Virtuality and VRML: Software Studies After Manovich

Virtuality and VRML: Software Studies After Manovich


A call for (and example of) material studies of software from Matt Kirschenbaum, spurred by the Digital Arts and Culture conference, 2000.

In 1999 I was asked to join a panel on “virtuality” convened by the artist and media scholar Johanna Drucker at the international Digital Arts and Culture 2000 conference in Bergen, Norway. See for the conference program. Her instructions to us were to organize our ideas around “the ideology of the virtual” or “virtuality and ideology.” Aware that the same word (“virtuality”) had also recently been used by N. Katherine Hayles to indicate “the cultural perception that material objects are interpenetrated by information patterns,” I was intrigued by the opportunity to think about ideologies of the virtual from a materialist perspective. Hayles, N. Katherine. How We Became Posthuman: Virtual Bodies in Cybernetics, Literature, and Informatics. Chicago: University of Chicago Press, 1999. 13. What would it mean to reverse the poles of Hayles’s formulation and address my own critical conviction that digital objects are “interpenetrated” by material patterns (and circumstances)?

By digital objects I mean tangible hardware devices such as processors, VDT screens, and Palm Pilots, but also, and especially, intangible software objects such as source code, operating systems, interface elements, and data representations of all kinds. I use tangible and intangible to distinguish between hardware and software because both hardware and software are material entities. The fact that you can’t reach out and touch software (only the shrinkwrap) is incidental. (I like to call that the haptic fallacy.) Software is the product of white papers, engineering specs, marketing reports, conversations and collaborations, intuitive insights and professionalized expertise, venture capital (in other words, money), late nights (in other words, labor), Mountain Dew, and espresso. These are material circumstances that leave material traces - in corporate archives, in email folders, on whiteboards and legal pads, in countless iterations of alpha versions and beta versions and patches and upgrades, in focus groups and user communities, in expense accounts, in licensing agreements, in stock options and IPOs, in carpal tunnel surgeries, and in the [former] Bay Area real estate market (to name just a few).

At the time I was highly critical of the general lack of historical materialist studies of new media, and the more polemical portion of my remarks went something like the following:

New media studies, as a field, has not yet shown that it appreciates the importance of material history. Too often instead, there is a kind of romance of the digital that prevails, celebrating either the medium’s putative immateriality or its putative newness and uniqueness. If the devil is in the details then much of what has been published under the rubric of new media studies has been positively angelic. We have numerous books on virtual reality, but no accounts of the rise and fall of VRML, the all-but-defunct Virtual Reality Modeling Language. Why is it that three-dimensional graphics, a representational form well established in both the gaming and the scientific visualization communities, has yet to find a foothold on the Web? What does that say about Web as an electronic environment? These are questions that I believe are answerable, and I believe it is our responsibility to answer them, for they have direct bearing on what we think we know about digital art and culture today. But they are questions that can only be answered by acknowledging that digital objects are the product of material environments and that those environments have histories that are or ought to be recoverable.

That’s what I said in 1999, though of course there were already studies that did at least some of what I was calling for, Hayles’s among them. Two years later, however, Lev Manovich published The Language of New Media, certainly the best book I’ve seen on digital culture and aesthetics. [Manovich’s book is reviewed by Geniwate in ebr.] In it, Manovich includes, among much else, a call for a shift from media studies to something he calls software studies or software theory. Manovich, Lev. The Language of New Media. Cambridge: MIT Press, 2001. For Manovich’s remarks on software studies, see p. 48. What is software studies? Manovich doesn’t give us much more than the term itself, though his book clearly stands as an extended self-defining example. To me, software studies implies something very close to what I was trying to get at in Bergen, the idea that the deployment of critical terms like “virtuality” must be balanced by a commitment to meticulous documentary research to recover and stabilize the material traces of new media - a remembrance of things past, but also the pre-condition for another of Manovich’s imperatives, a “theory of the present.” What follows is my own brief experiment in software studies and theories of the present, a thumbnail narrative of the rise and fall of VRML that attempts to answer some of the questions I have raised above.

Our story begins July 2, 1998, when unsuspecting users looking for (Cosmo Software being SGI’s brand name for its emerging line of VRML products) would have encountered an error message telling them their browsers were unable to locate the server. Industry insiders knew that SGI was looking to divest itself from the entire Cosmo line, and that a deal to sell Cosmo Software to Sony had just fallen through. But to quote one observer at the time, “even the most experienced people on the Internet can not remember when a Fortune 500 company, with an enormous investment in its presence on the Web, has simply turned off a major Web site.” SGI would later restore the site, claiming that its disappearance was only coincidence, the result of an ill-timed technical glitch. Yet days later, in the wake of the failed Sony sale, SGI pulled the plug on Cosmo, halting all product development, transferring all Cosmo employees to other divisions of the company, and eventually licensing Cosmo products to a third party outfit named Platinum. Although VRML was and is an open standard, meaning that it is not beholden to any one company or platform, Silicon Graphics had been its biggest industry supporter; with SGI out of the picture, VRML would loose its best browsing software and its only dedicated authoring tool.

VRML had had a checkered history up to this point. The first version of the standard was introduced in May 1994, relatively early in the Web’s overall development. Originally engineered by Mark Pesce, Gavin Bell, and Tony Parisi, VRML 1.0 was based on a Silicon Graphics 3D object format known as Open Inventor. Although the VRML spec itself was non-proprietary, this fact explains SGI’s early and central involvement in the VRML community. VRML allowed for the modeling and rendering of simple 3D objects and scenes that could be displayed by any Web browser with the appropriate plug-in. Unlike Apple’s QuickTimeVR, which is essentially a vehicle for displaying 360° panoramas, VRML brought the capacity for true 3D rendering to the Web. Objects were fully defined in a three dimensional coordinate system, and users could navigate between, behind, and around them, or else explore the infinite computational void in which they were situated.

Skeptics have noted that the VRML community from the beginning was infatuated with the most vulgar trappings of cyberpunk science fiction. The standard was actively promoted as the mechanism that would quickly transform the Web into an authentic Gibsonian information landscape. This prejudice was written into the VRML spec at the most literal level: VRML files, for example, were to be called “worlds,” with a.wrl suffix. But by late 1995, the VRML community, viewed by many as the Next Big Thing, had become heavily politicized, with a loose alliance comprising SGI, Netscape, and Sun on the one hand, and Microsoft on the other. The VRML Consortium, dedicated to keeping the standard open-sourced and community-based, was caught off guard. Matters came to a head when SGI and Microsoft each proposed rival specifications for VRML 2.0, the successor to the original roughly defined standard. SGI bypassed the VRML Consortium, and their spec, dubbed “Moving Worlds” (emphasizing animation and interactive scripting) was launched with considerable public anticipation. “Moving Worlds” quickly became the basis for the evolving Cosmo line of products. This was 1997 and early 1998, the heyday of VRML development. Cosmo was generally thought to be the best browser, though Microsoft’s Worldview, built by a company named InterVista (headed up by Toni Parisi, one of the original VRML triumvirate) was set to be included as an integral component of Windows 98. Developers, meanwhile, had started to realize that rather than virtual worlds and Gibsonian cyberspaces (for which there seemed to be little demand in the commercial sectors of the Web) the true future of VRML lay in embedded 3D animation, including banner advertising and the like. For a time, the VRML community enjoyed a weekly cartoon serial starring a character dubbed “Floops.” Such was the situation in the summer of 1998 when SGI, increasingly in dire straits financially, made the decision to sell off all of its non-essential product lines - including by this point Cosmo.

In the wake of the failed Sony deal, Cosmo was purchased by Platinum, a large but rather lackluster company specializing in corporate enterprise software. People at Platinum had begun thinking about something they called Process and Information Visualization (business visualization, or “biz viz” for short) and saw in VRML a way of revolutionizing the next generation of corporate middleware. They had also, about a month beforehand, acquired Intervista from Tony Parisi, which, with the addition of Cosmo, effectively gave them control over the only two VRML platforms in common use. Platinum planned to take the best features of both and release an integrated browser and developer’s tool tailored for corporate data visualization, product visualization, and what was nebulously called process visualization; all compatible with standard Microsoft Office packages like PowerPoint, Word, and Excel. So what went wrong? Platinum themselves fell on increasingly hard times financially, and eventually sold the Cosmo line to industry giant Computer Associates, where it remains mothballed to this day. The VRML community, meanwhile, has moved on, and reincarnated itself as the Web 3D Consortium, which is currently work on a standard known as X3D, an XML-compliant schema for describing 3D objects and scenes. See

So that’s what happened to VRML. But the question remains: why have true 3D representational technologies yet to prove broadly viable on the Web? This is a question that many have asked, and one line of thinking is typified by the following remarks posted to the VRML developer’s list:

Take a look at HTML [and Java]. None of [their] problems are stopping people from developing Web sites. Why? Because they perceive a need for the kind of content that they can create with these technologies. The perceived need is so strong that they workaround all the problems. We need content that demonstrates convincingly why 3D and VRML is useful and necessary to the masses.

Content may be king, as they like to say on Bloomberg and MSNBC, but I think part of what my brief narrative of VRML reveals is the extent to which this line of thinking slips too easily into a kind of false consciousness. Apple’s QuickTimeVR, for example, has built a dedicated following, and the true heir to the kind of dynamic animation once promised by VRML is to be found in Macromedia’s Flash product. Flash is currently being used for both games and visualization on the Web, but especially for splash screens and the kind of animated vignettes and shorts anticipated by the Floops character. Tracking the extent to which the rise of Flash parallels the fall of VRML is an exercise for a longer essay, but one of the reasons why Flash is flourishing is that it has overcome many of the problems that afflicted VRML, notably browser distribution (the Flash player is a standard installation option for both Netscape and Internet Explorer) and cross-platform compatibility. This immediately restores us to the eminently material world of data standards, licensing and distribution agreements, marketing strategies, and so forth. The salient question is not whether one can produce “better” content with VRML or with Flash, but rather the extent to which the kind of content we create for environments like the Web is determined by various social histories, histories that are often corporate, but always situated within absolute zones of material and ideological circumstance.

What is software studies then? Software studies is what media theory becomes after the bubble bursts. Software studies is whiteboards and white papers, business plans and IPOs and penny-stocks. Software studies is PowerPoint vaporware and proofs of concept binaries locked in time-stamped limbo on a server where all the user accounts but root have been disabled and the domain name is eighteen months expired. Software studies is, or can be, the work of fashioning documentary methods for recognizing and recovering digital histories, and the cultivation of the critical discipline to parse those histories against the material matrix of the present. Software studies is understanding that digital objects are sometimes lost, yes, but mostly, and more often, just forgotten. Software studies is about adding more memory.