Re: SUO: Re: Sign Relations & Communication
From: "Cathy Legg" <clegg@cyc.com>
> Fair question. To me, seeing something as an interpretation involves
> some kind of possibility that the interpretation might have gone another
> way. So for instance, I interpret a given cloud as looking like the
> Sydney Opera House, but others may disagree. If, say, someone meddles with
> my brain and plants the idea that the cloud looks like the Sydney Opera
> House - this is not a piece of interpretation on my part, this is my being
> causally determined.
Hmmm ... it seems to me that whether an interpretation is casually
determined (or not ) is a philosophical viewpoint which could be chosen
either way. Perhaps this is a place where we would need to diverge to
different SUO modules ... one where an Interpretation is the physical state
of affairs of a particular set of formulas which state is subject to cause
and effect, and another where it is an ideal abstraction immune to such
effects. Personally I prefer to peruse the former, but I would not want the
IEEE to exclude the latter.
> So my problem with assigning the term 'interpretation' to certain computer
> processes is their determinism (as long as nothing is broken, though of
> course if it is broken there is no interpretation either, just a big
> mess). Perhaps this is a naive John-Searle-ish philosophical take on
> interpretation, and I'm open to discussion about it.
But any computer's interpretation *could have* gone a different way too ..
change the database, or program, or the state of the program, just one bit,
and the response meaning of some stimulus (= database's interpretation)
would change too. And since the database+program+state can be an open
system amenable to change, so too the interpretations that it can generate
will change. Human minds and running-computers+databases+programs+states
are sitting in the same slot of the predicament. You may not prefer that
viewpoint, but I doubt you can disprove its logical possibility.
An interesting test case might be a Necker Cube ... the human mind chooses
to see it one way or the other ... so too could a
running-computer+database+program+state. The logically possible
configurations of the same diagram can change (not only with the data) but
also by where a computer process chooses to place its viewpoint and\or its
state. Do you doubt that we could contrive a database+program that could
spit out a string specifying in which state its internal representation of a
Necker Cube existed at any point in time ?
http://dogfeathers.com/java/necker.html
> > All I'm saying is that if you can take a system that is composed of
> > (database + programs) and design it to respond to strings of language
signs,
> > then the (database + programs) is indeed sitting in the Interpertant
slot of
> > Pierce's sign relation.
>
> Why? I don't see this. You might argue that the *responses* are 'sitting
> in the Interpretant slot', but I don't see why the database + programs
are.
I would put ~the responses~ potentially in the sign slot for other signing
relationships. These ~responses~ are physical events located in time and
space ... and as such they are merely marks {(a mark) (not isa)
sign.}available for potential interpretation by other signings. We might
want to decide whether we want to consider a Pierciean Sign Relationship
(PSR) a physical process (or not) and to actually try and locate it in the
SUMO ontology. Perhaps someone else will point out prior trains that
attempted to accomplish this ... to my knowledge those trains never
converged because I still can't find it in the database. Once we decide
on the KIF (and\or CycL) statements to express PSRs, then perhaps we can
determine whether a running-program+database+state could logically exist in
the domain of its interpertant slot.
> > Perhaps it doesn't feel right to you because the Interpretation is then
> > subjective to the particular (database + program) and is not some
objective,
> > mathematically ideal, context insensitive thing.
>
> No, that's not my issue.
So you say, but then you declare that your Interpretation must *not* be
subject to cause and effect. Or did I misread you?
> Seth, I can't speak for Cycorp managers (of which I am not one) and their
> decisions. However, I know people *are* working hard on OpenCyc at the
> moment.
Do you know if OpenCyc will be released webized or not?
http://www.w3.org/DesignIssues/Webize.html
... in any case, thanks for the dialogue ...
Seth Russell