Re: The Map/Territory Conundrum in Topic Maps vs. RDF

["W. E. Perry" <wperry@fiduciary.com>]

"Steven R. Newcomb" wrote:

> [Simon St. Laurent:]
> > Multiple maps (which are themselves territories) for shifting
> > territories seems far more useful to me than the notion of single and
> > final maps for fixed territories.
>
> No one should mistake a topic map (for example) for the truth.  All
> topic maps represent the opinions of their authors, and, inevitably,
> their world-views, as well.  Opinions differ, and world-views
> conflict, even among reasonable people.  Reasonable people, even in
> their everyday thinking, must make a distinction between (a) reality
> and (b) what people (including themselves) say about reality.
>
> Imprecision is endemic in the way people speak about "location",
> especially in computer-land.  Even in technical conversations, it's
> rarely clear whether we're talking about a location in some sort of
> representation (such as something that is called a "map"), or in some
> other kind of universe, such as (for one example), the "real"
> universe.

Up to this point there is nothing that I disagree with, nor would dispute.

>  Therefore, despite Walter Perry's annoyance with the
> "platitude" that "The map is not the territory", we should emphasize
> this platitude.  Part of the reason for this necessity is that people
> usually think fuzzily before they can begin to think clearly.  Another
> part of the reason is that an enormous amount of marketing effort has
> been spent erasing the "map vs. territory" distinction from the minds
> of consumers.  Microsoft's "Where do you want to go today" slogan is
> just one (actually comparatively open and guileless) example of what
> I'm talking about.
>
> The merging capabilities of topic maps were designed in anticipation
> of multiple conflicting maps for the same territories.  The scoping
> facilities of topic maps can be used to prevent us from losing track
> of which topic maps contributed which assertions to topic maps that
> result from merging other topic maps.  In this way, we can know who
> agrees (or agreed) with whom, because we can know whose opinions, as
> of what date, are reflected in each assertion.
>
> -Steve

In this second part of your post, I think that our differences arise
(unremarkably) because you are taking a broad view of the semantics of a
map, while I am speaking from an intentionally narrow perspective on a
particular syntactic form. My point was that a marked up instance is an
unusual sort of territory because the instance document includes both the
'bare' content as well as markup which is a gloss upon that content.
Nevertheless, that markup is as much part of the content of the instance as
is anything else which that instance includes. If the syntactic instance is
considered the territory (an assumption that I am making, but you may not
be), then the markup which is included in that instance is the basis for
one of many possible maps to that territory. This is what it means to me to
say that the marked up instance is 'self describing'. That particular map
is the elaboration of particular semantics from the syntax of the instance,
including both the markup and any other instance content.

That map is certainly not the only possible map, gloss, structure or other
opinion which might be constructed upon the non-markup content of that
instance, and in fact may not be the only one which could be elaborated
from that particular combination of markup and other content. But unlike,
perhaps, the 'real universe' or (pace Pontius Pilate) the 'truth', marked
up documents do consist of two types of content, one of which, the markup,
is in some sense a commentary upon, a suggested structure for, or a
possible map to some portion of the rest. That was my narrow point in
calling the map/territory distinction a conundrum in the specific case of
markup. Granted, if you are to visualize the map first, as an abstraction
with its own semantics, and then regard the markup syntax which is the
concrete expression of that map as merely a serialization syntax, then from
that particular point of view the map has a reality--albeit an abstract
one--apart from its expression in markup content in the instance. Yet even
in that case, when the serialization syntax is considered as the content of
a marked up instance, it is as such the sole concrete basis of both
territory and the particular map asserted over that territory.

Respectfully,

Walter Perry