Comments from pc on various issues

["Mark E. Shoulson" <shoulson@CTR.COLUMBIA.EDU>]

>Date:         Mon, 15 Mar 1993 15:27:42 -0500
>From: John Cowan <cowan%SNARK.THYRSUS.COM@CUVMB.CC.COLUMBIA.EDU>

>1) I bounced the idea of using unbound "da" within "ka" abstractions to
>signify a propositional function off pc; he agrees that it is the Right
>Thing.  Any other scheme, he says, for representing (lambda (x) ...x...)
>with an explicit arg-list would be intolerably clumsy.

Yes, but I dispute the fine points.  It's hard to explain this, but somehow
I don't like the idea of using "da" series words for lambda placeholders.
It doesn't fit what their job is.  If you miss the definitions of "da" in
Lojban, generally all that means is that you misunderstand or get lost when
someone uses them later, but only in that you might think they mean
"something" rather than "something in particular".  But with the new
method, well, somehow it feels like the change distinction in meaning is
more basic.  If I missed hearing that "daxipano" is unbound, or for some
reason thought it was when it wasn't, I might get a totally different
picture.  Moreover, people are likely simply to say "da" if it becomes
common usage, whether or not "da" has been bound (as we already see).
Personally, I'd rather see "ke'a" used for the place-holder. It already
partakes of the placeholder nature, and it's unlikely we'll use more than
one placeholder in these sentences (lambda-calculus notwithstanding), and
there's always ke'axire.  Perhaps a better move in general would be to
create a cmavo that's sort of a cross between "zo'e" and "da", that would
refer to something more definite than "zo'e", but never be bound.  So I
could say "mi se xadni pada" and not worry whether or not I bound "da"
a while ago.  Perhaps just convention, reserving "di" and its subscripted
variables for such use.

~mark