RE: Mein gott, Nelson was arguing for CSS

David G. Durand (
Wed, 10 Nov 1999 16:39:49 -0500

>> FWIW, the separation of presentation from structure/content predates both
>> CSS and Nelson's article by about a quarter of a century.
>What are you referring to as the predecessor?
>This could make for a pretty interesting FoRK one-up-mathon: what is the
>earliest identifiable recognition of the value of separating content from
>presentation? Here "recognition" could be either explicit, in words, or
>implicit, by systematic use.

I'm not sure exactly how to define the scope of that, in order to make it a
real competition. If we deal just with computer text-processing systems, at
least partial recognition of the value of separating structure from
formatting was present in Engelbart's NLS. Now NLS had a somewhat limiting
notion of structure, and highly limited presentation media, but one could
argue that the basics were there. I would be that this was part of the
earliest systems from '62-'64, but it was surely well developed '69 (along
with a layered, tool-based software-engineering approach).

I have a pre-Goldfarb paper on generic markup (a more specific issue than
just separating structure and content). I'll have to look for it to pull
out the details, but I think the date was somewhere between '69 and '74.

Goldfarb was proposing very complete generic markup ideas in the 70's, and
implementing them at IBM.

The interesting thing about Nelson's proposals is not the separation of
content and formatting (an idea he seems not to care about with the same
passion as many SGML/XML types), as separating the storage of structure
from the storage of the data. I think this is what people are calling out
as "parallel markup."

Coombs, Renear and DeRose's CACM article in the 80's points out the logical
problems in attempting a _full_ separation of structure and data. Much of
what we think of as "data" is in fact structural, like inter-word spacing.
This was an explicit invention (and not that early an invention, either).
Punctuation is another kind of markup commonly confused with data. These
aren't even purely academic questions, either, as anyone who has tried to
write typesetting code knows full well. It's really tricky to correctly
format word-breaks and punctuation in the presence of a formatting
boundary. Never mind the problems when you start tagging portions of

This is not to say that parallel markup is a bad idea, but to expose the
fact that neither embedded nor parallel markup canbe considered harmful in
themselves, but that they are extremes on a continuum where significant
tradeoffs are possible.

-- David
David Durand \ \ Director of Development
Graduate Student no more! \ Dynamic Diagrams
MAPA: mapping for the WWW \__________________________