Re: use/mention and reification

Comments below.

Martyn Horner wrote:

> Frank Manola wrote:
> 
>>Dan Brickley wrote:
>>
>>
> ....
> 
>>>In which case, the example has drifted somewhat from my original intent,
>>>which was based on their being two names for one thing-in-the-world, ie.
>>>some wierd guy who's kinda strong, sometimes wears glasses and sometimes
>>>wears a blue and red leotard.
>>>
>>>The point isn't that "lois's idea of superman" and "lois's idea of clark"
>>>are distinct entities worthy of our concern. I was trying to make a much
>>>more mundane and (I'd hoped) less woolly point. Lois lacks complete
>>>information about the name-to-world mappings. Her views, messages, diary
>>>entries and (we ought to steer clear of this) mental states will all be
>>>affected by her lack of the 'complete picture'. On the Web, we have a
>>>similar situation: no one document or agent has the whole story. Often
>>>they're wrong, or lack information. The partial information aspect of this
>>>is my main concern: if *everyone* had faultless access to the meaning of
>>>each and every URI name, I wouldn't have my current concerns about
>>>reification.
>>>
>>>Dan
>>>
>>I agree 100%, which I guess means I didn't understand something about
>>your original point.  Specifically, what does quoting the URIs in
>>reification have to do with addressing this?
>>
>>
> 
> I think DanBri's point is the key one. Lois lacks the information about
> the identity between the Clark and Superman URIs. In her interpretation,
> there is no conflict in believing `A is strong' and `B is not strong'.
> She can believe what she likes and, if there are no correspondences to
> show she is inconflict with herself, she can remain sane.
> 
> Kent Bach in http://userwww.sfsu.edu/~kbach/puzzle.html talks about this
> puzzle and gives several ways to deal with it. He points out the Perry
> White (the editor of the planet) is not entitled to hold these two
> beliefs because he knows about the identity (as we do). I favour Bach's
> `metalinguistic' approach: although Lois apears to be believing stuff
> about `our' URIs she lacks some knowledge about their identity which
> would stop us believing these two assertions. 


That's what I understood about the *situation*.  It's the syntactic 
distinction I'm having trouble with.


> 
> The reification of her beliefs must be understood as having its
> components interpreted according to Lois' (the subject of the
> reification `Lois believes...') knowledge. I think this indicates that,
> at a level of implementation in software, it is `better' or `safer' to
> maintain that the reified statement refers only as far as the literals
> (in their lexical space) and not to an interpretation made possible only
> by the reader's knowledge (hence DanC's pragmatic stance). To
> accidentally follow the interpretation through to the set of properties
> which have to hand regarding Clark and Superman would be to
> misunderstand the context in which Lois believes these two things and to
> misunderstand the reifification. It would also come up against a
> conflict which does not appear in Lois' mind.
> 


This last section of your comment is getting to what I wanted more information about.  

(NB:  the brackets in the examples are intended to indicate a "reification boundary";  
I'll write out the separate subject, predicate, etc. triples if that would help.)


Dan's example seemed to suggest that a piece of software, having

   lois accepts [superman is strong]

and being aware that superman and clark are coreferring terms, would go 
ahead and substitute one term for the other, determining that

   lois accepts [clark is strong]

ignoring the fact that this substitution must place *inside* a reified statement,
and inside the "context" (sorry!) of what "lois accepts", and that we 
have no explicit statement that 


   lois accepts [clark isthesameas superman]


(which would justify that sort of substitution).  The suggestion seems to be that
software would be less likely to perform such incorrect "reasoning" if it instead saw:


   lois accepts ["superman" "is" "strong"]


and I'm wondering why a piece of software might not just as easily 
"dequote" as "dereify" in cases like this.  I appreciate that there's a 
distinction between the two forms (and have no particular difficulty 
with adopting this "quoted" form), but I'm still wondering about how the 
two forms automatically convey the distinction you're having them make. 
  This particularly because I think we all agree that we can't assume 
universal agreement on all the properties of the things we talk about, 
even if we use URIs to refer to them.

--Frank

-- 
Frank Manola                   The MITRE Corporation
202 Burlington Road, MS A345   Bedford, MA 01730-1420
mailto:fmanola@mitre.org       voice: 781-271-8147   FAX: 781-271-875

Received on Friday, 25 January 2002 07:46:23 UTC