Posted by Ben C on 02/21/07 10:14

On 2007-02-21, dorayme <doraymeRidThis@optusnet.com.au> wrote:
[...]
> No one is sure how these things should work. My point has been
> that no help can be expected from certain quarters (mainly
> cumbersome obscure linguistic theories). If website making is to
> retain an ability to express information and style to accommodate
> all the important varying criteria without railroading the
> process into something that severely stifles creativity, then we
> are going to have to solve problems of a very much bigger order
> of magnitude than is supposed by the naively idealistic. The
> problem is a deep one to do with machines understanding meaning.
>
> AI has been in its "infancy" for a long time now and for very
> good reason. We have little idea about the operating system of
> the animal brain, we have some knowledge of the hardware but only
> in a rudimentary way - the truth being that we never will be able
> to describe the hardware properly without the descriptions being
> informed by a better theory about the overall "operating system".

Agree re "AI" and the fact that we're nowhere near a program that can
understand the user's intentions, also that cumbersome obscure
linguistic theories are unlikely to help much with that.

I don't know exactly what AD had in mind, or what current tools do
anyway. But I think the way to do it is to watch what good web authors
do and turn that into a tool. One that tries to replicate their
processes, not their intelligence. So a tool with actions to reclassify
and organize the information, rather than one that just pretends to be a
word processor or DTP, might work. Part of that might be actions in the
tool to author the "meta-structure" not just the structure, which is
where the obscure theories came in-- taken totally out of context, where
they made little or no sense, into one where they might make some.

[...]
> While I am at it, have you noticed how every few days if not
> every day on this ng, someone will ask how to automate this or
> that? There are sometimes nice replies, sometimes pie in the sky
> ones. All sorts of handy little tricks and tips and helpers are
> offered. The best we will do for a while is to have whole
> collections of specialist programs that do simple things well.

Even better than that a recognition that operating a computer for all
but the simplest tasks usually comes down to programming it in some
form.

You get much further with computers if you understand roughly what they
actually do, which is run programs. I've seen people invent recursive
programming for themselves using circular references in spreadsheets,
and a lot of creativity with search and replace and keyboard macros from
people who didn't realize they were programmers.

> To learn how to use them and use judgement. I am pessimistic that one
> program can coordinate all these things, can take over the "judgement"
> part for the foreseeable future.

Exactly right. Good tools don't hide what you're actually doing or try
to do the things you have to do for yourself, but they can give you
hints and prompts about the stuff you have to do, and help with the
details.

[Back to original message]