proof-planning

Now that I have the motivation to build my own proof-planning system, Alan Bundy's Proof Planning FAQ makes a lot of sense, and has tons of cool meta ideas, such as proof-critics and meta-tactic generation. Btw, I did reinvent proof-planning last month, only 16 years too late (this is narrowing... my last good idea was >50 years too late) (although I might have been inspired by the word "proof-plan" to think of AI Planning, I didn't know what it really meant)

The point is to think of the tactics (i.e. steps of the proof-plan) as common ways of getting closer to a goal, with specifications (e.g. if the goal is to solve an equation for X, a common tactic would be to bring all terms with an X to the 'left-hand-side'). Some tactics may have guaranteed post-conditions, while some may guarantee nothing: while 'collect-var-on-lhs' is guaranteed to leave no X's on the rhs, 'try-to-factor-out' is not guaranteed to narrow down the X to just one factor (e.g. in the case when the lhs is X^2 + X).

Btw, I got my rejection letter from Edinburgh yesterday. I might be able to get a proposal funded there in the future, but it won't be this year.

Anyway, this stuff is delicious cognitive AI, or as Bringsjord would call it, "Human-Guided Logic-Based AI", because it's about analyzing human intelligence: perfect for an introspective, navel-gazing phenomenologist like myself (I suspect that the most reliable symptom of this ailment is being into linguistics).

typography and calligraphy

I did post about this stuff before (damn, can't find the link, despite LJ... what a pain that LJ isn't searchable), but now I've found an image of Dutch handwriting

Notice the "8" on top of the page. The g's and a's also strike me as unusual, although I have seen this "a" a lot around here.

Why is it so hard to buy cheap, practical clothing?

Why is it so hard to buy cheap, practical clothing?

Because we're crowded out by fashion consumers, who are willing to pay a lot for their clothes? I'm not sure anymore, since you might say that the fashion market has nothing to do with the cheap clothing market.

Anyway, I went to the Albert Cuyp market this week, and managed to buy 6 T-shirts for 10 Euro. But I wish I could have done this at my local supermarket, instead of going to the other side of the city.

By the way, I've always been sensitive to clothing (e.g. labels rubbing on my back), and I hate finding out that models I wear have gone out of fashion and are no longer to be found (e.g. some kinds of slippers, underpants), because it means I have to readapt my skin or, if pockets are in the wrong places, readjusting my habits. I wonder if there's a market for models that are promised never to stop selling: it would suck to depend on a tailor for the rest of your life.

Ok, I'm exaggerating. It's not that bad.

What really sucks is having to look around and try things out, wasting away my time and patience. Can't I just order things the right model and size from home? Is biometrics that far from application?

ADDENDUM: MR on Vanity Sizing

LispBlog

Given a test and a list, BIPART divides the list in two: the ones which pass the test, and the ones which don't.

;;;I want to implement such a bipart function functionally: no PUSHs, and no LOOPS

;;using 'select' and 'not-p' from my personal LISP library, it's easy to come up with:

(defun bipart (test l)
  (list (select test l) (select (not-p test) l)))
;;PROBLEM: inefficient: will do each test twice

( Read more... )