[ExI] language

[Tomasz Rola]

On Thu, May 01, 2014 at 01:27:50PM -0500, William Flynn Wallace wrote:
[...]
> Do you (plural) think that it will be possible in the not near but not far
> future to say to a computer "I want you to write and execute a program to
> do xyz." ??  bill w

Depends on perspective. From my point of view it happens
routinely. All the elements on technical side are already in place,
affordable voice recognition being the last one to have come. The
first element, besides working computer, was a programming language -
in this case, Plankalkul appeared about 70 years ago. People working
on lambda calculus might have done something in this department even
earlier, in 1930-ties.

So, today, I fire up Emacs (an editor) and write "xyz" as a note to
the computer. "xyz" can be up to many kilobytes of text long
(hundreds, too, but rarely this long as a one long piece). Then I tell
to the computer: "gcc xyz.c -o xyz" which means "write program xyz
from description in xyz.c" and if all goes well, I can run it next,
sometimes it even works. Actually, most of the times I tell to the
computer "make", and it takes another note in Makefile, which
describes how to make and what.

There are people who do all of this with Emacs paired with trained
voice recognition, so here you are. I never so far saw necessity which
is why I prefer to write notes.

You could now protest that it is not what you meant. But I see no
problem. I use simplified English, interspersed with algebraic symbols
and numbers and it is in fact a language in which I talk to the
computer. How much simplified - depends on actual "dialect". Most
langs like C rely on about 30-50 words as their core vocabulary. In
Common Lisp there is almost 1000 words and I still have no idea what
majority of them mean. Unix-like operating system employs from
hundreds to thousands of words, each with special options to choose
and modify exact meaning of many of them. As far as I can tell, it is
always possible to create new words by defining their meaning with
words already known (at least in systems with which I am willing to
spend my time).

So this is how I see it. I guess such views may be grossly unknown
nowadays in the era of point and click "operating (hehe) systems". The
"click language" is very poor in meanings and actions. It makes a user
equal to foreigner who will never learn a language of a country in
which he is supposed to spend the rest of his life - I believe this is
only because lack of will, not lack of some brain part (although it is
very much possible that certain neuronal wiring patterns help a
lot). But this is not my problem, to be frank.

Now, I realize what you wanted to ask was "when one can talk to the
computer in a natural language like one would to the waitress or car
salesman - or one's own personal butler". I hope this will never
happen. Human language is too ambivalent in many places, often the act
of communication is designed to hide meaning or to lie about
it. Besides, it sometimes happens that humans don't know what they
want and so they cannot tell it properly. The world goes on somehow,
but I have doubts if this is thanks to our intellectual
abilities. Pure luck may be better culprit candidate. Likewise, that
we can somehow transfer meaning from one head to another is probably
lots of luck and common roots. The more different cultures, the more
misunderstanding, despite using the same language on the surface. The
computers either form or will form another such culture, and a very
different from the human one.

Regards,
Tomasz Rola

--
** A C programmer asked whether computer had Buddha's nature.      **
** As the answer, master did "rm -rif" on the programmer's home    **
** directory. And then the C programmer became enlightened...      **
**                                                                 **
** Tomasz Rola          mailto:tomasz_rola at bigfoot.com             **