DNSO Mailling lists archives


<<< Chronological Index >>>    <<< Thread Index >>>

RE: [ga] Question No. 1

Just fyi, I continue to think that Roeland's idea is an interesting 
one, and I've posted a brief comment at ICANNWatch about this issue, if 
anyone cares to comment.


I wrote:

>I think Roeland is right on target.  The DNS is a kind of language -- 
>it gives names to things.  Coordination is very valuable in a language 
>-- if Roeland calls a round spherical object a 'ball,' and I call it a 
>'goidosphere,' we're going to have trouble understanding one 
>another.  But we need imposed coordination of the ICANN variety about 
>as much as we need a Bureau of Definitions to tell us what words 
>mean.  We had this fight 200 years ago; really smart and sensible 
>people actually argued that we needed things like the Academie 
>Francaise to 'authorize' particular labels for things, lest chaos 
>ensue.  That argument lost, 
>decisively.  See  http://www.temple.edu/lawschool/dpost/DrakeTalk.pdf 
 >  Somehow we manage to agree to call it a 'ball', or not, quite 
>effectively without any single institution telling us we have to do that.

At 12:00 PM 4/11/01 -0700, Roeland Meyer wrote:

>There is a serious disconnect here. When dealing with finite physical
>resources, these distinctions make sense. OTOH, when dealing with abstract
>resources these distinctions make absolutely no sense. I am sorry to see
>that some have managed to rat-hole the discussion into "argument by
>analogy". There are no definitive analogies that matter. We need to grasp
>the concepts directly and deal with them ... directly.
>The facts are that names are simply labels. I may label a spherical rubber
>object a "ball" and you may label a square steel cube a "ball". Both are
>equally valid within their context. A third-party would have to deal with
>each of us, within our own context, and use the labels appropriately, when
>discussing those things within the given context.
>The fundimental problem here is that labels are not resources at all. They
>are a meta-object, a means for organizing a universe of entities such that
>we can distinguish one from another. They are a reusable tag. We can change
>the object that we place the label on. We can also place the label on
>another object ... at the same time. The problem here is that the contexts
>that the labels are valid within ... can also be labeled, which is valid
>within a meta-context. This is a recursive phenomenon. I submit that, the
>only other place we have this problem is in object-oriented design and
>analysis (OOD and OOA). Analogies don't work, for defining the problem,
>there either. Further, OOD and OOA technologies weren't available when the
>DNS system was designed.
>This is the fundimental theoretical problem that the original designers of
>the DNS system ran into. Thair answer was that, they drew the line in the
>sand, at the root-zone level, and said "we'll stop here, for now". The line,
>while occuring at a natural demarcation point, is still arbitrary. This
>reduced the problem to managable proportions, such that they could build the
>initial DNS system (the decendent of which is ISC BIND 8.2.3).
>What we have here, as a direct result, is a system which is, intentionally,
>context-limited. However, for labeling hosts on the Terabit project, it
>worked. The problems we are dealing with today are a direct resultant of
>applying the DNS system, out of its contextual referent, to problems that
>really have their own context rules. Refering to my previous example, my
>definition of "ball" differs from yours and unless we make an effort to
>define the context of "ball" we will never understand each other. In the DNS
>context, the label "ball" is a name space collision unless the relevant
>context is defined. The key is that, understanding the context matters and
>we'll never resolve this until we start dealing with context and
>acknowleging that there are other contexts.
>The practical application is that;
>ICANN cannot ignore the contextual referents of the other root-zone
>Trademark and IP interests cannot apply thier restrictions without also
>understanding the contextual references.
>The software needs to apply to a larger contextual scope than it has ever
>had to heretofore (the MultiBIND project has been announced).
>This message was passed to you via the ga@dnso.org list.
>Send mail to majordomo@dnso.org to unsubscribe
>("unsubscribe ga" in the body of the message).
>Archives at http://www.dnso.org/archives.html

This message was passed to you via the ga-full@dnso.org list.
Send mail to majordomo@dnso.org to unsubscribe
("unsubscribe ga-full" in the body of the message).
Archives at http://www.dnso.org/archives.html

<<< Chronological Index >>>    <<< Thread Index >>>