[sc34wg3] Individual contribution on the U.S. N.B. position on the progress ion of Topic Map standards

Mason, James David (MXM) sc34wg3@isotopicmaps.org
Tue, 30 Mar 2004 14:01:27 -0500


This message is in MIME format. Since your mail reader does not understand
this format, some or all of this message may not be legible.

--------------InterScan_NT_MIME_Boundary
Content-Type: multipart/alternative;
	boundary="----_=_NextPart_001_01C41689.67A04F27"

------_=_NextPart_001_01C41689.67A04F27
Content-Type: text/plain

I have been asked about my stand, as a member of the U.S. National Body, on
the NB's position on the progression of Topic Map standards (SC34 N477,
http://www.y12.doe.gov/sgml/sc34/document/0477.htm
<http://www.y12.doe.gov/sgml/sc34/document/0477.htm> ).  I have had several
quite informative side discussions about issues raised by this document, and
I know that several other people have also been engaged in side discussions.
Because this document will likely be a subject of discussion at the
Amsterdam meeting, which I cannot attend, I think I should explain my stand.
I must make clear that this stand is a function of my membership in the U.S.
N.B. and is independent of anything I do as Chairman of SC34.
 
(My role as Chairman is to see that development of all standards for which
we are authorized by JTC1 proceeds in good order. I have no official
opinions about the standards unless I see someone doing something out of
line with the JTC1 directives or I see bad writing.)
 
Before going further with my position as an individual, I must clarify that
I do not see the current Reference Model document (N460,
http://www.y12.doe.gov/sgml/sc34/document/0477.htm
<http://www.y12.doe.gov/sgml/sc34/document/0477.htm> ) as complete or
satisfactory, and I have occasionally castigated the editors for the opacity
of their documents.
 
Having said that, I will go on to say that my ideal RM is one of those
things that, to paraphrase Voltaire, we'd have to invent if it didn't exist.
 
My ideal RM would be <15 pages long (not counting ISO required stuff and
informative annexes) and would stick to some very simple ideas (which are
indeed represented in N460):

1.	The nature of a topic and its properties.
2.	Which of those properties allow a topic to stand as a proxy for some
subject (in the external world or not), and which are present just for our
convenience in building applications and talking about topics.
3.	How topics can be combined to make assertions about subjects.
4.	How the properties of topics allow the merging of information.

This should be done, in the original spirit of SGML/XML, in an entirely
declarative, nonprocedural way, and the declarative approach is essential.
 
Such an ideal RM is integral to ISO 13250 and should be the first normative
portion of it; all other Topic Maps-related standards must be traceable to
it. 
 
The RM should be the fundamental axioms of Topic Maps, and everything else
(XTM, TMDM, TMQL, TMCL, etc.) should be derived theorems from the axioms.
 
(I believe that I should have stepped out of the Chair's role in Baltimore
and made a real howl over the NWIs as proposed there before they went out
for ballot.)
 
Without the RM, <inflammatory_statement> everything we're doing on TMs is
sound and fury, signifying nothing.</inflammatory_statement>
 
Now, before someone gets truly inflamed, claims I've called him an idiot,
takes his ball ,and goes home and won't play anymore, I believe, as a
practitioner of TMs, that we really need those other projects, and both the
side discussions and my recent application development reinforce my opinion
that we need that work done.
 
I've lately seen some discussions of the RM Use Cases (N490,
http://www.y12.doe.gov/sgml/sc34/document/0490.htm
<http://www.y12.doe.gov/sgml/sc34/document/0490.htm> ) that say, in effect,
"we can do all these in a QL." To which I say, so you may well be able to,
but that's missing the point. Doing the use cases in a QL without the RM is
like doing chemistry in the late 19th century: you can make it work, but you
can't say why it works. Maybe you can even point to the Periodic Table and
give explanations of repeatable results based on it. But without
understanding the quantum mechanics of electron orbits, you can't explain
why the Periodic Table works. 
 
I'm not saying that the RM has to be so difficult as quantum mechanics (even
if the authors have occasionally tried to make it so). I'm just saying that
without the items listed above, you can't really say anything other than "I
have empirical results."
 
Doing TMDM and dependent projects without the RM in sort of analogous to
doing XML documents without a schema: you can write software to process the
stuff. You may even be able to show something's well formed. But you can't
show it's valid.
 
And I think we need the good establishment of fundamental axioms if we're to
hold our own against anything coming out of the W3C. I've heard it said
lately that we need to keep pushing TMQL and TMCL, etc. or we don't have a
chance of withstanding the RDF camp, and I've heard it said that we can't
afford to slow down those efforts. In response, I say to the latter that if
speed is essential, we've already lost. To the former, I say that I don't
see any need to slow down (at least from the perspective of one who's been
working on this stuff for 23 years).
 
"Curing the Web's Identity Crisis", to steal a phrase from the WG3 convenor,
demands a good source of identity. And while we can say, as the WG3 convenor
has done quite eloquently on several occasions, that TMs handle identity
better than RDF, we're still working on that empirical basis, not on the
basis of derivation from well established axioms. 
 
Axioms are, of course, not absolute truths. They are simply atomic
assertions from which to build other assertions. At some point, one has to
accept something on faith. At the moment, I contend, some of us are taking
too much on faith.
 
XTM is an example of something we're taking on faith. Yes, some of us argued
long and hard about its development, and later we voted it into the ISO
standard. But does it really do all we need? Is it, aside from being in a
more easily swallowed notation, really capable of all that HyTM was? Is
either of them really an adequate embodiment of TMs? Without the RM, we
don't have a real way of saying. And without a firm foundation for XTM, how
can we say that the TMDM (which is essentially an annotation of XTM), much
less TMQL, TMCL, CXTM, etc. really capture all of TMs? In another list there
has recently been a thread about the meaning of scope in XTM associations.
One of the early assertions in that thread was something to the effect "I
was on the XTM committee, and we discussed this, and we agreed on X." That's
a heck of a way to explain a standard! Sure, I'm a big advocate of corporate
memory (just get me started on old war stories), and I believe we need to
remember how we came to certain decisions. But without an axiomatic basis
for item 3, above, it's not going to convince everyone. Indeed, the
discussion seemed to resolve itself only when someone came up with an
example most of the participants could agree on.
 
It may be argued that some of the items I mention above are covered in the
TMDM draft. Some of them are indeed discussed, but always in reference to a
specific syntax. We're lacking the declarative, nonsyntactic foundation.
TMDM should stick to explanation of the syntactic instance at hand, and
leave establishment of definitions of what a TM is to the RM. Moving those
things out might work to the benefit of both the RM and DM. As it is now,
the TMDM is attempting both to create TMs ex nihilo and explain how XTM
works. That's too much for one document. The RM should take on the job of
justifying TMs, which would be simpler for an axiomatic document, and the
TMDM could be shortened and simplified as well. 
 
To summarize: The assertion "We can do X without the RM" is, to me, no real
argument. It's ignoring the question of what TMs really are. It's skating on
thin ice, and no matter how skillful the performance, it's still something
risky that may crash and hurt us. Accordingly, I support the U.S. position
that no work be carried to completion without establishing its traceability
to the RM foundation.
 
 
Jim Mason
 
 
 
 
 
 
 

------_=_NextPart_001_01C41689.67A04F27
Content-Type: text/html

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=us-ascii">
<TITLE>Message</TITLE>

<META content="MSHTML 6.00.2800.1400" name=GENERATOR></HEAD>
<BODY>
<DIV><FONT face=Arial size=2><SPAN class=269191714-30032004>I have been asked 
about my stand, as a member of the U.S. National Body, on the NB's&nbsp;position 
on the progression of Topic Map standards (SC34 N477, <A 
href="http://www.y12.doe.gov/sgml/sc34/document/0477.htm">http://www.y12.doe.gov/sgml/sc34/document/0477.htm</A>).&nbsp; 
I have had several quite informative side discussions about issues raised by 
this document, and I know that several other people have also been engaged in 
side discussions. Because this document will likely be a subject of discussion 
at the Amsterdam meeting, which I cannot attend, I think I should explain my 
stand. I must make clear that this stand is a function of my membership in the 
U.S. N.B. and is independent of anything I do as Chairman of 
SC34.</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN 
class=269191714-30032004></SPAN></FONT>&nbsp;</DIV>
<DIV><FONT face=Arial size=2><SPAN class=269191714-30032004>(My role as Chairman 
is to see that development of all standards for which we are authorized by JTC1 
proceeds in good order. I have no official&nbsp;opinions about the standards 
unless I see someone doing something out of line with the JTC1 directives or I 
see bad writing.)</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN 
class=269191714-30032004></SPAN></FONT>&nbsp;</DIV>
<DIV><FONT face=Arial size=2><SPAN class=269191714-30032004>Before going further 
with my position as an individual, I must clarify that I do not see the current 
Reference Model&nbsp;document (N460, <A 
href="http://www.y12.doe.gov/sgml/sc34/document/0477.htm">http://www.y12.doe.gov/sgml/sc34/document/0477.htm</A>) 
as complete or satisfactory, and I have occasionally castigated the editors for 
the opacity of their documents.</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN 
class=269191714-30032004></SPAN></FONT>&nbsp;</DIV>
<DIV><FONT face=Arial size=2><SPAN class=269191714-30032004>Having said that, I 
will go on to say that my ideal RM is one of those things that, to paraphrase 
Voltaire, we'd have to invent if it didn't exist.</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN 
class=269191714-30032004></SPAN></FONT>&nbsp;</DIV>
<DIV><FONT face=Arial size=2><SPAN class=269191714-30032004>My ideal RM would be 
&lt;15 pages long (not counting ISO required stuff and informative annexes) and 
would stick to some very simple ideas (which are indeed represented in 
N460):</SPAN></FONT></DIV>
<OL>
  <LI><FONT face=Arial size=2><SPAN class=269191714-30032004>The nature of a 
  topic and its properties.</SPAN></FONT></LI>
  <LI><FONT face=Arial size=2><SPAN class=269191714-30032004>Which of those 
  properties allow a topic to stand as a proxy for some subject (in the external 
  world or not), and which are present just for our convenience in building 
  applications and talking about topics.</SPAN></FONT></LI>
  <LI><FONT face=Arial size=2><SPAN class=269191714-30032004>How topics can be 
  combined to make assertions about subjects.</SPAN></FONT></LI>
  <LI><FONT face=Arial size=2><SPAN class=269191714-30032004>How the properties 
  of topics allow the merging of information.</SPAN></FONT></LI></OL>
<DIV><FONT face=Arial size=2><SPAN class=269191714-30032004>This should be done, 
in the original spirit of SGML/XML, in an entirely declarative, nonprocedural 
way, and the <EM>declarative approach is essential</EM>.</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN 
class=269191714-30032004></SPAN></FONT>&nbsp;</DIV>
<DIV><FONT face=Arial size=2><SPAN class=269191714-30032004>Such an&nbsp;ideal 
RM is integral to ISO 13250 and should be the first normative portion of it; all 
other Topic Maps-related standards must be traceable to it. </SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN 
class=269191714-30032004></SPAN></FONT>&nbsp;</DIV>
<DIV><FONT face=Arial size=2><SPAN class=269191714-30032004>The RM should be the 
fundamental axioms of Topic Maps, and everything else (XTM, TMDM, TMQL, TMCL, 
etc.) should be derived theorems from the axioms.</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN 
class=269191714-30032004></SPAN></FONT>&nbsp;</DIV>
<DIV><FONT face=Arial size=2><SPAN class=269191714-30032004>(I believe that I 
should have stepped out of the Chair's role in Baltimore and made a real howl 
over the NWIs as proposed there before they went out for 
ballot.)</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN 
class=269191714-30032004></SPAN></FONT>&nbsp;</DIV>
<DIV><FONT face=Arial size=2><SPAN class=269191714-30032004>Without the RM, 
<SPAN class=269191714-30032004>&lt;inflammatory_statement&gt; </SPAN>everything 
we're doing on TMs is sound and fury, signifying 
nothing.&lt;/inflammatory_statement&gt;</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN 
class=269191714-30032004></SPAN></FONT>&nbsp;</DIV>
<DIV><FONT face=Arial size=2><SPAN class=269191714-30032004>Now, before someone 
gets truly inflamed, claims I've called him an idiot, takes his ball ,and goes 
home and won't play anymore, I believe, as a practitioner of TMs, that we really 
need those other projects, and both the side discussions and my recent 
application development reinforce my opinion that we need that work 
done.</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN 
class=269191714-30032004></SPAN></FONT>&nbsp;</DIV>
<DIV><FONT face=Arial size=2><SPAN class=269191714-30032004>I've lately seen 
some discussions of the RM Use Cases (N490, <A 
href="http://www.y12.doe.gov/sgml/sc34/document/0490.htm">http://www.y12.doe.gov/sgml/sc34/document/0490.htm</A>) 
that say, in effect, "we can do all these in a QL." To which I say, so you may 
well be able to, but that's missing the point. Doing the use cases in a QL 
without the RM is like doing chemistry in the late 19th century: you can make it 
work, but you can't say why it works. Maybe you can even point to the Periodic 
Table and give explanations of repeatable results based on it. But without 
understanding the quantum mechanics of electron orbits, you can't explain why 
the Periodic Table works. 
<DIV><FONT face=Arial size=2><SPAN 
class=269191714-30032004></SPAN></FONT>&nbsp;</DIV></SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN class=269191714-30032004>I'm not saying that 
the RM has to be so difficult as quantum mechanics (even if the authors have 
occasionally tried to make it so). I'm just saying that without the items listed 
above, you can't really say anything other than "I have empirical 
results."</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN class=269191714-30032004>
<DIV><FONT face=Arial size=2><SPAN 
class=269191714-30032004></SPAN></FONT>&nbsp;</DIV>
<DIV><FONT face=Arial size=2><SPAN class=269191714-30032004>Doing TMDM and 
dependent projects without the RM in sort of analogous to doing XML documents 
without a schema: you can write software to process the stuff. You may even be 
able to show something's well formed. But you can't show it's 
valid.</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN 
class=269191714-30032004></SPAN></FONT>&nbsp;</DIV></SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN class=269191714-30032004>And I think we need 
the good establishment of fundamental axioms if we're to hold our own against 
anything coming out of the W3C. I've heard it said lately that we need to keep 
pushing TMQL and TMCL, etc. or we don't have a chance of withstanding the RDF 
camp, and I've heard it said that we can't afford to slow down those efforts. In 
response, I say to the latter that if speed is essential, we've already lost. To 
the former, I say that I don't see any need to slow down (at least from the 
perspective of one who's been working on this stuff for 23 
years).</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN 
class=269191714-30032004></SPAN></FONT>&nbsp;</DIV>
<DIV><FONT face=Arial size=2><SPAN class=269191714-30032004>"Curing the Web's 
Identity Crisis", to steal a phrase from the WG3 convenor, demands a good source 
of identity. And while we can say, as the WG3 convenor has done quite eloquently 
on several occasions, that TMs handle identity better than RDF, we're still 
working on that empirical basis, not on the basis of derivation from well 
established axioms. </SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN 
class=269191714-30032004></SPAN></FONT>&nbsp;</DIV>
<DIV><FONT face=Arial size=2><SPAN class=269191714-30032004>Axioms are, of 
course, not absolute truths. They are simply atomic assertions from which to 
build other assertions. At some point, one has to accept something on faith. At 
the moment, I contend, some of us are taking too much on 
faith.</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN 
class=269191714-30032004></SPAN></FONT>&nbsp;</DIV>
<DIV><FONT face=Arial size=2><SPAN class=269191714-30032004>XTM is an example of 
something we're taking on faith. Yes, some of us argued long and hard about its 
development, and later we voted it into the ISO standard. But does it really do 
all we need? Is it, aside from being in a more easily swallowed notation, really 
capable of all that HyTM was? Is either of them really an adequate embodiment of 
TMs? Without the RM, we don't have a real way of saying. And without a firm 
foundation for XTM, how can we say that the TMDM (which is essentially an 
annotation of XTM), much less TMQL, TMCL, CXTM, etc. really capture all of TMs? 
In another list there has recently been a thread about the meaning of scope in 
XTM associations. One of the early assertions in that thread was something to 
the effect "I was on the XTM committee, and we discussed this, and we agreed on 
<EM>X</EM>." That's a heck of a way to explain a standard! Sure, I'm a big 
advocate of corporate memory (just get me started on old war stories), and I 
believe we need to remember how we came to certain decisions. But without an 
axiomatic basis for item 3, above, it's not going to convince everyone. Indeed, 
the discussion seemed to resolve itself only when someone came up with an 
example most of the participants could agree on.</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN 
class=269191714-30032004></SPAN></FONT>&nbsp;</DIV>
<DIV><FONT face=Arial size=2><SPAN class=269191714-30032004>It may be argued 
that some of the items I mention above are covered in the TMDM draft. Some of 
them are indeed discussed, but always in reference to a specific syntax. We're 
lacking the declarative, nonsyntactic foundation. TMDM should stick to 
explanation of the syntactic instance at hand, and leave establishment of 
definitions of what a TM is to the RM. Moving those things out might work to the 
benefit of both the RM and DM. As it is now, the TMDM is attempting both to 
create TMs <EM>ex nihilo</EM> and explain how XTM works. That's too much for one 
document. The RM should take on the job of justifying TMs, which would be 
simpler for an axiomatic document, and the TMDM could be shortened and 
simplified as well. </SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN 
class=269191714-30032004></SPAN></FONT>&nbsp;</DIV>
<DIV><FONT face=Arial size=2><SPAN class=269191714-30032004>To summarize: The 
assertion "We can do <EM>X</EM> without the RM" is, to me, no real argument. 
It's ignoring the question of what TMs really are. It's skating on thin ice, and 
no matter how skillful the performance, it's still something risky that may 
crash and hurt us. Accordingly, I support the U.S. position that no work be 
carried to completion without establishing its traceability to the RM 
foundation.</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN 
class=269191714-30032004></SPAN></FONT>&nbsp;</DIV>
<DIV><FONT face=Arial size=2><SPAN 
class=269191714-30032004></SPAN></FONT>&nbsp;</DIV>
<DIV><FONT face=Arial size=2><SPAN class=269191714-30032004>Jim 
Mason</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN 
class=269191714-30032004></SPAN></FONT>&nbsp;</DIV>
<DIV><FONT face=Arial size=2><SPAN 
class=269191714-30032004></SPAN></FONT>&nbsp;</DIV>
<DIV><FONT face=Arial size=2><SPAN 
class=269191714-30032004></SPAN></FONT>&nbsp;</DIV>
<DIV><FONT face=Arial size=2><SPAN 
class=269191714-30032004></SPAN></FONT>&nbsp;</DIV>
<DIV><FONT face=Arial size=2><SPAN 
class=269191714-30032004></SPAN></FONT>&nbsp;</DIV>
<DIV><FONT face=Arial size=2><SPAN 
class=269191714-30032004></SPAN></FONT>&nbsp;</DIV>
<DIV><FONT face=Arial size=2><SPAN 
class=269191714-30032004></SPAN></FONT>&nbsp;</DIV></BODY></HTML>

------_=_NextPart_001_01C41689.67A04F27--

--------------InterScan_NT_MIME_Boundary--