Re: [idn] draft about Tradition and Simplified Chinese Conversion[version01]

[Eric Brunner-Williams in Portland Maine <brunner@nic-naa.net>]

This pattern of exchange has been going on since an earlier "contributor"
made claims about some (undocumented) "supreme system".

Someone points out (correctly) that labels are seqences of octets, and
octets can have some adjacency semantics (serialized-octet-sequence for
CCS other than ASCII, infix rule for "-" for ASCII), and that except for
this adjacency semantic, a property interior to octets forming isolated
characters, and with the sole exception of "-", independent of character
adjacencies.

This observation has as a conclusion that the correct characterization
of the growth properties of equivalent labels is non-linear.

Someone else (inevitably) makes the claim that because existing labels
in some label collections, and strings in non-label-contexts have some
character adjacency properties, e.g., the LDH distribution in ASCII labels
isn't random, and SC-TC mixes aren't random, and that these statistical
properties suggest linguistic semantics additionally apply.

This observation has as a conclusion that the correct characterization
of the growth properties of equivalent labels is linear.

Now, if the someone else were any old "contributor" with a "supreme"
solution, I wouldn't care. The reply to the authors of TSCONV however
is written by the co-chair of this working group.

I'd like to hear from the other co-chair if, in his opinion, James is
writing as a contributor, in which case he can be ignored as just another
argumentative idiot, at least on this issue, or if he is debating the
authors of a draft, substituting his (odd) technical judgement for that
of the working group as a whole.

James' opinion of himself, or his opinion of myself, isn't that hard to
guess, so lets just look at what syntactic guidance we've got, resulting
label semantics, and scale.

> > For 2, TSCONV intend to reduce 2^n , every one do not have to register
> > 2^n
> 
> TC-SC is not a 2^n problem. TC-SC have basically around 2-4 variant in
> practice. Others nonsense variant can be constructed but it bears no
> meanings usually.

Incidently, the earlier "contributor" of some (undocumented) "supreme system"
actually "got it", and dropped that part of his proposal that had equivalent
scaling properties to that we are discussing when we are in fact discussing
the necessity and sufficiency of intermediate transformations -- mappings
and the tables or algorithms required. 

Eric