RE: Some TAG review of "Cool URIs for the Semantic Web"

Pat Hayes:

| >This raises a question for me. Suppose I have no clue about 
| >webdesign, and serve
| >a picture of me from:
| >
| >
| >
| >1 October 2007 I change my mind, and serve my CV from
| >
| >
| >
| >and the picture from
| >
| >
| >
| >What would you now say:
| >
| >[1] is an IR which 
| denoted Marc's picture
| >before 1 October 2007 and Marc's CV after 1 October 2007.
| Better, which denoted a thingie which looked (?) like the picture 
| until then and the CV after then. The problem with this isn't the URI 
| or the denotation, but the fact that this is a very odd thingie. Its 
| naughty to create such thingies, they confuse people and resist 
| normal classification. 

Agreed. But URI's change ownership, so what is now my blog could be a furniture
catalog tomorrow (not that imaginary, since sells
furniture indeed). And that does open the question whether it is the same IR
before and after changing hands.

| But consider
| html?wpisrc=newsletter
| which gets me something different every day. We don't have a problem 
| with that, yet its always "changing". Its OK because we have a 
| category for this kind of changing thing, its a 'daily column', and 
| we know about those and can work with them. Similarly for webcams and 
| so on. But pictures that morph into CVs aren't in our conceptual 
| space, so they bother us. When I moved from England to upstate NY I 
| had a hard time at first with things that went back and forth between 
| a clock and a thermometer, but you get used to them eventually.
| >Or:
| >
| >[2] Marc, you have no clue about web design, you served a 
| different IR before
| >and after 1 October 2007, go read Tim Bernes-Lee's "Cool URI's don't 
| >change" [3]
| >and mend your ways.
| I don't think this is really to do with Web architecture as such. The 
| problem arises here because Web technology makes it so easy to create 
| new 'kinds' of document-like things that behave in all sorts of 
| irrational ways, like images that change into text for no reason.
| >I always assumed IR was intended as a sort of refined notion 
| of "document" in
| >[3], and I would thus say [2]. Webarch suggests as much.
| What REST says is, an IR is a function from times to representations. 

Fielding wrote in his thesis: "REST accomplishes this by defining a resource to
be the semantics of what the author intends to identify, rather than the value
corresponding to those semantics at the time the reference is created." and
"Defining resource such that a URI identifies a concept rather than a
document...". True, the time thing is there as well, but there is more. I think
function of representations over time is fine *as long as the URI does not
change in the "Cool URI's" sense*, and we just need some vocabulary to cover
cases where the URI does change hands or is badly handled by its owners.

| Of course that allows for either answer, but it sure allows [1]. I 
| think that in order to get to the intuition that justifies [2], you 
| have to ask, "representation of what?" and ask if that is the kind of 
| thing that should be represented.
| BTW, philosophers have been here. There is a famous example by 
| Goodman called grue/bleen. Grue means: green up to 1 October 2007, 
| then blue. Bleen is the opposite. Question: are these colors as good 
| as green and blue? Seems obvious they are not. But we can't base this 
| intuition on the form of the definition, because if we had been 
| brought up to like grue/bleen, then green is defined as: grue up to 1 
| October 2007, then bleen; and similarly for blue. So as far as 
| definitions are concerned, there is perfect symmetry here. So if my 
| lawn suddenly turns blue, what justification do I have for saying 
| that it changed color? It was just a grue lawn all the time.

I should have given him credit, since this indeed inspired my example.
| >
| >I think saying [1] may be coherent, but then we still need 
| some notion of
| >"concept" to replace "IR" in [2].
| AArgh, don't go there. This is taking the entire bottom off the can 
| of worms. At the very least, call them something like 'documents'.

Fine with me. I don't think Web Architecture should require peeking in the heads
of people either.



| >Computers can't do much with it, but for
| >people it is often extremely easy to distinguish them, and sentences 
| >such as [2]
| >are pretty essential to (improving) the Web.
| <aside>
| I wonder, does the Web really need improving in this regard? Does it 
| even make sense to try coining URIs that will last for 2000 years? 
| Given the results I often get from Google, seems to me that part of 
| the problem with the current Web is that many URIs last TOO long. The 
| Web is cluttered with instantly-accessible information that is now 
| outdated and often false (people's old addresses, old blogs that are 
| archived, etc..). Might be better to allow quite a lot of URIs to 
| just die quietly, unremembered and unmourned. Let the second law of 
| thermodynamics do its gentle work here, as in the rest of the 
| universe. It takes less energy, anyway.
| </aside>
| Pat
| >
| >Marc de Graauw
| >
| >
| >
| >[3]
| -- 
| ---------------------------------------------------------------------
| IHMC		(850)434 8903 or (650)494 3973   home
| 40 South Alcaniz St.	(850)202 4416   office
| Pensacola			(850)202 4440   fax
| FL 32502			(850)291 0667    cell

Received on Friday, 5 October 2007 15:21:42 UTC