OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help



   Re: [xml-dev] Abstract Equivalence (was Microsoft FUD on binary XML...)

[ Lists Home | Date Index | Thread Index ]

Bob Wyman wrote:

> But whatever my intent, and whatever you believe you see, I will always
> know that it is merely the shadow of my fingers.

That's easy enough to grant, but the subject here is not you, the creator,
nor your intent, nor what you know in your inner heart, but rather the
text which you *publish*, which has a particular physical manifestation,
which may be empirically examined, and which of itself can manifest some
properties but not others. Examination of the text reveals that certain
qualities inhere in its physical manifestation. Among those qualities are
document order (which is the basis of narrative technique and
compositional style); metre, rhythm, scansion (which are the structure,
not just of poetry, but of poiesis generally); and various identifiable
rhetorical figures ('rhetoric' in that sense consisting of what can be
shown in the physical manifestation of the text). What does not inhere in
a text is e.g. the 'numberness' of runs of characters which are to be
understood as integers or floats, nor the 'boleanness' of other characters
which are to be understood a booleans, nor indeed any sort of data typing.
You might 'know' that it is merely the shadow of your fingers, but if what
you render, and publish, is a silhouette which can be cut as a stencil and
used to mass produce recognizable rabbits, then knowing that the shape is
merely a finger shadow is as immaterial as knowing that character 'X' in
encoding 'Y' is merely the manifestation of bytes 'xxxx'. And that's just
the point:  XML is just text. It is not a level lower than text,
consisting of the bytes, or in another context the glyphs, which are
rendered from particular encoding as the characters of that text. Nor is
XML an abstract precursor to text, nor any abstraction of the syntax in
which text is manifested. XML is text; a body of XML syntax is
fundamentally a text (whatever else it might be agreed, or even
misunderstood, to be).

> But, I'm also Aristotelian in that I think that "The Rhetoric" is
> relevant reading whenever one is dealing with communications
> systems. As Aristotle teaches, it is important for the speaker to
> understand the warrants of his audience. Both speaker and audience must
> share a common understanding of the speech and the context in which it
> is made -- otherwise, what is heard may not be what was said.

Nope. Plato was the first overt enemy of poetry, and Aristotle
appropriated Plato's misunderstandings of poetry wholesale. You refer to
Aristotle's opinions of how to produce an intended effect upon an
audience, which is a vastly different matter (and two degrees of
abstraction removed) from producing a text. Poiesis is the work of
producing the text. How further processes might abstract from that text
its correspondence to a schema, a data vocabulary, or a typing scheme is
way beyond the scope of the text itself. And how still further processes
might engender particular effects on a particular audience out of the
abstract conformance of a text to schematic or vocabulary expectations is
entirely unknowable to the straightforward process of first creating and
publishing the text. What is heard is what is in the text, an instance of
syntax which of itself manifests narrative and compositional technique,
metre, rhythm, scansion, assonance and other rhetorical figures. If what
is in that text is not what was 'said', then that is a mechanical error in
the production of the instance text. But it is a vastly different matter
to say that what is in the text is not what was intended by its creator,
or what was expected by its audience. Those intentions and expectations,
unlike the lexical and rhetorical substance, are nowhere found in the
text. They might be elaborated from it by the operation of suitable
processes (provided, of course, that the particular instance text conforms
to the expectations or suits the intentions of such processes). Otherwise,
what is heard (that is, interpreted in accordance with an audience's
expectations) may well not be what was said (that is, intended by the
author). But at an empirically examinable level, barring error, what is
said is by definition what is encoded in and published as a text, and what
is heard (however it might differ from what is understood from that
hearing) is that same text.

> If I wish to show flexible fingers, yet you see rabbits, we have not
> communicated...

My expectations, or other elaborations from the text which you publish
might well not match your intentions upon publishing it--it happens all
the time, but has nothing to do with the text which we share, only with
the difference in the processes which you and I apply to that text. As to
whether we have communicated:  well, if you are satisfied that what you
published was, from your understanding, what you wanted it to be, and for
my part I am satisfied with the use I have made of the text from my
understanding of it, then how much more than that might we be said to have
communicated usefully. Within the internetwork topology, certainly, I do
not think that we can improve upon that (already satisfactory) result.

> Schemas help us communicate and share warrants -- they are the ground,
> implicit or explicit, upon which all communications
> are built.

They attempt to constrain us to identical understanding--that is
processing--of a particular text. To each gain value from our separate
processing of a text does not require that we process it identically, nor
upon the same abstract understanding of it. And identical understandings
are certainly *not* the ground upon which all communications are built. We
have had a period of less than forty years in computing during which an
enormous number of transactional processes, executing staggering volumes
of transactional outcomes, have been based on identical understandings of
the shared data structures upon which those processes operate, but that is
only because that was the only widely-understood way to implement those
transactional processes. In the wider context of transactions, by far the
greater number (and I suspect all the interesting ones) are arbitrages
between different understandings of the the same text, data or other
substantial content, rather than counterparty operations which share an
identical understanding of the stuff against which they are executed.

> Of course, by communications, I mean "utterances" that are made with
> "intent."

Of course that's what you mean, but that is not the creation of a text:
it is the attempt to cajole the use of the text to a particular end.

> I do not believe it useful to consider all processes which cause
> modifications in the entropy of remote systems
> to be "communications".

Then you deny the usefulness of poiesis, and of the creation of texts,
compositions, and published works generally. By contrast, I consider the
production and publication of texts to be the basic stuff of human
creativity and expression.

> > the 50 years or longer struggle in the 20th century
> > that was required for classical philology to
> > understand the nature of oral poetry demonstrates
> > why the physical, rather than any abstract nature of
> > a text is worth insisting upon.
> You declare here the end of a battle that still rages. Perhaps *you*
> have accepted this view, but many others -- including myself, have not.

In the specific case I cite, the understanding of oral (traditional,
formulaic) poetry, the battle is over. The 'learned Germans' of the
nineteenth century have been vanquished. There is no longer any
respectable intellectual opinion which maintains any thing other than that
the stuff or components of formulaic composition are syntactic fragments
fitted to the occurrence of specific grammatical morphology within
specific metrical constraints.

> But, the subject here is interprocess communication -- not poetry.

A major point of mine here is that they are separate examples of the same

> Thus, I will resist the temptation to flame about what passes as
> teaching in the literature departments of today's universities...
> > ASN.1 and abstract syntax generally are incapable
> > of a precise and unambiguous encoding of inherent
> > fundamental textual properties without resorting
> > to a priori agreements between the creator and the
> > consumer of a document, and from the very nature of
> > document processing such agreements are unreliable
> > and negligible.
>         This is simply not true. Technical inaccuracies do little to
> advance your cause...

As above, the stuff out of which we are composing are lexical
formulae--syntactic fragments which express necessary grammatical
morphology within particular metrical constraints. These have no
discernible abstract precursor:  they are themselves the fundamental
components of composition. Abstractions might be drawn from them, but only
in the pathological case where an 'abstract' syntax were itself the set of
all occurring formulae (that is, the collection of all detected concrete
syntax) could it undoubtedly schematize all of the forms which we might
encounter in a text.

> There is no question that an ASN.1 schema can be defined that encodes,
> without loss, all of the emanent qualities of any particular piece of
> text.

Again, as above, only if that 'schema' were the enumeration of all
possible syntactic formula which might occur.


> XML, ASN.1, BER, etc. are all used, in one application or another, to
> deal with both documents and data. You seem to argue for rules that are
> mostly applicable to these things you call documents -- to the exclusion
> of solutions that make data interchange as easy andaccurate as it might
> be.

No, I argue that textual presentations (encodings if you must) by their
very textual nature end in the production of a text. Parenthetically, that
text does not have to be written--it might be recited--or might be, for
example, music. The salient requirement is that a text must be fixed by a
rendition on a particular occasion, and as a fixed entity might then be
examined or further processed. Upon examination such texts display the
characteristics I describe as inherent in text:  the compositional
techniques displayed in document order; metre, rhythm, scansion; basic
rhetorical figures, etc. They do not, as texts, display the fundamental
'type' qualities of data, which you and I seem to agree are abstract,
precisely because conveying those qualities in, or elaborating them from,
a syntactic instance requires processing that instance in a particular

> It seems that to maintain this position, you would have to insist that
> XML is an inappropriate tool for "data applications." Yet, you appear to
> be taking the alternate rule of insisting that all XML based data
> systems be forced to accept the "fundamentally" different constraints of
> a document-oriented system. How can you justify this?

On the contrary, I believe that in the internetwork topology XML is the
*most appropriate* tool for "data applications", though using it requires
understanding data in textual terms, a very significant leap which ASN.1
and other abstractions of syntax fall short of. Yes, I do notice that to
use XML requires accepting  textual and  document-oriented premises.
Essential among those premises is that text does not convey the abstract
nature of data, though in ASN.1 the expected data is schematized around
just that abstract nature which text does not convey. Only the application
of a particular process--based on a particular understanding of the data
structure upon which that process is operating--will elaborate some
recognition of the abstract nature of particular data from a textual
rendition of that data. Accepting this inescapable premise allows us to
build processes which not only do not expect textual data to convey
particular abstractions, but allow us to arbitrage profitably among
different understandings of particular texts.

> Why do you care what the "data people" do if your concern is so clearly
> focused on non-data things?

I am simply pointing out that in using XML they are using a demonstrably
'non-data thing', given that the very abstract qualities of data which
ASN.1 is at pains to schematize are just those qualities which are not in
the inherent nature of text to convey.


Walter Perry


News | XML in Industry | Calendar | XML Registry
Marketplace | Resources | MyXML.org | Sponsors | Privacy Statement

Copyright 2001 XML.org. This site is hosted by OASIS