Re: One major NLF2DWS flaw: density / compactness

> [YA]  "non-NL"?!  Why not just say "linear"?
They're not really exact antonyms.

> [YA]  You've lost me here!
>  - Aren't "information" and "entropy" antonyms?
No!

See http://en.wikipedia.org/wiki/Information_entropy for the mathy version.

>  - Aren't "predictability" and "unpredictability"
>   antonyms?
Yes.

> I correlate "information" with "form",
> "organisation" and "structure".
>
> I correlate "entropy" with "arbitrariness",
> "disorganisation" and "unstructuredness".
This is why I specified that I meant it in the technical sense, as in
the link above.

It is arbitrariness in the sense that it is unpredictable given
context, i.e., that it contains /new information/.

> "information" is essentially surprising; after
> all, no news is no *news*.  "information" is,
> by its nature, unpredictable and surprising.
Exactly so.

> I'm sure you're onto something, whatever
> you label it; have you ever tried reading
> through a massive government publication
> for sense?
Uggh. :-P

> [YA]  I think the answer is - it depends very
> much on the reader.  Why else would Word
> come with readability statistics, like the
> Flesch-Kincaid readability score, or reading
> age?  I know, as someone who writes daily for
> consumption by ordinary people, that I can't
> expect to be understood well enough if I pitch
> my writing at a level much beyond six or seven
> years of schooling.  There's a probability dis-
> tribution involved here, probably a normal bell-
> curve or something quite similar.  So, as always
> in communication, you will need to design a
> workable compromise between efficiency and
> efficacy.  Some people will never get it; for
> some others, you'll be labouring the obvious.
I think this is an entirely separate thing. One could still use
"simple" concepts and short words (or their NL-symbol equivalent) and
have a high entropy content.

In any case, I don't mind being snobbish about the intelligence needed
to use the system; I don't think anyone who's not in the top
percentiles would even be interested anyway, because most others seem
to think that existing systems are perfectly sufficient for their
needs.

But as I said, I consider this unrelated to the entropy question...

> I. Wrapping, linking and decomposing
> One of my major beefs with many printed works,
> including web pages, is that the text is too small
> to read
FWIW, in the majority of cases you can set your client / browser to
display it at a level you find comfortable.

> An unanswered question I have on your ideas for
> a fully 2D writing system is this: is it bounded?
I believe I have addressed this actually.

a) you can have pagination or 'wrapping' as you describe (though I
consider this to be a kludge)

b) you can use computer media to have it be more like 2.5D (thanks to
Gabe for pointing this out) - namely, you have zoom and pan functions
when reading it, in addition to 'follow the link' or other related
ones

The latter I think is the solution for texts of arbitrary size and mutability.

For fixed presentation (e.g. a mural), you could have a more
structured layout, with a result somewhat like a pointilist painting -
using reader distance to resolve the various 'levels'.

I don't think it would adapt well to 8.5x11 page books, unless it's as
a set of single or double page size chunks (like, say, a book of NL
poetry).

>  1. read what is written in your F2DWS, without
>  calling on someone else or using a magnifying
>  glass;
That's purely an issue of display size; I don't see it as a problem
except perhaps in limiting how much you can fit on a page.

>  2. apprehend your main thought readily;
>  3. explore the details of that thought;
>  4. *ignore* the details of that thought.
Done by having levels of zoom, whether fixed or dynamic.

> II.  Scan direction, or "What next?"
[...]
> Ok, enough background!  However you organise
> your F2DWS, you should consider (as one aspect
> of its usability) that people need clear expecta-
> tions of how to scan the writing efficiently.  It
> wouldn't be much help, for instance, if the writing
> could go off an arbitrary distance in an arbitrary
> direction while still saying something central to
> the utterance.
Good point.

I think however that the very idea of 'scan direction' would be
seriously messed with by a NL language, because it's an inherently
linear (and fixed-parse-order) concept.

This is actually one of the things I think would make for interesting
research - namely, how do people scan different implementations of NL
systems? I think it could vary considerably depending on
implementation (eg the parts what you mentioned).

I think one thing one could put as a starter spec is that scan
distance from any given idea to any connected one, and from any idea
to its details, should be minimized.

One counterpoint though: speed reading works in very large part
BECAUSE common text is so low on entropy - one can recongize whole
chunks at once because they are so predictable. High entropic text
would make this impossible, because the amount of data lost in speed
reading (that is regained through context and reasonable assumptions /
guessing) would be high enough to cause real (unrecoverable) data
loss.

> III. Trivia:
> -------------
>
> 1. A F2DWS is necessarily a NLWS.
I wonder. Only thing I've come up with though.

> 2. Unless you plan to give up writing on paper or
> computer screens, a NLWS is necessarily a F2DWS.
I see no reason to restrict it to paper & computer screens - 3d
implementations are surely in the future. :-)

> 3. Ergo, the phrase abbreviated NLF2DWS is a bit
> redundant.
Not quite. The reason I added the F2D part was to distinguish it from
systems such as Glyphica Arcana, Ithkuil's cartouche-based system, etc
- ones which are in some sense nonlinear and in some sense 2d, but not
in the senses that *I* mean. So I want something that clearly names my
type of system alone, so as not to cause the sorts of
misunderstandings (and arguments) that happened on CONLANG when I
first presented the idea. (Ray Brown may be able to comment on this
better, since he is as it were a 'convert'... Ray, what do you think
of this issue of trying to distinguish these different types of
non-linearity / 2d-ness?)

 - Sai

P.S. Two requests for ease of reading:
1. Increase your linebreak cutoff; it's fairly narrow as is
2. Cut out more of your quoted content, to make it so that you're only
quoting what you are directly referring to in the text below; this
makes it clearer and reduces bulk.
Thanks!

Reply