knowledge into code.

Which is why we have problems doing stuff with computers that humans
find easy - image recognition is Hard [tm] as is, in fact, any form of
recognition. Where we don't control the input then things get tricky. 

Which is where AI comes in. And the three camps of AI are that we shoudl
strive to build a human like intelligence, that we should build
something that is intelligent but not like a humna (in the same way a
plane can fly but not like a bird) and that we shouldnt build generic
AIs and just build specific task ones.

Of course then there'es the fourth camp that says that AIs are a bad
thing and that one day they'll rise up, enslave us and make us their
human bitches but then they've never read Asimov.

Anyway, sorry to drag the tone down (you bring up Christopher Alexander,
I counter with Slashdot) but this a ranty but interesting post on
methodologies, people hacking, cheating and MMORPGs (yes, games again,
cling to what you know Simon :)

http://slashdot.org/comments.pl?sid=10406&cid=448344

Simon