Re: [lojban] centripetality: subset vs component

[PILCH Hartmut <phm@A2E.DE>]

> The bottom line is that we are disagreeing as to which is the "important" 
> information.  I am claiming that the important information is whatever the 
> listener does not know, which is context dependent, with the context 
> hopefully known equally well to the speaker as to the listener.

Things are becoming even more comlicated.

The address structures we are talking about may contain elements that
suffice to identify the whole thing and let you skip over some containers.
This is especially true of place addresses like "FairFax", and of names.
It is not true of dates.  So let's, for simplicity's sake, stick to the
dates first. 
 
> The bottom line is that neither I nor anyone I know seems to feel a need to 
> switch things around that you feel.  Therefore no time travel nor switching 
> is necessary to human thought.

At least all simultaneous interpreters feel this need.
They have to memorize a lot of details until they arrive at a point where
they can release them.
The same is a case when I listen to 

	"On 8th of May 1945 .... "

although the strain on memory is not very heavy here, since the time is
short.
 
> Sorry, but around here I say I live in Fairfax, whereas with distance, I 
> say I live in Fairfax VA, or even Fairfax VA USA.  

Again, when I listen to "Fairfax VA USA", I feel an unfriendly strain on
my memory.  I have to listen up to "USA" until I can finally recollect
that almost already forgotten syllable "Fairfax" and place it on my mind's
map of the USA.  And how do you know that I really want to maintain such a
detailed map of where people live?  Perhaps I would be happy to just
memorize that Lojbab lives in the USA.

> Note the standard order, which again is used in almost all postal
> systems.

>From Japan through China and Russia to Germany different orders are used.
But they are all under assimilation pressure by the US system.

> If you already know that Fairfax is in VA or the USA, the redundant
> information is unimportant, container or not, and goes later.

If I already know that, then I don't need an addressing system.

We are, I hope, analysing the situation from the communication
perspective, not from the speaker perspective. 

Things can of course become infinitely complicated by the fact that
addresses may contain known placenames at any level of the hierarchy.
But this doesn't really change anything, it just adds confusion into the
analysis.  That's why I suggested to limit the debate to dates for the
time being.

> >Again, your thinking proceed from the user than from the recipient.  For
> >the debat about whether a certain language design is appropriate to the
> >requirements of human thinking, the latter is far more important.
> 
> Per above, if I am the recipient, and I already know the container 
> information, it is not important at all.  Thus its importance is context 
> dependent.

"the listener already knows the container information" can mean two
things:

(1) The speaker has assumed a too far outward container, e.g. "usa"
    although everyone knows that we are talking about the geography of 
    Virginia.
(2) The speaker is using placenames in the address hierarchy that
    don't need to be further expanded, because they are known, such as

    galaxy->world->usa

    or usa->va->fairfax	
	 
Case 2 is the extra complication that you introduced at the beginning.
We should restrict the discussion to case 1 first.
 
> Except that because of the unidirectionality of time, there is no way of 
> knowing that the expansion is needed until it is too late.  I say "the 
> 25th" and you look at me confusedly, and I say "of June"; I cannot go back 
> and expand before the date.  Additional information has to be ADDED, and 
> cannot be prepended.

Again you insist on thinking from the lazy speaker's viewpoint.
Of course, a lazy speaker will first assumes a narrow frame of reference
and then try out how far outward he has to go.
 
> >then the listener can decide that "197x" is all he wants to know and stop
> >listening to the rest.
> 
> But what if the listener already knows the 197x and doesn't need/want to 
> listen to it?

Then he gets the redundant information at the beginning rather than at the
end.  That doesn't make things difficult, because the route from the
periphery to a known center is previsible and requires no strain of
memorisation.
 
> Oh, you are observing that numbers are big-endian.  But I contend that they 
> are nothing of the sort.  They are words, units of meaning that the brain 
> processes as a whole, without considering the breakdown into 
> symbols/phonemes.

I contest that even for written language.
But for oral speech it is even clearer that you don't listen to 

  four thousand three hundred and eighty seven point three nine four ...

as a whole.   You might even stop collecting more information after having
noticed that it is more than four thousand.  So you will be glad that the
number is in big-endian order.

There are language habits of putting numbers into little-endian (like
"fourteen"), but that only works as long as nobody is making serious use
of numbers.  	

  (Likewise I suspect that dates and addresses and names 
> are usually taken as single units despite being written as multiple words, 
> and are not switched around or manipulated or anything else in most 
> people's brains.)

If you want to use the information, you will have to manipulate it.  Only
by manipulating it can you hope to be able to memorise it.  Any mental
training will tell you that.
 
> I don't sort history dates.

Do you never subtract 1.9.1939 from 8.5.1945?  Subtracting is just one of
many kinds of comparisons.  It is quite unpleasant to first subtract the
days, then the months and then the years.

Whenever you want to get a clear picture of what happened during some
duration of time, you will compare and sort dates.

> What's good enough for most postal systems is good enough for me.  As
> I said, Lojban is not a reform movement.

That amounts to saying "The ways of bureaucracy are good enough as a
system for organising my thinking.  They will certainly be proven superior
by the S.W. hypothesis".
 
> >As I already said repeatedly, if it is obvious, it will either not be
> >there at all (in case of good speaker-listener consensus) or the listener
> >will treat it like a recapitulation of something he already knows
> 
> which I contend is not listener friendly because it is putting unimportant 
> information first (unimportant because it is already known).

see above.
I keep trying to differentiate this argumentation into several levels, but
you keep mixing them back into one.
 
> We cannot know.  We just know that all language patterns are based on human 
> thought.  That is because they cannot be based on any other kind of thought.

They are rather based on the inertia of social conventions.

If they were based on human thought, nobody would be fascinated by a
Logical Language.
 
> >That is because normal email addresses are more humanized than uucp
> >addresses.
> 
> i.e. they match human thought patterns rather than computer thought 
> patterns.  

no: match anglosaxon conventions more than logical patterns.

> Email addresses are not similar to any other kinds of addresses, 
> so their ordering must be based on new human thinking (as opposed to 
> caveman articulation).

they are modelled on anglosaxon place addresses.  The 'at' letter even
suggessts that they are modelled on colloquial English.
 
> >  The uucp form is a very good man-machine interface convention,
> >and the UK for a long time insisted on writing email addresses like
> >
> >         john@uk.ac.cambridge.cs
> 
> But not
> uk.ac.cambridge.cs:john

right.
 
> which would be what your argument would predict.

right.
 
> And indeed, why was UUCP changed at all, if it was a good system.

It wasn't changed.  But it is (and always was) a system that operates at a
low level which people want to be isolated from.  UUCP addresses contain
the whole route that a message must travel.  People want to leave those
details to the computer.
 
> >For the listener / forwarder, the toplevel domain is essential for getting
> >started.  Whether one wants to continue and up to which stage one wants to
> >go on (a2e? phm?), is a secondary question.  If you want to go on up to
> >phm, then you also must first go to de.
> 
> Not if I am already at a2e.de

right.
a big-endian address starts at some common container, preferably the
one that is appropriate to the communication situation.
a little-endian address starts at the speaker's side and proceeds outward 
to look for the common container.
  
> > > ONE part of the language has leftward expansion.  All the rest is 
> > rightward.
> >
> >This one part is the part that is at the origin of date structure
> >conventions in all languages that I know of.
> 
> Then why doesn't it match the order patterns for those languages?  

It largely does.  

> English adjective-noun corresponds to tanru. But we are little-endian
> in dates and addresses.

(1) French/Latin is at the basis of many English patterns, although they
    are also verbalized by English non-tanru:  5th of May.
(2) English has been tending away to big endian (e.g. May 5th) while
    French hasn't
 
> Language evolves by subconscious instinct, not by conscious manipulation.

Language is determined by social conventions, which of course also depend
on the instinct of a lot of people.  These conventions are tacit and below
the level of conscious planning, but they have a life of their own and do
not directly reflect individual's subconscience.  And they can be raised
to a conscious level, if enough people in the language community want it
to happen.

-phm