- From: Pat Hayes <phayes@ihmc.us>
- Date: Thu, 4 May 2006 16:04:07 -0500
- To: "Mark Birbeck" <mark.birbeck@x-port.net>
- Cc: <public-swbp-wg@w3.org>
>Hi David, > >> Just to avoid confusion, DanC and I are addressing a little >> different issue. DanC and Pat and I are discussing: >> >> What is an "information resource"? >> >> Whereas you are talking about the issue: >> >> Is is okay to use the same URI both as the name of >> a resource and as the name of a Web page that describes >> that resource? Example: Is it okay for Mark Baker to >> use http://markbaker.ca/ as a name for himself *and* >> for his blog Web page? > >Exactly! And I am saying that they are inseparable. That's why I ended my >post with: > > I feel this is important because in my opinion, none of the > recent discussion has got to the root of the issue, and without > this, talk of trying to 'clarify' what 'information resource' > means leads to a lot of confusion. Issues like whether the > *content* of the resource change over time, etc., are as far as > I can see, irrelevant to the issue. > >In other words, I'm saying that the very fact that discussions abound as to >whether the URI <http://markbaker.ca> can represent both a person *and* a >web-page, show that people are misunderstanding the notion of 'information >resource'. Not necessarily. It might mean that some people are tolerant of ambiguity. Or, it might mean that some people think that the distinction between a person and an information resource is clear, but that there are two different notions of 'represent' being confused in statements like this. So the same URI might - I had better not say 'represent', so - *identify* the information resource but *denote* the person, at the same time. And if indeed identifying and denoting are distinct then this is not an ambiguity, nor necessarily a problem that needs to be solved. It hasn't seem to have been much of a problem so far, at any rate. As you say in your blog http://internet-apps.blogspot.com/ (thanks for the pointer), "at the moment, millions of web pages break this fundamental principle...". Stop right there. If millions of web pages are working fine, while violating a 'principle', then the thing to do is to re-examine the principle. Maybe its not so fundamental after all. So, read on. You predict that trouble is coming: "These processors can get away with [breaking the information resource/resource distinction] though, because they are running distinct 'silos' of data... But if you try to mix the two databases you'll get trouble, because you'll find that the URI of my blog is being overloaded to represent me, a person, the house for sale, and of course, a web-page." OK, suppose that you do. Now, what is wrong with overloading? Programming languages do it all the time. Natural languages do it all the time. Formal logical languages (like Common Logic and its recent offspring IKL) do it: a CL name can represent an individual, a function of any arity and a relation of any arity, all at the same time (in IKL add a proposition to the list.) In every case, it works because there is an adequate way to disambiguate every use of the overloaded name, if it needs to be disambiguated. In English, the surrounding context tells you whether 'rose' is something moving upwards in the past, or a flower. In CL, the syntax tells you if the name is being used in a relational way. And so on. You only get trouble - and not in fact always then - if you overload in a way that cannot be disambiguated, but all the examples I've seen can be. If I tell my banker to put in a bid for the real property referred to by <insert URI>, then what agent, human or software, is going to be confused about my meaning or intentions? Will I accidentally buy you, or your website? If I tell my email handler to email a message to <insert same URI>, what will go wrong? Will an email message get sent to your webpage or your house, instead of to you? If I do an http GET with the URI, am I in any danger of being sent you or your house or even your mailbox? (If someone were to use the same URi for two websites, or for two different people, there would be a real cause for worry: but those cases aren't what we are discussing here.) Your conclusion welcomes disambiguation moving us "away from the world of having to rely on lots of pockets of implicit knowledge", but it doesn't. This is the most fundamental point: when we start relying on descriptions, rather than transfer protocols, to fix referents, as we *must* when we want to refer to non-information resources (i.e. virtually everything in the universe), there is absolutely no way, in principle, to avoid relying on implicit knowledge, be it in 'pockets' or not. This is often called the 'grounding problem' in KR discussions: the fact that names in descriptions are *inherently* ambiguous about what they refer to. And the more you let yourself say about them - the more distinctions you let yourself make by using a richer descriptive vocabulary - the more ambiguous they get. We all know that 'bank' is ambiguous. How many meanings does it have? Three or four? One of them is a thing like Barclays, right? But if you let yourself talk about places and times and types of things and social topics, and so on, you can start distinguishing between the bank as institution, as legal entity, as a building, as the location of the building, as the interior of the building, as the contents of the building, as a role played by the agents who are employed by the institution, etc., etc.. Do we have to have distinct URIs for all of them? Its no good protesting that you don't care about such pedantic distinctions: URIs have global scope, so as long as *somebody* wants to make the distinction, it needs to get made by *everyone*. And as long as the TAG insists that URIs refer unambiguously, then no URI is safe, because there isn't any way to know when you finally get something pinned down uniquely. Arguably things like social security numbers serve to identify people uniquely, but that's not good enough. I can point you at practicing ontologists who will argue about what a person really is, and have incompatible formalizations of personhood. Do they have to use different URIs to refer to me with, because one of them thinks I'm a continuant and the other thinks I have temporal parts? Or can we tolerate a certain ambiguity here? To set out with torch and sword in hand to eliminate all ambiguity on the Web is about as constructive as setting out to travel to infinity and beyond. It just ain't going to work for non-information resources, which can't be wholly characterized by a single finite message. There is nothing you or I or anyone else can do about this, so let's stop trying. >Or to put it another way...it doesn't matter which of the two >issues you answer, yours or mine, answering one will answer the other. > >In passing, I have to say that anyone who is trying to follow all of this >and is confused by the issues...I don't blame you! I don't think it has been >explained very well, and as is often the case with subtle things like this, >people seem to keep repeating the *conclusions*, without reminding the rest >of us how they were arrived at. I would give a lot to know how the TAG arrived at its conclusions. I have tried to discover the historical roots of the WebArch terminology such as 'resource', 'identify', 'represent' and 'name', and have not been able to track it down. Engelbart, for example, is often cited as the source for the term 'resource', but it seems clear that he had a much narrower conception in mind, roughly what WebArch refers to as an information resource. The idea that a resource is 'anything with an identity', whatever that means, seems to have arisen spontaneously somewhere within the W3C around the turn of the millennium, and has no detectable connection with any earlier writings or traditions. Similarly, the WebArch notion of 'represent' has no apparent connection with any earlier useage of that word in philosophy, logic, Krep or computer science. Pat > >Regards, > >Mark > > >Mark Birbeck >CEO >x-port.net Ltd. > >e: Mark.Birbeck@x-port.net >t: +44 (0) 20 7689 9232 >b: http://internet-apps.blogspot.com/ >w: http://www.formsPlayer.com/ > >Download our XForms processor from >http://www.formsPlayer.com/ -- --------------------------------------------------------------------- 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 phayesAT-SIGNihmc.us http://www.ihmc.us/users/phayes
Received on Thursday, 4 May 2006 21:06:28 UTC