October 16, 2004
The Vision: Cyberspace Protocol Requirements
In 1993 Electric Communities was a three-person consulting partnership: Randy, me, and our former Lucasfilm Games coworker and frequent collaborator, Doug Crockford. We had several clients, but by far the biggest and most important was Fujitsu. Fujitsu had hired us to do a business plan for bringing Habitat back to the United States, starting from the Fujitsu Habitat system they had developed in Japan. This was the project that eventually became WorldsAway (and which they eventually hired us to staff and manage during its formative years).
We, for our part, were far less interested in Habitat per se, which for us was yesterday’s thing, than we were in newer, more cutting edge stuff. The biggest limitations in the Habitat architecture, from our perspective, were that it relied on a unitary, centralized server, and that the environment was not user extensible. We wanted to put together a system which overcame these limitations.
We had some ideas about how to do this. We were heavily influenced by lengthy discussions with Dean Tribble and Mark Miller, who introduced us to the wonders of the capability paradigm for security and optimistic computation for implementing distributed systems. Dean and Mark worked at Xanadu during the time Randy and I worked at American Information Exchange. We got to talk to each other a lot because, for about five years, AMiX and Xanadu shared offices in Palo Alto (both companies being part of the Greater Autodesk Coprosperity Sphere at the time). But that’s another story.
Our arrangement with Fujitsu was that they would pay us to spend half our time working on WorldsAway and the other half doing advanced conceptual studies, planning, and prototyping for a next-generation system. They’d own the WorldsAway work and we’d own the conceptual stuff, though they would have the right to make use of the latter however they wished. It was a pretty sweet deal, especially when viewed in the light of these post-bubble years. We got to sit around and think great thoughts about pretty much whatever we damn well pleased while getting paid rather well for our trouble.
For the most part, I worked on the conceptual stuff (with a little bit of WorldsAway), Randy worked on WorldsAway (with a little bit of conceptual stuff), and Crock worked on figuring out how to present all this weirdness to the people at Fujitsu. After a couple of years of producing various presentations, studies, and reports, the ultimate end product of this effort was a hefty document gradiosely entitled Cyberspace Protocol Requirements, which we post here in public for the first time. (Note that although this is the final revision of this document, dating from February, 1995, the principal work was published in the March, 1994 in a form only very slightly different from this.) This document is really quite embarrasingly dated, but we decided it would be worthwhile to get it on the record. It’s probably also worth pointing here to the manual for Dean Tribble’s Joule language, which was included as an appendix in the dead tree version of the Protocol Requirements document. Joule was our first round pick for the foundation language for the infrastructure we proposed building.
An important piece of historical perspective to keep in mind, should you actually attempt to read this: there is one seriously major development that had not yet taken place at the time this was originally drafted, namely, the rise of the Internet. The Internet was just beginning its explosive takeoff at that point; indeed, we allude to this. However, though people certainly expected that some kind of dominant, global, electronic network was going to be the wave of the future, at the time the Internet per se did not appear to be the foregone conclusion that in retrospect it obviously was. It was a time when all the various large media companies (phone companies, television networks, cable TV operators, etc.) were announcing mergers and new strategic alliances almost daily, in an absurd and furious spectacle that we referred to as “The Dance Of The Dinosaurs”. Nothing concerning the look of the future was very clear. It was a time when people were still getting used to the idea that commercial use of the Internet was even legal (which it hadn’t really been for most of its existence up to that point, since it had been a creature of the world of academia). In particular, we were getting along with FTP and Gopher and none of us realized the the World Wide Web would be the big winner. Indeed, those of us who had been associated with Project Xanadu were especially surprised by the success of the Web, since its technical deficiencies were, in our eyes, so profound that we never quite believed that anyone could ever take it seriously (as Mark Miller put it, “if we had known at the time that the thing didn’t actually have to work, we would have been done a lot sooner.”)
Another disclaimer is that there is a lot of visionary stuff in here that I either no longer believe or believe in a profoundly modified form. I’ll have a lot more to say about this in my next post, which I’ll have up here soon.