Re: [lojban] Re: A (rather long) discussion of {all}

[John E Clifford <clifford-j@sbcglobal.net>]

--- Maxim Katcharov <maxim.katcharov@gmail.com>
wrote:

> On 5/17/06, John E Clifford
> <clifford-j@sbcglobal.net> wrote:
> > MK's proposal in a nutshell.  (For
> corrections,
> > additions, explanations, etc.)
> 
> A draft, yep.

Very much so.
 
> > 1.      1.     Lojban's method of introducing
> new
> > referents is inefficient, ineffective and
> > incomplete.
> 
> No. I believe that it's clearly efficient, and
> it's effective, 

Glad to hear that.

> but:
> a) it does not cover situations where context
> is ambiguous
> b) it is ineffective where it's critical (or
> where you want) to be
> fully unambiguous
> 
> I'd like it to be very clear that I am not
> arguing /against/ the
> current method, but that I'm essentially
> arguing for a method to cover
> areas where the current method fails.
> 
> I see the method that I'm presenting as the
> proper way to say things,
> where the method currently used is a /very/
> useful shortcut. But we
> shouldn't have to rely on this shortcut - a
> paved road should be
> available if we need to haul something
> important.
> 
> I argue against anyone who says that the
> current method covers
> everything. I argue that this method will cover

"this method" = the one you are proposing (not
the current one)

> (a), and will just
> about completely cover (b), and so is better
> than the current method
> (shortcut) on its own.
> 
> > Lojban descriptions depend upon unmentioned
> > factors  which conversants infer from the
flow
> of
> > the discourse and from the physical
> environment
> > in which the discourse occurs.  In
> particular,
> > the set of things having a mentioned property
> is
> > restricted to those things of that sort that
> are
> > relevant to the discourse at the moment.  The
> > inferences involved in correctly identifying
> the
> > referent of a given description may be long
> and
> > complex and the factors on which they are
> based
> 
> No. They needn't be long and complex at all.
> They are usually quite
> simple, it's just that there is a significant
> number of cases where
> there are a lot of simple interpretations.

I don't get this point.  Most inferences are not
complex (glad to hear you say so, though I think
a very large percentage are complex) but ...
what?
Is this just ambiguity, cases where several
interpretations are equally well justified and
there is no deciding factor?

> > may not be obvious to all the conversants. 
> Thus,
> > there is a strong possibility that the
> referent
> > will be missed, requiring further discussion
> to
> > correct the mistake.  But especially, this
> > relativization of reference makes it
> impossible
> > to refer to the things which have the
> property in
> > question but are not relevant to the
> discourse so
> > far, for example, to jump from the bears we
> have
> > been talking about to all bears absolutely
> (ever,
> > existent, imaginary, merely possible, etc.)
> or
> > even just all in the area, including those we
> > have not been talking about.
> 
> It's not strictly impossible to refer to things
> completely now. It's
> just that the method proposed for doing so is
> very crude and attempts
> to solve problems after the fact. And sure, you
> can clear up what you
> meant after the fact, but sometimes that's not
> very useful, and it's
> never as useful as clearing it up
> pre-emptively.

Since I would introduce talk about all bears,
past, present, future, merely possible, and also
the impossible ones, by saying something like
just that, unless the flow of the discussion had
made some parts of that obvious, I don't see what
could be better.  And, of course, this is just
the method I would propose for doing this.  How
is yours less crude (come to that, how is this
crude)?  This is not argumentative, just seeking
more information about just what your complaints
and your correctives are.
 
> > 2.      In every situation (a discourse
> carried
> > on in a given environment), for any object or
> > group of objects, there is a description 
> that
> > applies (and will be seen to apply) to
> exactly
> > that object or group.  Further, this
> description
> > does not rely on the flow of discourse  and
> > relies on the environment only for ostention;
> > that is, it relies only on overtly mentioned
> > properties and deixis.  It thus avoids the
> > difficulties that make Lojban descriptions so
> > fallible.  And, being not relativized to the
> > discourse, it can introduce things that are
> not
> > already relevant to this discourse.
> 
> Yes, roughly, though I'll have nothing to do
> with the phrase "that
> make Lojban descriptions so fallible" - they
> aren't "so" fallible:
> they're very good for most circumstances. It's
> just that they don't
> cover things as well something could be
> covered.

Glad to hear it, again.  I put these remarks
about the failings of Lojban in to get some
motivation for the proposal.  I see that a lesser
motive was meant and will suffice.
 
> darves: I hope that John has addressed your
> objection, because I'm
> uncertain of how it relates to this point #2.
> It seems to relate to
> what I said previously regarding how we all
> have a limit of what we no
> longer consider a bear. Most of the time, our
> differing limits will
> not cause a problem. Even the nominalist, for
> all practical purposes,
> will understand me when I talk of some three
> bears, and won't argue
> that they aren't bears because they don't
> perfectly match the "ideal"
> bear.

A nominalist will certainly not talk abbout an
ideal bear; he is committed to there being no
such thing (and, indeed, no general property of
being a bear).

> > 3.       Lojban needs a way to use these
> > descriptions ("complete descriptions")
> clearly
> 
> I call them "complete restrictions" - that is,
> no other restrictions
> implied by context are applicable.
> 
> > marked as such.  Since these descriptions
> tend to
> 
> You don't quite mark them. You simply say that
> you mean *all* (every
> single...), and this incidetally always is a
> sign that the restriction
> is complete.

But, since the present system uses {lo ro broda}
to refer to a contextually restricted (rather
than overtly restricted) group, you are taking
{ro} to be used only for complete restrictions;
that is, it is a clear indicator that the
description is complete (else it would be
incorrect to use this expression).
 
> > be longer than the ones Lojban ordinarily
> uses,
> > and the ordinary ones work well enough for
> > ordinary cases (when we mean to stay within
> the
> > already given bounds of relevance), Lojban
> should
> > keep most of it present descriptions for the
> > ordinary cases, but some redundant or little
> used
> > form should be set aside for complete
> > descriptions.
> 
> Yes, roughly. But this implies that I want to
> include some marker or
> whatever, which makes this sound somewhat
> weird. Let's just say that I
> want {ro} to mean "damn well *all*. Every one.
> Not just the ones in
> context - no. I mean every one, that will
> exist, exists, has existed,
> [and so on]". 

I am afraid I don't see the distinction you are
making here.  I suppose it is that {ro} does not
SAY the description is complete, only that it is
proper to use it only if the description is
complete (and thus the parallel for the present
system where {ro} does not SAY "all relevant
cases," but is used in such a way that tha is
what the referent turns out to be).

>I'm saying that the idea that {L_
> ro cribe} should mean
> "all such that are relevant and such that are
> bears" is *strange* -
> {L_ ro cribe} should be "all such that are
> bears". If you want to say
> "the most contextually sensible number", just
> leave the inner blank,
> and the listener will assume exactly that.

Well, I wouldn't, but we are dealing with your
understanding of Lojban for now, so I'll take
that as a given.
 
> > 4.       The form {lo ro broda} is redundant
> for
> > relativized descriptions and so could be used
> for
> 
> how are you defining "relativized
> descriptions"?
> 
=== message truncated ===
<<> 4.       The form {lo ro broda} is redundant
for
> relativized descriptions and so could be used
for

how are you defining "relativized
descriptions"?>>

That is one which picks its referent only from
those established as relevant by the flow of
discourse so far.


<<>  complete ones (and only for them).  In
> relativized descriptions, {lo ro broda} refers
to
> all the brodas relevent to the discourse so
far,
> but this is just what {lo broda} refers to in
> relativized descriptions.  Further, since a

Yes, I find the blank inner and the inner {ro}
redundant.

> complete description exactly specifies its
> referent, we are indeed interested in all such
> things.  The expression {lo ro broda} would
thus
> be accurate in terms of Lojban meaning, which
the
> present, relativized, {lo ro broda} is not: in
> the latter {ro} does not mean "all" but "all
> relevant ones" or some such.

I'm not sure I understand what is being said
here, but yes, {lo ro
broda} by my proposal does /not/ mean "all: such
that are brodas
and/'intersect' such that are relevant/in
context".

> 5.      So, Lojban should adopt the convention
> that {lo ro broda} indicates that {broda}
(which
> may be complex, of course) is a complete

What is meant by "which may be complex"? That you
can {poi} it? Yes,
you can, and most usually will {poi} it.>>

I just meant that {broda} stands in for (as it
always does) what ever the total predication
expression is, including {poi}s and tanrus and
whatever else can go there (modals, ....).

<<> description of the intended referent.  And,
> wherever using a relativized description
> involves vagueness or ambiguity ? or even just
> complications that extend the needed inferences
> unduly -- in the background on which  
> description,
> complete descriptions should be used instead.

Yes, this seems correct.

> And, of course, always for cases where the
domain
> of relevance in a discourse is being changed,
> whether by restriction or expansion.

No, not always. If it serves your purpose to
change context in the
usual way (which I've referred to as "saying
something that makes no
sense in the given context, therefore you must be
talking in a
different context"), you can go ahead and do
that.>>

These seem to me usually just to be cases of
being more explicit about what is meant.  Of
course, since this is a shift of relevance, it
will appear incongruous, but that is not a
significant part of its functioning.

<< But if you want to
be precise in this context shift, you now have
that ability.

>  A.    Tied in with this, though not essential
to
> it, is the claim that, for every property, P,
> there is a set (or whatever) of all the things
> that have (now, at some other time, in the
realm
> of possibility, ?)  and that this set (?) is
the
> referent of {lo ro P} (as corrected by the

Yes, though the bounds of this set may vary from
individual to
individual. This is rarely a practical problem,
especially if one
assumes that the speaker's bounds (assuming that
the bounds are
somewhat sensible) at the time they said it are
the bounds that should
be considered.>>

This emendation is problematic for me since it
seems to say both that there is a single unique
set of everything that has a certain property and
also that this set may be different for different
people.  How can this be? If the set is unique it
is not relative to a person and contrapositively.
 

<<This is the 'vagueness' that isn't quite
covered by my proposal, and
it surely isn't covered by the current usages.
And the odds of it
being a problem are drastically slim.

> proposal).  Relativized descriptions may be
made
> complete by explicit restrictions carving out
> subsets of this set.

If I understand what you mean correctly, then not
quite: they're made
complete when you say "listener, I mean *all* of
them, and therefore
it is implied that you shouldn't use context to
cut the set down
further".>>

I guess my point is that, when a description is
relativized so that it does not refer to all of
them, it can be made complete by adding
explicitly what was supplied implicitly by
context, so that it does refer to all of the new them.