Connective, Collective, Corrective:

The Future of VRML

Delivered at VR World
Boston, Massachusetts
30 November 1995


At the outset, Iíd like to thank the organizers of VR World, who have given Tony and I carte blanche, a platform from which we can speak to matters of importance to ourselves. That means a great deal to me; working outside the boundaries of the corporation has meant that I can always say whatís on my mind - and more importantly, that I can speak my heart on matters of importance. Iíd like to do that here, today.

This talk is the first of three that Iíll be delivering over the next two weeks; the others will happen in two weeks time, one on the other side of town, at the World Wide Web conference, the other at the VRML Symposium in San Diego. Together, they trace an evolution - understanding our origins so as to be receptive to our destiny. I must state that I am speaking for myself in this talk, and not for www-vrml, or the VRML Architecture Group; these are my own words. All of that said, let me begin.

Growing VRML

Two years ago, Tony moved to San Francisco, and this whole strange trip really began.

This year has been wild ride, beyond anyoneís imaginings. I've traveled around America and Europe, watching VRML grow. What Iíve learned may help inform us where VRML is going in the future, so that we can plan for it -or at least get out of the way.

The seed of the idea for VRML - which we called Labyrinth - became a functioning reality in February of 1994. At that time, almost no one cared - the Web was still entirely an academic curiosity - but we did find someone who did care, someone who did encourage us: Tim Berners-Lee, the father of the Web. He saw the value in our work, and invited us to present at the First International Conference on the World Wide Web. At this conference, Dave Raggett coined the phrase, "Virtual Reality Markup Language," and we found a broad base of support to turn our concept into something viable for a whole community of users.

The beginning of VRML can be directly connected to the launch of the www-vrml mailing list, in the first week of June, last year. It quickly swelled to nearly a thousand members, and has remained that size ever since. From this list came the VRML 1.0 specification, the product of a process of a community of connected users, collectively creating something greater than any single individual could produce. We got there amazingly fast - from nothing to a draft specification in just five months. We refused to invent everything ourselves - like nature, the VRML community would take the best of what was available and bend it to meet our own ends.

We spent another six months refining our work - and developing the first VRML browsers, released in May of this year. The enormous popularity of our work - it has fired imaginations around the world - is proof that, collectively, weíve created something worthwhile. More than this, it demonstrates that dedication to a vision is the root of our community.

So now we take a look at our work, and seek to improve it, marrying Java and VRML to create the foundation of a global, interactive cyberspace. All our work before this is merely prelude; everything after might well endure for decades. Weíve yet to see even the first fruits of cyberspace, but in a yearís time weíll understand the power of the tool weíve crafted, and weíll be able to share that knowledge - and our vision - with others.

Wiring Under the Board

The growth of VRML has three distinct phases; first, a period of connection, when we established www- vrml and defined our goals for a VRML process; second, collective intelligence, the emergent property of connectivity and vision, producing a specification; lastly, correction, as the VRML community reflects upon its collective efforts and adjusts its sights, reinterprets its goals, and continues forward.

I discovered this three-fold process while analyzing the history and growth of VRML, but later found that it applies in many other areas. Let me illustrate with a few examples:

The Internet was established in the late 1960ís as a strategic defense initiative for the nationís computer networks - particularly those used by the military. The connective phase, spanning 1969 through 1989, brought hundreds of thousands of individuals into a single community - a single room, as it were - creating space for collective intelligence and collective action.

In 1989 a software engineer at CERN, the European Center for Nuclear Physics, developed a system which turned the entire Internet into a continuous document space, with a universal user interface. The World Wide Web, which set out to solve the problems of scientific data management, harnessed the collective intelligence of the Internetís academic and scientific communities, to produce the Web, an entity much greater than the sum of any of its parts. The great success of the is built upon the collective hard work of hundreds of thousands of members of a global community. The Web is not a space of ownership, in fact, it tends to favor a rich pastiche and equal access for all, while it deprecates restriction and authorship.

The Web is not enough; weíve built it on a shaky foundation known as the Universal Resource Locator, or URL. Its incomprehensibility is the price we pay to interface with the machinic intelligence of the Web; for example, let me tell you the address of my home page; itís Now, you might remember that, for a few minutes. Maybe long enough to write it down. But in an hour or a day, youíll probably forget it. And my mother - who has an AOL account and can cruise the Web - doesnít really understand the URL at all. Itís just gobbledygook. Yet, weíve built the Web upon it.

If I were to turn that example around, and tell you, "Get to Market Street, take a left on Sanchez and a right on Henry", that would be information you can understand. Itís human-centered, and expresses the essence of VR: information made sensual makes sense. Computers have been trapped behind ďsophisticatedĒ interfaces for fifty years, hiding an embarrassing truth - how stupid they really are.

So, as the Web community reflects on their collective effort, VRML is born, as a correction to the abstract URL space of the 2D Web. VRML offers tangibility, sensuality, and sensibility in a thoroughly conceptual space. Somehow, space is the right idea, sensuality is the right idea, and so the Web grows - correcting a mistake and encompassing a solution - VRML. Here it is again: connective, collective, corrective.

Corrective Intelligence

There are many forces converging on VRML; in part because itís suddenly become a hot topic, in part because it established a beachhead in cyberspace - something people have been waiting years for. Now commercial interests are grappling for a foothold in the community, and the government is asking how it can help, and everywhere the base of VRML users are pushing the envelope - in browsers, in content, in wild ideas. And we need to be able to satisfy everyone.

In the beginning, it was easy; www-vrml served as the Agora of our polis, the place where we could all meet to discuss the great issues of the community. It was and is an incredible environment - some of the brightest minds in simulation spend serious time working to create a common base of knowledge, as codified in VRML specifications and practices. Iím always in awe of it, how open people have been to a process designed to provoke the best they have to offer. The list is a true meritocracy - like all viable Internet communities, youíre judged not on the color of your skin, or your education, or your background, but on the content of your postings.

Still, this was not enough; we had a collective intelligence, but even this had to be coordinated in a substantial way. We had to bring form to our efforts as a community - this became acutely clear by the beginning of August of this year. For this reason the VRML Architecture Group, or VAG, was formed, to serve as the coordinators of this collective intelligence, gathering the best of the ideas generated on www- vrml into substantial documents - specifications, clarifications and examples - and then sending these back out into the community for further refinement. The VAG walks the fine line between autocracy and rule of the mob; we see ourselves as custodians and stewards rather than owners of the specification process. We have no power, only the authority of our actions. This keeps us honest, because we canít enforce, we can only suggest - and do our best to look like intelligent adults, instead of blathering fools.

The VAG has been incredibly effective; in the four months since our first meeting, weíve fleshed out a series of enhancements to VRML which increase its viability, and meet the vast majority of the requirements of its users. Weíve done everything possible to open the VAGís machinations to public scrutiny - although weíve been very strict at keeping the membership down to 10, weíve published all of our notes, papers, and other materials. We feel itís important to keep the community heavily involved in a process which theyíve initiated, and although it falls to us to argue out the arcane points of syntax or functionality, the VAG understands its relationship to the community - that of servant, not master.

When the VAG had its first meeting, we agreed to a set of criteria - measurements - by which we could gauge our success. Foremost among them is the success of VRML across a broad user base, commercially, academically, and with the community of do-it-yourselfers who are the shock troops of the Internet. What we hadnít planned for was our success. In fact, we see a VRML so successful that all of a sudden the VAG finds itself confronted with a thousand differing demands, overwhelmed trying to deal with political, economic and technical imperatives - but the VAG was established only to extend VRML, purely a technical objective.

The greatest threat to VRML comes not from some competitive standard - we can expect those, and can expect to move faster than any competitor. Our threat comes from within the community; if we lose cohesion and fragment into multitude of different communities, each with their own peccadilloes, specifications and implementations, weíll actually be worse off then when we began. Thatís something to be avoided, at almost any cost.

Furthermore, we must find a way to engender a community that can remain open to all, and answer the interests of any of its members. The informal community works, to a point; but, for example, we canít ask the Fortune 500 to bank on an ad-hocratic process forever. Theyíll come along for a while, but unless they see systemic cohesion, theyíll find our creative anarchism to chaotic for their tastes, and move outside the community to meet their needs.

In other words, VRML must grow up. Itís been a fine childhood and adolescence, but now, jobs are in the balance, and the direction of computing for the next decade is in our hands. Itís a responsibility that canít be taken lightly. We need to pass into adulthood - responsible and focused.

So, our community, connected, collectively works through the VAG; but to correct the shortcomings of the VAG, we need some other organization.

The "C" Word

We need a consortium. Itís been my feeling that weíd have one as soon as we needed it, and, once we reached that point, it couldnít be stopped. After floating trial balloons at SIGGRAPH, I came to learn what people wanted in a consortium, and Iíve been talking with both the VAG members and other members of the VRML community in order to develop some concrete plans and goals for such an organization.

The function of a VRML Consortium must be threefold.

First, it must be dedicated to research; VRML is a multi-year research project, taking place in laboratories and offices throughout the world. The Consortium can serve as a focus for this research, and can conduct research that is considered vital to the future of VRML, but isnít being done privately or academically. Clearly, the Consortium would work closely with other organizations like the World Wide Web Consortium, and academic institutions like Brown and UCSD; it can serve as a focal point for the exchange of ideas, and give shape to a formless research process. It can and will maintain the basic of public domain source code which is at the heart of VRML. Finally, we must also plan to establish a laboratory where the best and brightest of VRML can come to create the wild-eyed projects that extends the horizons of our work.

Second, the Consortium must draft and maintain both specifications and standards practices for all areas in VRML; the specification of the language, its integration with other environments such as Java, the implementation of browsers and VRML practices and techniques, such as cache management, scalability, and so forth. The Consortium must subsume the VAG, and the VAG itself must grow into at least four different bodies; one to deal with graphics issues, one to deal with networking issues; one to deal with browser issues, and one to focus on languages and interactivity. Already there are persistent complaints that the VAG is too small, too insular, that it isnít open to the VRML community. Enlarging the VAG will address these issues, and will continue to broaden VRMLís base of support.

Third, the Consortium must handle compliance and conformance issues. Weíve now started to see the beginnings of what could be a disaster for VRML - implementations that donít interoperate. There must be one standard, and developers must be able to test their implementations against this standard, throughout the development process. At this time, various members of the community have created their own sites designed to put VRML browsers through their paces, but thereís no support - something that a Consortium would be able to provide.

These three functions - research, specifications and compliance - form the core of the Consortium; they exist to address the needs of a community which could already make use of them, if they existed.

Conclusion: Last Try

This is our final chance - if we donít have a consortium soon, expect to see VRML disintegrate and disappear within a year. This isnít an alarmist statement - the forces tearing at VRML are already substantial, and will increase drastically over the next year. We need to create a community thatís cohesive technically, politically, and economically. Without it, weíll be gone, overrun by another community that had more foresight than ours.

I am asking all of you - both personally and as a member of the community - to think through these proposals. Nothing comes for free - and it will take serious finantial backing to get a real consortium off the ground. We need to have a real discussion about a consortium, about what itís meant to do.

Iím putting myself on the line - I think itís absolutely necessary that we begin creating this consortium today, and for the next few months, itíll be the focus of all of my efforts in the VRML community. I hope weíre all ready to grow up. If not, weíll find ourselves at the starting gate. Again. Iíd like to think weíve grown past that.

Thank You!

Mark Pesce
28 November 1995
Schipol to Dulles