W3C home > Mailing lists > Public > public-rdf-wg@w3.org > October 2011

Re: <sandro> PatHayes, can you formally define g-box for us?

From: Pat Hayes <phayes@ihmc.us>
Date: Thu, 13 Oct 2011 08:54:55 -0500
Cc: ian.davis@talis.com, public-rdf-wg@w3.org
Message-Id: <9BFB8DC8-F6EA-4027-BE13-E64ABB573F89@ihmc.us>
To: William Waites <wwaites@tardis.ed.ac.uk>

On Oct 13, 2011, at 3:38 AM, William Waites wrote:

> On Thu, 13 Oct 2011 08:50:27 +0100, Ian Davis <ian.davis@talis.com> said:
>    iand> This is a very interesting formulation of the web
>    iand> architecture. More generally one might say "when an IRI
>    iand> identifies a resource that IRI then also denotes the state
>    iand> of the resource"
> I find it hard to see how to wiggle out of mentioning time in the same
> breath as "denote" but maybe that's because of some lack of formal
> background in logic or philosophy.

Well, a brief summary. It is *usual* (in logical circles) to assume that denotation is 'timeless'. Assertions are true or false, period, not true or false *at a time*. The idea being that if you say soemthing true and write it down on a clay tablet and it gets dug up a long time later, it should still be true then. So for example if you write "it is raining' then thats not going to stay true, and if you write "it is raining now' that might be true but we have no way to know since we don't know when 'now' was, but if you write 'it is raining on 08/09/2011' then this stays true while time passes. Which is obviously better for communciation across times. So putting the "context' (or as much of it as necessary to fix the truth of what you are saying) into the assertion itself is a basic good rule for data which is supposed to last for a while and still be true. 

In a 'contextual logic', you dont assume timeless truth, but instead assume that there is a context in which you are asserting things and which can be assumed as implicit without being stated. We talk like this to one another, obviously, because it saves time. But if we expect information to retain its truth (or usefulness) when it is written in one context but read or used in a different context, then this style of implicitly using a mutually assumed context clearly isnt going to work. (Like using "now" on the clay tablet, even though "now" might be fine if you are talking to someone.)

One very common kind of contextual logic is presentism, which is the assumption that we are talking about the present, ie the same time that we are talking *in*. Again, we all do talk this way, in fact, of course; but it isnt a good way to manage data which is intended to be used over a long time period, or for reasoning using data from a range of different times. And, btw, it goves you a very peculiar logic in which truth is kind of dynamic and shifting, and ordinary inferences (such as that P&Q entails P) can go wrong. 

> Does this mean that to identify a
> resource means to denote all possible states of that resource at all
> times in the past and future?

No, just to identify the resource as an object that is extended in time, so that it is the same resource later and earlier as it is now.  Making timeless assertions does not mean being God.

> Or in other words can the IRI denote
> more than one state of the resource at the same time?

The IRI identifies the resource itself, which is (more or less) a function from times to representations. So for example I can identify you, but I can't predict everything you are going to say in the future. 

> Or does "denote"
> require an observer, the agent doing the dereferencing thinks that the
> IRI denotes state s because that's what they got with an HTTP GET?

No, it doesnt require that. 


> Sorry if this is muddy thinking, not enough coffee yet.
> Cheers,
> -w

IHMC                                     (850)434 8903 or (650)494 3973   
40 South Alcaniz St.           (850)202 4416   office
Pensacola                            (850)202 4440   fax
FL 32502                              (850)291 0667   mobile
phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes
Received on Thursday, 13 October 2011 13:55:29 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:04:09 UTC