The Great Leap Downward

The Great Leap Downward: Mark Pesce on the politics of VRML

Not everyone is convinced that 3D environments will provide quality entertainment. "The real problem is content, will always be content," says Andries van Dam, a computer science pioneer at Brown University. "What are we going to say to each other which is meaningful? Are we going to be reduced to the same banalities that we exchange at cocktail parties? We can do wonderfully well on the tech side, but the content is still a challenge." Read on for the downside of VRML.

Like all of the Silicon Valley myths, this story begins with two hackers hot on the trail of the Next Big Thing. It all seems obvious now: the Web that brought us text and images could bring us objects as well, a three-dimensional representation of the world. You could explore that new space, move through it; navigating the Internet would finally be more than just a metaphor. That was my starting point -- mine and Tony Parisi's. Three years ago this week, Tony and I followed a link in Mosaic and found -- to our delight -- that we looked into a 3D universe of objects.
Tony had only one question: "What do we do with it?"
This is the story of the birth of a standard, a bloody and painful meeting between the forces of commerce and community, forces that are at work anytime a new medium comes into being. The Next Big Thing is a very big thing; to own it would be to rule the world. But we made that impossible, Tony and I -- because we took what we had created and we gave it away.
In olden times -- that is, before 1993 -- Tony and I would have done the sensible thing and immediately sought patent protection for our work. But after reflecting on my own failures, I argued that the best of all possible outcomes might be had by giving our work away. It had worked for TCP/IP. It had worked for the Web. Why not follow in their footsteps, and hope for a similar success? The alternative -- fighting uphill to get a proprietary standard accepted as a basic protocol on the Internet -- seemed nearly impossible, and thoroughly against the spirit of the nascent Web.
So we wrote up our work at the request of Tim Berners-Lee, and presented it at the First International Conference on the World Wide Web. However primitive, it embodied all the concepts which would become central to VRML, and, in that spirit, we presented it -- hoping that others would find the work compelling and help us to create something more enduring. After that first success, I began to build a community to realize a shared vision. We got lots of help, in those first days, from WIRED magazine. Brian Behlendorf -- the technical brains behind the first version of HOTWIRED -- convinced his boss that WIRED should give back to the Web as it had been given; so WIRED gave us disk space and a mailing list that would become the ground zero of VRML community. From that list, we reached out to scour the world for the best ideas and the best technology to use in an industrial-strength VRML. Our clock was ticking; I wanted a completed specification by the next Web conference, just five months hence, in Chicago.

Lack of time forced us to be very practical: rather than creating something from whole cloth, we sought to identify and, if necessary, modify, an existing tool, something that had been around long enough to be mostly bug-free. We found a few candidates -- mostly academic projects -- but the one that clearly shone, Silicon Graphic's Open Inventor, was a product. You actually had to buy it. Even better, Open Inventor project leader Rikk Carey offered up the Open Inventor file format, with changes to accommodate the Web -- all in a nice, royalty-free package. What more could we ask for? Still, some objected that it hadn't been invented here, in the heart of the community; others protested that borrowing an existing product would inevitably profit the company that originally created it. I wasn't inclined to disagree. Sure, Silicon Graphics had an advantage in the creation of VRML browsers and tools, because they had spent the man-hours working through the technical nitty- gritty of VRML 1.0. But why should that matter? I saw it as a simple, and unavoidable, quid pro quo.
In the four months after we canonized Open Inventor as VRML 1.0, I heard very little from Silicon Graphics. Then -- just about two years ago -- I got a call from one of Silicon Graphics' marketing partners, who asked me to sign an non-disclosure agreement, so I could learn the details about the "launch" of VRML. This "launch" had nothing to do with the VRML community. We had no input in the process. It made business-page news all around the globe; meanwhile, the VRML community could only sit on the sidelines as Silicon Graphics and its hand-chosen partners basked in the media spectacle. Very few of those partners have remained involved in VRML. But it was "official": the VRML business had begun.

It's been my observation that any "open" process has three distinct phases; first, connection, when the community gathers; second, collection, when a collective mind generates ideas and implements them; last, correction - as the community reflects upon its results and feeds this back into the collective mind. By mid-Summer 1995, the need for a set of "Secret Chiefs" of VRML had become painfully apparent, so Rikk Carey and I organized the VRML Architecture Group (VAG), a group of engineers dedicated to amending the existing flaws in VRML. We had no mandate, asked no one for permission; we just went ahead and did it. All of our work would be available for review and comment by the VRML community, sufficient franchise to prevent an outrage.
After lengthy debate, we came up with some new directions for VRML; we fixed existing problems and added new capabilities. But through all of this a cry rose up from commercial interests -- most notably Netscape and Sun -- that unless the VAG expanded to include them, they might not pay any attention to its deliberations. "Fine," we said, "go your own way. Be proprietary. And die for lack of market share." We had made sure to include Microsoft in the original VAG. Neither Sun nor Netscape could make a standard stick without Microsoft's buy-in, but the Redmond Empire could do as it pleased.
Despite our best efforts, at the end of 1995 we saw the VRML universe explode. On December 4, Sun, Netscape and Silicon Graphics announced COSMO -- a proposal for a VRML/Java hybrid -- at a joint press conference. Three days later, Microsoft announced Active VRML, another approach with strong technical merits. Both products positioned themselves as the legitimate heir to VRML 1.0, but neither had been presented to the VRML community for approval, or even comment. Under the threat of a revolt from the VAG members who didn't cash SGI paychecks, Silicon Graphics submitted to an open referendum on the future of VRML.

FEED reader Peter Frey writes in, "All of us who are professionally involved in the growth of the 'net have prophesied and lamented the loss of the original spirit of the 'net which gave it the impetus to grow into the world wide phenomenon it is today. However, the reality is that Sun's over touted phrase: 'The Network is the computer' is a reality, and therefore the battle isn't over web browsers or standards: it's over survival."
Click here to post your responses in our Feedbag discussion area.
In a recent piece titled Software Makers Evangelize VRML, Ed McCracken, CEOof SGI offers a few words to bolster the VRML community's faith: "Once again people are asking questions about why do we need 3D. On the Web does [3D] mean something like cartoon-like avatars? Is it Neuromancer and Snow Crash or the Home Shopping Network? I know we can answer these questions about 3D network computing."

Stunning and atmospheric VRML enter- and edu-tainment can be found in SGI's VRML Excellence Award Winners. Caveat: you will need to download SGI's VRML browser/plug-in to look at the sites.

Find out more about the companies born of VRML, including ThinkFish: "Most 3D worlds today resemble the highly polished, sterile look of Toy Story. . . ThinkFish's approach to 3D allows designers to create spaces with a more artistic, rougher appearance. . . [like] a 3D field of sunflowers in the style of Van Gogh, through which users could navigate."	This time the electoral process resembled an American election; money and power defined the strategies of the candidates -- in addition to a few dirty tricks. Microsoft barely skirted anti-trust laws in an attempt to slow Silicon Graphics' development efforts on COSMO; Silicon Graphics and Netscape launched a PR air-strike designed to take the wind from Microsoft's sails. Every party to the referendum -- myself included -- displayed a lust for results that said more about our political aims than technical merits. Microsoft lost. As inconceivable as it seems, the world's largest software company went down, guns blazing. And Silicon Graphics -- who had picked up a lot friends along the way -- won an overwhelming victory in "Moving Worlds", the basis for COSMO. But the real lesson to be learned from the open "debacle" which gave us VRML 2.0 was the pressing need for some non- partisan organization in the VRML universe, a body without commercial interest, that could work to the common goals of the community. When the VRML 2.0 specification -- argued over and amended by a thousand developers -- went to press in August of last year, the need for a consortium to maintain coherence became apparent to everyone. Suddenly, Microsoft -- which had been humbled in browser wars and defeated in the Java skirmish -- stepped back up to the plate. We formed a tiny working group to define a consortium which could meet the needs of all; Netscape, Microsoft and Silicon Graphics each had major roles. The logic was simple: if we could satisfy the "Big Three," we'd have something that would satisfy everyone. Or so I thought. I wrote a charter for the VRML Consortium, one which did its best to balance the need for a stable specification against the need for innovation. Using the U.S. Constitution as a rough guide, I attempted to make no distinction between multi-billion dollar firms and the latest start-up. And I did my best to provide a way for anyone to contribute to the process - a meritocracy of ideas, given backbone by the Consortium. But VRML had become big business. Silicon Graphics stood to make millions a year from VRML tools and workstations; Netscape needed a strong VRML implementation to fight back against Microsoft's Explorer efforts; and Microsoft needed in at a fundamental level, so that they'd never suffer another humiliating defeat. So each of them looked for partners. Microsoft found one in Intervista -- the product of Tony Parisi's never-ending developmental efforts. Netscape wavered between standing on its own and cozying up to Silicon Graphics. Meanwhile, Silicon Graphics courted Apple Computer. Trying to get a read on Apple Computer is a lot like learning about quantum physics; you can never know Apple's position on a technology, and its direction, simultaneously. Apple had been dancing around the edges of VRML for some time but had burned the community badly by committing to deliver a specification for a binary format of VRML (which would cut download times significantly), then never delivering it. Further, it had privately informed developers that choosing VRML over Apple's proprietary QuickTime 3D technology would mean that Apple would be less than interested in supporting them in their projects. But eventually Apple -- crippled by the internal crises of 1996 -- came around to the inevitability of VRML. But they had some bridge-building to do, as they'd already angered most of the developer community. Other than arch-enemy Microsoft, only Silicon Graphics had managed to avoid Apple's to-and-fro. So, sometime in mid- November, the two companies inked a secret agreement to bring COSMO to the Macintosh. In return, Silicon Graphics would use QuickDraw 3D in COSMO. Using every path open to them, Silicon Graphics moved to raise Apple's prominence in the VRML community. On its own, Silicon Graphics already a had a great deal of say in VRML's future; allied with Apple, it might control the entire industry. So, in the first week of December, Silicon Graphics engineered a power-play, trying to get Apple onto the Board of Directors. Other consortium members -- lacking my detailed knowledge of Apple's attempts to delay or derail VRML -- greeted the idea positively. But I lost my temper, and said that Apple would only have a board seat over my steaming corpse. At first I stated this privately to Silicon Graphics. They contended that, after an "investigation," they had every assurance of Apple's commitment to VRML. This -- before any word of the Silicon Graphics/Apple deal had been made public -- seemed a little too slick for me, so I continued to protest. I wanted every company on the Board to have a history of dedication to VRML -- something Apple clearly lacked. I could smell something -- I couldn't tell you what, precisely, couldn't tell anyone who asked me -- but I knew this to be part of a larger strategy. At this very delicate moment of beginnings, a commercial interest sought to poison the centerpiece of the consortium. I was sure of that. The stakes certainly made it worth the attempt; whoever won here would control the next age of "virtual" computing, of interfaces so intuitive that they disappeared into the environment. I had only one path open to me: I had to go public with the whole affair, and hope that, by dragging it into the light, we'd all be able to get a clear picture of what was going on. So, on December 6th, I very publicly quit the VRML consortium, resigned from the VAG, and advised everyone involved to reexamine their own commitment to a process that seemed destined to be dominated by a single company. I didn't mince my words. This was my own work in jeopardy, my own community hanging in the balance. I refused to see it given up. It worked; Silicon Graphics backed down and the Board was restored to its earliest -- and most balanced -- form. I had won the battle. But I was on the outside now, looking in. Three months later, it's clear to see that the battle for cyberspace has been joined; the world has divided into two camps. On one side, Microsoft with its smaller partners; on the other, Silicon Graphics, Netscape and Apple. It's an uneasy truce, because each party has a direct economic stake in the failure or success of the other, a foolish competition around something that must remain open to all to succeed. No single company can dominate cyberspace, just as no single company can hope to dominate the Web. It takes an ecology -- a community -- to ensure success. Of course, we must be reasonable; in the late-capital commodification of all, when matter has completed its inevitable progress from Magna Mater to private property, there is no thing that is not owned - ideas, visions, dreams. Without that ownership, we are told, there would be no progress toward our common goals. Yet in the eighteen months VRML spent "below the radar" of corporations, we moved forward faster than ever we have since. A collective mind can do great things, even in the absence of a profit motive. It may indeed be impossible to turn the clock back to an age before the commerce of VRML. But I wonder if we have not condemned ourselves to eternal warfare. (February 14, 1997)
Have a look at Feed's Dialog on interface design from last September. In "Beyond the Desktop," Pesce and three other notables talk tech, new and old: "'Do you throw books away because you live in the real world? Or ignore the real world because you have books and newspapers?' It's almost ludicrous - but sympomatic of the continuous obsolescence of our technologies - to consider junking anything that has proven to be an effective interface."
Pesce's quest for the holy grail is only partially metaphorical. The wand-bearing wizard is right out of Narnia and he's got x-ray vision when it comes to finding digital treasures. Glimpse his booty at Hyperreal online, including other written works, lectures, and papers that chronicle the birth of VRML.

Click here to post your responses in our Feedbag discussion area.

Not everyone is convinced that 3D environments will provide quality entertainment. "The real problem is content, will always be content," says Andries van Dam, a computer science pioneer at Brown University. "What are we going to say to each other which is meaningful? Are we going to be reduced to the same banalities that we exchange at cocktail parties? We can do wonderfully well on the tech side, but the content is still a challenge." Read on for the downside of VRML.	Like all of the Silicon Valley myths, this story begins with two hackers hot on the trail of the Next Big Thing. It all seems obvious now: the Web that brought us text and images could bring us objects as well, a three-dimensional representation of the world. You could explore that new space, move through it; navigating the Internet would finally be more than just a metaphor. That was my starting point -- mine and Tony Parisi's. Three years ago this week, Tony and I followed a link in Mosaic and found -- to our delight -- that we looked into a 3D universe of objects. Tony had only one question: "What do we do with it?" This is the story of the birth of a standard, a bloody and painful meeting between the forces of commerce and community, forces that are at work anytime a new medium comes into being. The Next Big Thing is a very big thing; to own it would be to rule the world. But we made that impossible, Tony and I -- because we took what we had created and we gave it away. In olden times -- that is, before 1993 -- Tony and I would have done the sensible thing and immediately sought patent protection for our work. But after reflecting on my own failures, I argued that the best of all possible outcomes might be had by giving our work away. It had worked for TCP/IP. It had worked for the Web. Why not follow in their footsteps, and hope for a similar success? The alternative -- fighting uphill to get a proprietary standard accepted as a basic protocol on the Internet -- seemed nearly impossible, and thoroughly against the spirit of the nascent Web. So we wrote up our work at the request of Tim Berners-Lee, and presented it at the First International Conference on the World Wide Web. However primitive, it embodied all the concepts which would become central to VRML, and, in that spirit, we presented it -- hoping that others would find the work compelling and help us to create something more enduring. After that first success, I began to build a community to realize a shared vision. We got lots of help, in those first days, from WIRED magazine. Brian Behlendorf -- the technical brains behind the first version of HOTWIRED -- convinced his boss that WIRED should give back to the Web as it had been given; so WIRED gave us disk space and a mailing list that would become the ground zero of VRML community. From that list, we reached out to scour the world for the best ideas and the best technology to use in an industrial-strength VRML. Our clock was ticking; I wanted a completed specification by the next Web conference, just five months hence, in Chicago. Lack of time forced us to be very practical: rather than creating something from whole cloth, we sought to identify and, if necessary, modify, an existing tool, something that had been around long enough to be mostly bug-free. We found a few candidates -- mostly academic projects -- but the one that clearly shone, Silicon Graphic's Open Inventor, was a product. You actually had to buy it. Even better, Open Inventor project leader Rikk Carey offered up the Open Inventor file format, with changes to accommodate the Web -- all in a nice, royalty-free package. What more could we ask for? Still, some objected that it hadn't been invented here, in the heart of the community; others protested that borrowing an existing product would inevitably profit the company that originally created it. I wasn't inclined to disagree. Sure, Silicon Graphics had an advantage in the creation of VRML browsers and tools, because they had spent the man-hours working through the technical nitty- gritty of VRML 1.0. But why should that matter? I saw it as a simple, and unavoidable, quid pro quo. In the four months after we canonized Open Inventor as VRML 1.0, I heard very little from Silicon Graphics. Then -- just about two years ago -- I got a call from one of Silicon Graphics' marketing partners, who asked me to sign an non-disclosure agreement, so I could learn the details about the "launch" of VRML. This "launch" had nothing to do with the VRML community. We had no input in the process. It made business-page news all around the globe; meanwhile, the VRML community could only sit on the sidelines as Silicon Graphics and its hand-chosen partners basked in the media spectacle. Very few of those partners have remained involved in VRML. But it was "official": the VRML business had begun. It's been my observation that any "open" process has three distinct phases; first, connection, when the community gathers; second, collection, when a collective mind generates ideas and implements them; last, correction - as the community reflects upon its results and feeds this back into the collective mind. By mid-Summer 1995, the need for a set of "Secret Chiefs" of VRML had become painfully apparent, so Rikk Carey and I organized the VRML Architecture Group (VAG), a group of engineers dedicated to amending the existing flaws in VRML. We had no mandate, asked no one for permission; we just went ahead and did it. All of our work would be available for review and comment by the VRML community, sufficient franchise to prevent an outrage. After lengthy debate, we came up with some new directions for VRML; we fixed existing problems and added new capabilities. But through all of this a cry rose up from commercial interests -- most notably Netscape and Sun -- that unless the VAG expanded to include them, they might not pay any attention to its deliberations. "Fine," we said, "go your own way. Be proprietary. And die for lack of market share." We had made sure to include Microsoft in the original VAG. Neither Sun nor Netscape could make a standard stick without Microsoft's buy-in, but the Redmond Empire could do as it pleased. Despite our best efforts, at the end of 1995 we saw the VRML universe explode. On December 4, Sun, Netscape and Silicon Graphics announced COSMO -- a proposal for a VRML/Java hybrid -- at a joint press conference. Three days later, Microsoft announced Active VRML, another approach with strong technical merits. Both products positioned themselves as the legitimate heir to VRML 1.0, but neither had been presented to the VRML community for approval, or even comment. Under the threat of a revolt from the VAG members who didn't cash SGI paychecks, Silicon Graphics submitted to an open referendum on the future of VRML.
FEED reader Peter Frey writes in, "All of us who are professionally involved in the growth of the 'net have prophesied and lamented the loss of the original spirit of the 'net which gave it the impetus to grow into the world wide phenomenon it is today. However, the reality is that Sun's over touted phrase: 'The Network is the computer' is a reality, and therefore the battle isn't over web browsers or standards: it's over survival." Click here to post your responses in our Feedbag discussion area.
In a recent piece titled Software Makers Evangelize VRML, Ed McCracken, CEOof SGI offers a few words to bolster the VRML community's faith: "Once again people are asking questions about why do we need 3D. On the Web does [3D] mean something like cartoon-like avatars? Is it Neuromancer and Snow Crash or the Home Shopping Network? I know we can answer these questions about 3D network computing."
Stunning and atmospheric VRML enter- and edu-tainment can be found in SGI's VRML Excellence Award Winners. Caveat: you will need to download SGI's VRML browser/plug-in to look at the sites.