farm wuzzies

["John E. Clifford" <pcliffje@CRL.COM>]

One of the logic fights I most regret losing in the formation of Lojban --
in the transition from Loglan to, in this case -- is that one about
keeping most qualitative adjectives as inherently comparative, rather than
absolute or metrical.  That device (firmly grounded in natural languages,
by the way) would have spiked a lot of the nastier aspects of the fuzzy
discussion -- as well as being generally more useful.  It fell afoul of
the Lojban users' initial lack of linguistic sophistication and
controversy about the meaning of _lo'e broda_, the usual default value in
the "than" place of the comparatives.

McCawley does not (on a quick check) have much useful to say about "tall."
However, his remarks about "most" in contrast to "more than half" are
entirely relevant and apt.  I do not know what the average height of
American men, say, is nor do I measure the men I meet.  Yet my judgment
about who is tall and who is not is pretty much in accord with other
people (who don't refer to these putative standards either) and that is
all that communication requires.  I tend to know by the angle of my neck
talking to them and normal talking distance and that is, I think,
unconsciously ('til now) correlated with repeated observations of local
dominance -- a tall person is regularly one of the taller ones in a
variety of groups.  Of course, there is (maybe, "are") a jargon "tall" for
various health sciences and sociological checks and so on and that
probably is a metrical concept and related to means in the population.
And the membership curve may even make sense in that.  But, it should be
remembered that, after a certain point on those fuzzy curves, being taller
does not make you more fully a member of the tall-set nor make it more
true that you are tall.  All these sets have 1's.

Pretty much the same goes for the old sophistical arguments, sorites
("heap", not "polysyllogism") and pilatus.  If you buy the premise,
"Taking one hair/grain from a head/heap does not make it bald/a non-
heap,"  then the sophist has you, for you have admitted that counting
counts -- indeed that it is determinative.  The right thing to say is "No,
the presupposition of this claim is false" (for which Lojban -- a win,
that -- has a word).  Sometimes, taking one away does make it no longer a
heap, as when a trail cairn (fast four-stone version), loses its top stone
and becomes a mere ring (no longer heaped up).  At other times it does
not.  And, of course, it is not the change in number but the change in
arrangement that does it.  A man would be just as bald, if, like the eyes
of a halibut, his hair migrated from the top to the sides with nary a
strand lost.  And again, we can tell heaps and baldness without counting
or even having any idea how to go about counting or what the significance
of a given count might be.  Baldness does not relate to averages,
interestingly, but to ideal, since -- at least by my age -- most men are
bald.  Heaps don't relate to any of these things, but are about shapes and
the way the shapes are built up.  (All these points have been made since
roughly Aristotle but have been ignored or overborne by particularly loud
sophists repeatedly since, only to be rediscovered anew in each age.
Qoheleth was an optimist.)

As for hills and mountains, we all thought the Hugh Grant movie this
summer was a very funny absurdist comedy and the reason why it was absurd
(as opposed to the other kinds of comedies it was) was just the notion
that a couple of feet could really -- as opposed to "for purposes of His
Majesty's Ordinance Survey" -- change a hill into a mountain.  My favorite
altitude mountain story is of a friend writing me that he had just climbed
some famous mountain in New Hampshire.  He mentioned the height of the
summit which was about 10 feet lower than my living room (east of LA at
the time) and I did not even live on a hill (proven by the fact that the
subdivision nor the city did not contain a "hill" word in any langauge).

In short, on positive doctrines Peter is generally right (note, for
example, that the OED and AHD do not contradict his definition of "tall"
but specify a typical criterion of the sort he claimed was needed).  But
that does not mean that fuzzy operations of all sorts are not useful
things to have or that they do not correspond to things that speakers want
-- perhaps even often -- to do (though the fuzzy truth values are a bit
hard -- psychologically and historically -- to separate from confidence
levels, probabilities and likelihoods).  The addition of the information
about the Guttman scales, if nothing else, will be welcome to the List's
intellectual armory -- along with the (I hope) the warning against
applying the mechanics of one kind of scale to another, a charge which
might be leveled, not entirely unfairly, against some proselytizers for
fuzzitude.