W3C home > Mailing lists > Public > www-tag@w3.org > July 2003

Re: resources and URIs

From: pat hayes <phayes@ihmc.us>
Date: Wed, 23 Jul 2003 10:53:07 -0500
Message-Id: <p06001a0ebb44535be7cf@[10.0.100.23]>
To: Walden Mathews <waldenm@optonline.net>
Cc: www-tag@w3.org

>  > >  > OK, then in the sense I was understanding the word "on" here, its the
>>  >>  *representation* which is on the network, not the thing it is a
>>  >>  picture of.  But I realize now that others are using 'on the Web' to
>>  >>  mean what I would phrase as 'referred to somewhere on the Web'. The
>>  >>  problem for me that is not that this is meaningless, exactly, but
>>  >>  that it is useless. Being 'on the Web' in this sense isn't
>>  >>  well-defined, cannot be checked for accuracy, provides no
>>  >>  architectural or semantic content.  So OK, I don't give a damn
>>  >>  whether something is or is not 'on' the Web in this sense, and I see
>>  >>  no reason why I or anyone else should give a damn what the TAG group
>>  >>  thinks about it either, as it makes no difference to anything.
>>  >
>>  >Pat, above you seem to use 'network' and 'Web' interchangeably;
>>  >is that intentional?
>>
>>  I was being somewhat careless about the distinction, I confess. I
>>  assumed we we talking about the Web which is a kind of network.
>
>Have you figured out "what kind" of network it is?

I would defer to others more expert than I on that topic. There is a 
way to read the REST description which makes perfect sense to me as 
an idealized description of the Web architecture, though one has to 
tread carefully to get through Roy's section 5.1 :-)

>  It's tempting to
>say that the Web's nodes are representations and its arcs are something
>composed of URIs and the method of "following" them [not sure how
>to word that], but that definition lacks the mutability and indirection of
>the modern Web.  How would you, then, correct the model?

I like the idea that the nodes are '(re?)sources' of representations 
which are 'accessed' by the use of URIs. That makes less sense for 
URNs, but I think the whole idea of conflating URLs and URNs was a 
mistake in any case. (Not conflating their syntax into a common 
model, but conflating their semantics into a common referential 
framework: that was just a fundamental, disastrous semantic error, 
like declaring that north is the same as up.  We have been trying to 
clean up the resulting mess ever since.)

>  > >If you told me there was a random bag of representations available
>>  >on the Web, I would't give a damn.
>>
>>  Wouldnt that depend on what they were about? This isn't a bad
>>  description of quite a few blogger pages, for example.
>
>I meant truly random, in the sense that there could be no collective
>'about' about it.  My interest would severely diminish.

Well, OK. But still, it would be possible, you agree? Nothing in the 
architecture would break or be violated. And in fact I might find it 
strangely fascinating if it were done with some flair.

>  > >But if you told me there would
>>  >be a set of representations over time which express the state of
>something
>>  >of interest, then I would.  How is the architecture supposed to indicate
>>  >that certain groups of representations (maybe over time) "hang" together
>>  >meaningfully?  Or is that not the job of architecture?
>>
>>  Good question. I guess it depends on what counts as hanging together
>>  meaningfully.  I would say that this sounds more like a semantic
>>  condition than an architectural one, though to do it full justice
>>  would probably need both.  One way to say it might be that there is
>>  something which endures and which the representations accessed at a
>>  time are in some sense 'about' or 'of' the state of thing at that
>>  time, ie there is an implicit 'now' in the semantics of the
>>  representation that is tied to the architecturally-defined notion of
>>  the time of the transaction which produced the representation.
>
>That "something which endures" ... not a "resource"?

Maybe it is. OK, lets run with that idea: a resource is a thing 
referred to by the representation and which endures through time, 
indexed to the time of accessing the representation.  (Quite a lot of 
web pages now have no resource at all in this sense, but lets run 
with it.)

>If the representation is "of" that thing, then is it reasonable to say that
>the
>thing is a "source" of representations, even if it doesn't physically "emit
>the bit", so to speak?

No. Take the remote-galaxy example, or Sherlock Holmes. Of course 
what the representation is about isn't the 'source' of the 
representations in any kind of network architectural sense. In fact, 
its extremely unlikely that this would ever happen, though I guess it 
could, eg a webcam looking into a mirror, maybe.

>And then the final step of the composition would be
>
>     source + time = resource

I have the sense Im being led carefully into chapter 5 of Roy's thesis.

>  > Im
>>  thinking here of something like a webcam (Roy Fielding's example)
>>  which when pinged emits a picture out of window, it being the view
>>  'then' of the 'same' scene, or the current weather in some place
>>  illustrated by a recent radar image.  Cases like this also fit the
>>  current 'single denoted resource' account very well, but I think its
>>  simplicity fails to do them justice, in fact.
>
>Do you have an example of that failure?

Well, take this case. What *exactly* is the constant resource here? 
Is it the webcam itself, or the view from the window, or (suppose the 
webcam is in Peoria) Peoria? Ive seen discussions like this quickly 
dissolve into into incoherence because there is no way to even make 
sense of the question (All of them; What difference does it make? 
Whatever you like, its up to the owner of the webcam to say what it 
is, stop asking silly questions, the Semantic Web will decide things 
like that, etc.) Just in passing, when you have to do semantics 
seriously, it does make a difference.  If we are strict about 
resources being network entities, the answer is clear: its (some 
suitable abstraction of) the webcam. That is what is attached to the 
network and responds to requests for information transfer. It emits 
representations OF a view in Peoria, true; and the owner of the 
webcam might indeed declare that his site is about Peoria; but that 
is a semantic claim, not a network claim.  People might be 
disagreeing about that claim while the network is happily working, 
oblivious to their semantic debates; and as I have tried to explain 
on other threads, if you start doing semantics seriously there is 
usually no need to fix on a single referent in any case: that idea is 
left over from the good old days when "resource" meant 'addressable 
network node'.

>  > But consider another case of an exotic art site which when pinged
>>  chooses a random picture from some large set of sources and sends it
>>  back. No coherence or rationality at all: that might be the artist's
>>  point. But *architecturally* these seem the same to me: when pinged,
>>  they deliver a picture. The 'state coherence' is a semantic notion,
>>  even though it refers to an architecturally described time parameter.
>
>"No coherence or rationality" is at odds with "exotic art" above.

My dear fellow, have you been to many galleries lately? (Seriously, 
artists are exhibiting things like this already.)

BTW, I meant that the site itself was the artwork, not that it was a 
site *about* exotic art, like a catalog or a dealer.

>Does 'state coherence' denote something other than 'identity'?

Oh my goodness yes: that is the very point. It doesn't depend on 
anything as simple as identity (identity of *what*??) For example, 
consider a website which displays a new chapter of a novel each week.

Pat


-- 
---------------------------------------------------------------------
IHMC	(850)434 8903 or (650)494 3973   home
40 South Alcaniz St.	(850)202 4416   office
Pensacola			(850)202 4440   fax
FL 32501			(850)291 0667    cell
phayes@ihmc.us       http://www.ihmc.us/users/phayes
Received on Wednesday, 23 July 2003 11:53:11 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 22:56:00 UTC