4 Ocak 2010 Pazartesi

frivolous heuristics

but why this rule and not the other?
_some of the heuristics might have resulted from some particular practical expediency, thus might be explained explicitly.
_but the only reason for some of them to be chosen and be adopted so fervently might be just that they are not apparently useless, and that we need some concrete direction to move on; just moving on: an abstract scheme of practicality. you don't calculate every outcome before moving on, you are constantly moving on, so in a way continuously scanning a region of potentialities, and not the other regions, and that's all.

when it comes to architectural design, we have both types of heuristics, but it's hard and sometimes unnecessary to identify, in terms of this distinction, a particular heuristic that we frequently utilize. indeed most of our 'global' heuristics that we express as "i like the idea", or "this doesn't seem right," without explicating the reasoning mechanisms (which most of the time elude us) are quite versatile and multifaceted, in that, they, at one swoop define a whole strategy and direction that enable us to move on. it seems that we have global strategies stored as 'patterns'. we feel/experience them as ideas, feelings, tastes or just frivolity; we like to seem strange sometimes. operating in a world like ours, our computer programs should incorporate frivolity and gratuitousness, this is not to exalt free will, this is practicality.

4 yorum:

kakadece dedi ki...

tamsayı serilerin tanınması / çözülmesiyle ilgili (yani: 1,2,4,6,3,7,8 ... sonraki sayı nedir, ya da bu serinin kuralı nedir türünden prolemler) çıldırtıcı sıkıcılık ve uzunlukta bir pasajın ardından (bu problemleri çözen bir yapay zeka uygulaması üzerine çalışıyorlarmış da zamanında) hofstadter şunu yazmış:

"Despite their protestations, though, mathematical sophisticates are often
the very first to spot patterns, both simple and complex, in a Seek-Whence
puzzle. And often they are the only ones to find the deepest patterns. Like
everyone else, mathematicians have an unconscious esthetically-based sense of
which kinds of extrapolations are more appealing or more plausible than
others. Indeed, when working professionally, they make full and unashamed use of their esthetic intuitions about patterns, in the act of guessing which pathways would be best to explore in inventing new concepts, making hypotheses, devising proofs of theorems, and so on. In mathematics, pattern sensitivity
acts like an abstract sense of smell, and informs all of the hardest decisions.
In fact, mathematics books and articles are filled with appeals to the obvious" extension of a pattern. Readers are told that the proof of the general
case is left to them, after they've seen a proof of one specific case. Or they are told that the other cases in a proof follow "bysymmetry" or "by analogy", where the symmetry or analogy is presumed obvious. An infinite sequence (such as the integers running down the diagonal of a continued fraction) or infinite series (such as a Taylor series or Fourier series) is indicated by giving the first few
terms, and then writing three dots to indicate that the rest is "obvious". An
infinite diagram is presumed to be clearly communicated by showing a small
typical piece and then using some kind of two-dimensional dot-dot-dot notation
to suggest the rest. All over the place, one is asked (subliminally) to buy into other people's sense of pattern. In effect, an "objective" or "natural" esthetics of pattern is presumed by the waymathematicians communicate." (fluid concepts, creative analogies, d. hofstadter p70)

ne kadar tanıdık geliyor. genel olarak zekanın temelinde analojilerin ve örüntü tanımanın yattığını düşünüyor hofstadter. araştırmacı da öyle düşünüyor mu?

gönülsüz dedi ki...

araştırmacı en başından beri bu fikrin peşinde sayın kkdc

kakade dedi ki...

aynı kitap p86:

"It is my firm belief that pattern perception, extrapolation, and generalization are the true crux of creativity,and that one can come to an understanding of these fundamental cognitive processes only by modeling them in the most carefully designed and restricted of microdomains."

gonulsuz arastirmaci dedi ki...

şu anda ne o kadar tanıdık geliyor ne anlamlı... kaldırsam mı o yorumu??