An e-mail all the way from Argentina cited my HyperCard eulogy and informed me that the product had been mentioned in Tim Berners-Lee's original WWW proposal. Son of a gun, there it is. I never knew that.
The same search also pulled up some historical notes on HTML, which reminded me of other projects of the time that might have influenced the WWW. (We'll have to ask TBL to find out for sure.) Anyway, back in October 1988, a year after HyperCard's release, Jan Walker of DEC and John Leggett of Texas A&M organized something called the Dexter Hypertext workshop (named after the inn where the group first met). I attended on behalf of Apple and HyperCard, and most of the functioning systems of the time or before were also represented by team members. There wasn't a really firm agenda, other than to compare experiences on architecture, usage, wins and losses for mutual benefit. (These tales obviously belongs to a more innocent, less IP conscious time.)
Two things eventually came out of the series of four Dexter workshops. One was the widely cited 'Dexter' formal data model of hypertext, driven by Mayer Schwartz of Tektronix and Frank Halasz of Xerox PARC, with input from the rest. Another was a practical experiment in exchanging content among hypertext systems, which as I recall was instigated during a bull session while driving to the Houston airport after a meeting. The practical result was that I hired Jeremy Bornstein of Brown University, who had already worked on the early HyperCard-based Perseus project, as a summer intern with a charter to see what could happen. Jeremy collaborated with Victor Riley at Brown to design and test an SGML-like text markup, based on the Dexter formalism, that could move content among several systems (it was never used as a native format). This was demonstrated at the Hypertext '89 conference (written up in an ancient Jakob Neilsen trip report) and eventually published in the proceedings of a NIST hypertext standardization workshop.
While the Dexter Hypertext Interchange was SGML-influenced, there was another project going on separately that was specifically SGML-based. HyTime started as an effort to express music in SGML. It was driven by Charles Goldfarb, inventor of SGML, and Steve Newcomb, then at Florida State. At some point I became aware of this effort, and invited Goldfarb over from IBM to present to my Apple group. As I recall, a two-hour conversation between myself and Goldfarb in the Apple City Center parking ramp afterwards led to my concordance of Dexter and Hytime which is cited in the HyTime history. Hytime was very elegant, but also quite baroque, and too much overhead for wide adoption at the time.
Both of these efforts were pretty well known at the time. (The cited NIST initiative died after one meeting, when the industry representatives decided a standard was premature and bailed out. We were right.) Both Dexter models and SGML based representation were well discussed at the ECHT90 European Hypertext Conference, attended by both TBL and Robert Cailliau. In March 1991, Victor Riley posted information on Hytime to a list where it was picked up by the CERN team It's clear that by some time in 1991 SGML related features to the evolving HTML were in play at CERN, and by 1992 HTML could be expressed in an SGML DTD. Unlike Goldfarb, TBL and team were willing to cut loose from the overhead of a full and formal SGML system in favor of simplicity, and the rest is history. We got markup that could be banged out in a line mode text editor, and put off the reckoning with formalism until the advent of XML.
(NB: Most of this from memory and Googling around. It does remind me that somewhere in the attic is a collection of old conference notes and incriminating historically significant photos. If anyone cares about this stuff, I can dig some of it out. Send e-mail.)