Lists Home |
Date Index |
1. Enumerations are an awkward means to address.
Most purely name systems are awkward that way.
They require a bounding specification,
some set of related information to establish
identity within acceptable probabilities. For
example, if you search on my legal name,
Claude Bullard, in Yahoo, you will get two
returns in Alabama. Which one is Len? It helps to
know where Len lives. This must be specific to a
name in a different system. The naming
system is insufficient to establish identity unless
it is paired to the precise location. The limits
of precision in location can create a random choice
that is made arbitrarily (by chance or policy: see
2000 American presidential election for a study
in system limits given an unresolvable granularity).
We can say a URN is a URI is a URL only because
we use tools to make this true. Otherwise, a
name is not of necessity a location and a name
and a location may not be sufficient to establish identity.
A URI is insufficient to establish an information item is
on the web. It makes it possible to put it there
because it is a name paired to a name (location). The
resources at the location establish identity.
Identity is not in and of itself, a value: it is a
derived or testable assertion, a mapping.
2. Enumerations on an infinite space can exhaust
system resources unless the range of choices
can be restricted.
The point is that given equal choices,
establishing identity is a coin toss. Systems
are designed to make choice non-random. Communication
systems are particularly designed to eliminate
randomness or noise. It is useful to have clear
To be effective, the designer bounds the space
to a range of some kind and will choose a means
to identify a member. Otherwise, the system becomes
entropic (ie, Boltzman entropy). The effectiveness
of the means limits the reliability.
It is possible, as in the example of the circle,
for the boundary to contain infinite numbers of
points. Abstractly, this is so, but practically,
a unit will be declared, a means to identify a unit
will be created, and so on. That unit might be
identified by position relative to the system
boundary or by position relative to another member of a
set, or by a unique name, known to be unique only
because it is in the system set.
Namespaces in XML are not boundless. One doesn't
need order. One needs structure and paired names
(the QName and local name) to create boundaries.
The circle provides a structure that bounds
a potential infinity, but not a practical one.
So is the Web bounded or boundless? Is the
Web a bounded information space?
From: email@example.com [mailto:firstname.lastname@example.org]
Sent: Tuesday, April 16, 2002 8:59 PM
Subject: RE: [xml-dev] Boundless Space and Identity
> Len Bullard wrote,
> > Enumerated space. Is that addressing by name or position?
> Could be either. You could address them as 1st, 2nd, 3rd ... from
> some arbitrarily nominated origin, in which case it's addressing by
> position. Or you could assign the points distinct names, eg. "red",
> "blue", "green" ...in which case enumeration is just the name
> resolution mechanism, ie. to find the point named "blue", start
> anywhere, and keep on going till you reach the point with that label.
but in the case of naming "red", "blue", "green", we have to think about
matching name to order. but if there is no regulations,
it's impossible in boundless space to impose name to each point..
Am i right?
> cp. ordinal vs. nominal numbers.
> > Are the points on the curve or is the curve a boundary for the
> > points inside the circumscribed space?
> No, the points _are_ the curve ... there's nothing in between or
> outside, in the same way that 0.5 isn't in between 0 and 1 on the
> natural number line.
i think that many cases in XML, namespace is a kind of boundless space
not ordinally arranged, but nominally identified with key name.
There is no order in components in that space.
and we just can think about virtual curve to circumscribe every components.
so each points in that space are inside of the curve.