Re: D1. Separate computer science ontology from philosophical ontology
John,
As a rule of thumb, let me say I will always be wary of arguments
which make bald statements about what something "is" or "is not."
These are staples of on-line discussions, and in my experience lead
only to ever more narrow definitions, which eventually become
meaningless. Or perhaps it is better to say they become so specific
that their original meaninglessness eventually becomes clear.
If you say Wittgenstein's games can be objectively recorded, we first
have to argue about the meaning of objective (of which I think there
are not a few already in the SUO archives.)
Similarly the statement that "context dependence does not imply
subjectivity" needs a definition of subjectivity, and context
dependence, before it means much.
By which it should be clear that, no, I do not submit to the authority
of dictionaries. They are in my experience very poor guides to
"meaning." Most admit as much by filling themselves with examples.
Rather than bald statements about what things are, I prefer statements
about how things behave. Relationships among words to some extent
avoid the problem of what words mean and provide content beyond
dispute. Statements such as that the nature of truth necessarily
involves its own contradiction, contain meaning which is to some
extent independent of the constituent words.
So I would rather hear statements about things, not assertions of what they are.
Anyway, this is philosophy. It seems people have been arguing about it
for years and the point of view we agree (mostly) about won.
This thread is not about that. It is about whether we should ignore it.
My message is that we don't need to, because the hottest area of IT
practice is moving in accordance with this conclusion of philosophy
anyway. Ad-hoc labels are the ones proving useful, and people have
gone from ad-hoc labels, to ad-hoc hierarchical relationships. Maybe
theory could suggest more.
As I say my own interests tend more to the complexity underlying all
of this. I think a more complete solution is possible working from
language. Defining meaning from first principles in terms of
relationships among words to get "meanings" which vary with context
(and defy any single summary because they imply their own
contradiction.)
Either way, my answer to this thread is: don't separate computer
science from philosophical theory. If for no other reason than because
you don't need to.
-Rob
On Jan 26, 2008 2:30 AM, John F. Sowa <sowa@bestweb.net> wrote:
> Rob,
>
> I agree there is a lot that we agree about.
>
> But statements like the following raise serious
> disagreements:
>
> RF> Since the discussion (as it has developed) is all
> > about our philosophical inability to find objective
> > meanings....
>
> There most definitely are objective meanings. Every
> major dictionary, such as the OED or the Merriam-Webster
> Third, is filled with objective summaries of how people
> have been using many thousands of words.
>
> RF> We need to stop looking for things which are true or
> > false independent of contextual premises.
>
> With suitable qualifications, I would agree with that.
> But context dependence does not imply subjectivity.
>
> Wittgenstein's language games, for example, are highly
> context dependent, but any particular game can be
> objectively recorded and summarized in an appropriate
> set of rules.
>
> One reason why NLP is so difficult to automate is that
> people can freely invent new language games, and they
> can mix and match different language games at any time,
> even within a single sentence.
>
> I would grant that any new language game that anyone
> invents is invented for some subjective reason. But as
> soon as two or more people reach an agreement on how to
> play a game, the ways in which words are used in that
> game are objectively observable.
>
> John