Re: performatives and trust

>pat hayes wrote:
>
> > No, I did not mean to imply that it was like HUMAN telepathy. I meant
> > that the kind of internal/external contrast that is so central in
> > human communication (as when you are not sure what someone is
> > thinking in spite of what they are saying) seems to be missing for
> > web ontology agents: they can see it all, as it were, and what they
> > see is all there is.
>
>Ah, I think that this last is a source of significant miscommunication among
>us.  Web ontology agents *can't* see it all.  I mean this not just in
>practice -- there's too much all out there for anyone to pragmatically
>collect it, altavista/google's attempts notwithstanding -- but in principle
>-- it's constantly changing, growing, customized and generated-on-demand,
>and there is and can be no sensible notion of "all", even limiting yourself
>to a particular moment in time.
>
>I think that Pat will fundamentally disagree with what I just said, and I
>expect that most who come from logicist backgrounds will join him in this.

No no, I have expressed myself badly. That isn't what I meant at all, 
and I entirely agree with what you said in the previous paragraph. My 
interest in the logical issues that arise from this 'open-ended' 
quality of 'web logic' is what got me into the business in the first 
place. I might add, however, in response to your quick assumption 
about my narrowmindedness, that the idea that the "logicist 
background" is somehow restricted to narrow, confined, closed worlds 
is itself a profoundly misleading impression (which might arise from 
thinking of logic only in the context of logic programming?); 
philosophical logic, for example, takes almost all of metaphysics 
into its scope, and has its roots in an analysis of the meaning of 
human language, which is just as open-ended and generated on-demand 
as the web. One of the beauties of Tarskian model theory is precisely 
how very open and agnostic it is about meanings, how few restrictions 
it places on reasoning.

What I meant was that once one has access to a URL where an ontology 
is located - a web-site containing some DAML, to make the example 
concrete - *all* the DAML on the website is fully and immediately 
available, and moreover, there isn't anything else 'behind' it, in 
contrast to the thoughts, intentions, desires and so on that give 
rise to speech acts. If we think of ontologies as like formalised 
knowledge (most ontologies seem more like formalised thought than 
formalised speech), then the prototypical communicative act is more 
like a direct transfer of mentalese than a speech act. It might make 
sense to decide that a DAML ontology was mistaken or false, for 
example, but it simply wouldnt make sense to decide that the web page 
was lying to you, or deliberately not telling you everything it knew, 
or trying to confuse you, or being sarcastic, and so on.  About the 
only performative acts that can be done are to ask for the content, 
and to receive it, in its raw unmodified form. (The person who wrote 
the DAML might have been lying, etc., , but that is more like the 
creator of the thing involved in the speech act, than the actor.)

The subject line refers to 'trust'. But, to emphasise the point, we 
are using that word here in rather a limited sense. One can have more 
or less trust in an ontology in the sense that one has more or less 
confidence that what it asserts is in fact true, but not in the sense 
that one has more or less confidence in its sincerity.  Web resources 
aren't the kind of things that have psychological states or take 
social positions.

Its a small point, maybe, but I made it in response to Tim Finin's 
pointer to work on speech acts in this context. I think that the 
analogy of web agent communication to human communication is often 
taken for granted; but that it should be looked at more carefully, as 
it isn't as exact or as automatic as is often assumed. Even on the 
semantic web, software agents really aren't much like human agents, 
and they really do not have a social life, all the hype 
notwithstanding.

>On the other side, I think that Tim and others from the web world are likely
>to agree with me.  I suspect that a certain amount of where we get lost with
>one another derives from this fundamentally different base assumption.

I don't think so, as I think we agree on everything you have said 
about the web. In particular, you should not assume that a claim that 
(for example) web logic is pretty much like non-web logic in many 
ways, is necessarily a symptom of a too-limited view of the web. It 
might be a symptom of your too-limited view of logic :-)

>I could say more about the rest of Pat's paragraph, but I think it would
>only stir up more trouble :o)

Actually I would be interested in them, and on the above, as the more 
we can come to understand one another, the better, but by all means 
take them offline if you want to keep the waters calm.

> > There aren't any other thoughts behind the ontologies, and they don't have
> > to worry about whether the ontologies
> > are thinking one thing and saying another. If ontology agents were
> > humans - which of course they are not - we would call this telepathy.
> > I probably shouldnt have used such a poetic way of expressing myself.
> > Sorry if I caused confusion.

Pat Hayes

---------------------------------------------------------------------
IHMC					(850)434 8903   home
40 South Alcaniz St.			(850)202 4416   office
Pensacola,  FL 32501			(850)202 4440   fax
phayes@ai.uwf.edu 
http://www.coginst.uwf.edu/~phayes

Received on Wednesday, 6 June 2001 22:07:16 UTC