[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: fuzzy logic (was: scalar polarity)

On 951123 "Peter L. Schuerman" <plschuerman@UCDAVIS.EDU> writes:

> The problem with this has to do with the conceptualization of truth as
> having "degrees."  It simply doesn't.  I know the "fuzzy logic" folks must
> think I am very quaint and atavistic for saying this, but I would
> challenge them to give an example of a *real* situation in which
> conventional logic cannot be used to arrive at a solution, or a *real*
> situation in which the terms "true" or "false" cannot be applied.

I probably shouldn't add to net pollution, and yours is the 50th of 225 messages
in my box most of them from the Lojban list, so I can't tell if this example has
been given already, but here goes...

Probability is the clearest example where the fuzzy logic model can be useful.
In the casino, we know that of a large number of suckers participating, well
half will be fleeced (i.e. win less than they bet), and also that on successive
days on which one specific sucker gambles, the losing days will definitely
outnumber the winning days, so that the casino operator can say without
recourse to fuzzy logic that his gambling hall will turn a profit.  But one
sucker asks, "will I win today?"  There, a yes-no answer is certainly
impossible; suckers do win, and it isn't even rare.  The correct statement is
that "I would win more than I bet on 43% of the trials".  (Obviously
oversimplified, as most of the bets aren't 50:50.)

The rules for combining probabilities are well known, but a neat and intuitive
way to express probability statements would be a big help to language speakers.
And's {xoi <number>} idea looks attractive, particularly with a quasidigit
such as {piso'i} (a lot).

James F. Carter        Voice 310 825 2897       FAX 310 206 6673
UCLA-Mathnet;  6115 MSA; 405 Hilgard Ave.; Los Angeles, CA, USA  90095-1555
Internet: jimc@math.ucla.edu (finger for PGP key)