Bob Jacobson is one of the first contacts I made through the Internet. When I first got onto the Net back at the end of January 1994, I posted a message relating to the avaiiability of this newsletter on the USENET newsgroup "comp.infosystems.gis" or as a mailing list, "GIS-L." I was "flamed" by a university graduate student for my "blatant commercialism," but it was Bob who came to my defense by posting a follow up message supporting my announcement. We met in person shortly thereafter, and I learned more about what he is attempting to do with his firm in terms of research and development activities.

It's been nearly a year-and-a-half since I wrote about the relationship between GIS and virtual reality (VR). The article which appeared in the September 1993 edition of this newsletter was inspired by the work of one of Bob's former students at the University of Washington. As I mentioned in the previous article, the relationship between the two is such where VR is used in conjunction with GIS to enhance spatial visualization abilities of the user. Realizing this fundamental concept, I've intended to present a follow up to the September 1993 edition all along and what better source of information could I find than one who is helping to define and shape the VR discipline! My sincere thanks go out to Bob for putting together this guest column.

Facing Up to GIS Interfaces

Real-time multidimensional interfaces for GIS applications are imminent. In part this is due to the evolution of technology, but more important, a new audience of users is ready to demand it.

GIS users are growing in number and diversifying forcing radical changes in GIS technology. The interface is a key target of software developers. Map outputs are no longer adequate for a population of users that is overwhelmingly map illiterate. New interfaces are appearing in laboratories and at trade shows: charts and graphs, video and acoustical images, and integrated multimedia. Although formal studies of human-GIS interactions are few and far between, anecdotal evidence suggests that the new interfaces are enjoying wide public acceptance. Moreover, new products being put on the market by manufacturers (including market leaders ESRI and Intergraph) and public domain software increasingly feature non-map interfaces. This suggests that private market research and user demand are compelling developers to accelerate interface innovation.

While equally dramatic developments are taking place among other GIS componentsÑspatial databases and spatial database engines will soon be all the rageÑthe interface remains GIS's "public face." Adepts may differentiate among GIS's based on the systems' operational capabilities, but for the vastly larger number of non-professional users, GIS's are best known by their interface characteristics. ESRI's ArcView product, for example, essentially offers a new way of using a mature product, ARC/INFO; ArcView is all interface. How crucial the interface can be to the success of a product, GIS's included, is demonstrated by the recurring battles over "look and feel" between Lotus and Borland, Microsoft and Apple Computer, and now even among designers of Web pages for the WWW (World Wide Web).

The interface, being so important, in turn determines how other elements of the GIS must function. For example, video images may be central to a particular application (for example, surveys to monitor the condition of highways). Video servers must be accessed. The GIS must be able to link geospatial data with video images and do so quickly, so that the servers can provide the massive amounts of data that video comprises, fast enough to sustain the illusion of movement. As the user prowls the phenomena under study, the GIS must learn his or her virtual location with relation to these phenomena and call forth the appropriate data. The interface becomes a key part of the GIS. Conventional GIS as recently as 1990 did not offer a video interface. Today such interfaces are common, if pricey.

Beyond Conventional Solutions

The value of these more user-friendly interfaces, however, is diminished by the two-dimensionality of computer and video presentations, which creates a kind of tunnel vision, a frustrating "window" effect. In a movie theater, the large screen at least subtends much of the viewer's peripheral vision, increasing the user's involvement. Computer images that are relaffvely small and sounds that emanate from only one direction, however, even if in stereo, are inherently less compelling and often difficult to share. Moreover, it is nearly impossible to interact with video and similar linear media without interrupting the flow of experience while the technology (for example, a CD-ROM reader) catches up with the data search. Even so-called "2-1/2D" images, computer images that display apparent dimensionality, can usually only be rotated on the screen: the user cannot easily look around inside the model.

There are two problems associated with this situation; both affect user satisfaction. One is the method by which a GIS displays information, and gathers user instructions, for future action. The other is the structure of the data being processed by the GIS, to be displayed to the user. On the one hand, lifting images off the screen, especially in a shared immersive environment, is extremely powerful as a way of increasingly the salience of data. On the other hand, data that can be converted into multidimensional imagery is likely to be better understood by human beings, whose sensory apparatus is spatially organized.

The use of virtual worlds to achieve these goals is preferred by my firm, Worldesign Inc., and by our clients. We are determined to create an interface based on virtual worlds that shows off GIS's to their greatest advantage as spatial data processors. This strategy is controversial and certainly not assured of success, but it is built on powerful theory. Given the pronounced failings of conventional interfaces, it is worthwhile to pursue the virtual world solution; it holds forth promise.

Why Virtual Worlds are Important

What is a "virtual world," and what does it mean in the context of GIS? Briefly, the term "virtual world" is a scientific term of art with two distinct but related meanings, in environmental psychology and computer science respectively. The virtual world's first meaning is as the internal mental map that a person or a group of persons construct

and refer to as a portrait of the physical and social worlds they inhabit. Donald Schon, the well-known organizational theorist, is credited with this definition of the virtual world. The virtual world's second meaning is as the model of relevant phenomena and their relationships maintained in a computer memory. The theory of virtual worlds, as we apply it at Worldesign, holds that the more complete and successful is the interface that connects a person or persons with the computer, the closer the correspondence will be between these virtual worlds. The degree and quality of this integration is the measurement of a successful virtual world.

Currently, there are six major commercial virtual "worldbuilding" and object modeling software packages (from Sense8, Silicon Graphics, Division, Autodesk, and the UK's Virtuality Group and Superscape) and several challengers (including software from Computer Explorations, Paradigm Simulations, MultiGen, and VREAM). Public domain software includes the University of Alberta's MR Library and 386REND and its derivatives, created at the University of Waterloo. The platforms on which these programs operate run the gamut from high-end SGI PowerChallenge supercomputers to modest 386 PC's; generally, the heavier the metal, the better the resolution, update rate, and tracking in a virtual world. This year, however, several new players are entering the field with new technology, including Hewlett-Packard and Sun Microsystems, and a raft of accelerator chips and boards that promise to turn modest Pentium and P6-equivalent PCs into extremely powerful virtual world platforms. Equally stunning has been the development of software that permits the sharing of virtual worlds over the Internet, albeit still with limited speed and only on computer screens, for now. Building virtual worlds, however, remains a labor-intensive process, but this too may soon change (see below).

Joining Virtual Worlds to GIS: Worldesign's Solution

Virtual worlds have a vital role to play in GIS. For most analyses, what is critical is that the analyst and the individuals or organization who base decisions on the analysis fully appreciate the spatial factors at work in what is aptly called the "decision space." A multidimensional, immersive, "inhabitable" virtual world definitely conveys the sense of space, both concrete and abstract, as no other interface can. Military simulators are used for just this purpose. The data they process is generally highly constrained, either by type or geography, so that the database associated with each mission remains manageable. As a result, simulation technology, while able to generate compelling images, usually do not support interfaces with enough flexibility to fit the many varieties of problems addressed and data processed by GIS's. Ñ

The database problem has proven equally vexing for virtual worldbuilders, but the development of spatial databases and fast spatial database engines promises to liberate GIS-oriented worldbuilders from arbitrary technology limits. The new "SDBE" announced by Jack Dangermond at GIS'95, in Vancouver, BC, in April portends significant changes in the future of GIS. For the moment, however, current GIS offer an uncomfortably tight fit to worldbuilders, who generally find the map-building output from contemporary GIS's unsuitable for multidimensional worldbuilding.

One solution is to craft an analytical engine whose output can be configured to generate the data objects necessary to define and populate a virtual world. Tellus, the software subsidiary of Puget Power (Puget Sound Power and Light Co.), sells a number of GIS-like modules for the planning and management of electric power networks. One of these, the AEPD^TM ("Automated Electric Plat Design") software, has been used by Worldesign as a foundation for AEPD-3D^TM, a software engine that automatically converts the output from AEPD into a true three-dimensional virtual world. The AEPD via ARC/INFO generates a two-dimensional schematic map of an optimized electrical network; the AEPD-3D converts it into above-ground and underground 3D models of the subdivision the network serves. Using the above-ground view, non-technicians (like real estate developers, city planners, and future home owners) can see and comment upon the location of transformers, streetlights, and similar components. The underground view is ideal for other infrastructure providers, like telephone companies, who prefer to follow utility-dug trenches and nnust configure their networks accordingly. (AEPD-3D was extensively discussed in the cover story for GIS World, Deeember 1994).

In the near future, versions of Worldesign's software will find application in related fields like AEC (Architecture/Engineering/Construction), facilities management, and real estate sales. Ultimately, it may find general application as a tool for realization of spatial phenomena.

Software Developments in Other Quarters

Worldesign is not the only organization working to unite virtual worlds and GIS software, though it is (to the author's knowledge) the only commercial firm to pursue this goal. Others working on the problem include Larry Hodges and Andreas Dieberger (a visiting researcher) at Georgia Tech's Visual Graphics Lab, Ron Li at the University of Calgary, Jonathan Raper at Birkbeck College in the UK, Greg Turner at the Army Research Laboratory, and Antonio Camara's exciting group at the New University of Lisbon. The GIS manufacturers, even the "innovative" ones, have lagged badly in this area, but Kurt Buehler of the Open GIS Foundation continues to advocate serious attention to virtual worlds as a necessary component of future open GIS's. In 1994, Jack Dangermond of ESRI, in a small way, helped to sponsor Worldesign's virtual worlds display at A/E/C SYSTEMS '94; perhaps, in 1995, he will direct the leading GIS manufacturer to take a more active role.

Displaying GlS-Generated Virtual Worlds

As noted earlier, worldbuilding in the GIS environment is one of two problems. The other remains how to best display the output of virtual worlds-capable GIS's. For most serious applications of virtual worlds, in all fields, the head-mounted display (HMD), or "goggles," has become passe. Many laboratories and users are content to work on the computer screen, dealing with the same limitations faced by all interface designers. Others, like Worldesign, prefer to work in a true immersive environment. In 1993 and 1994, we cobbled together the "Virtual Environment Theater", or VET^TM, which proved a surprisingly durable and attractive product. The VET is a three-screen booth with rear projectors and 3D sound capable of surrounding from four to 20 individuals with sights and sounds that can be navigated by a "guide." In 1995, at A/E/C SYSTEMS'95, we are experimenting with the VisionDome7M, a turokey immersive environment manufactured by Alternate Realities Corporation in Research Triangle Park, NC. In it we will be showing the AEPD-3D and a popular world, the Seattle waterfront's future development, that was rated "best of show" at last year's A/E/C SYSTEMS show.

Another immersive environment is the CAVE, or Computer AudioVisual Environment), developed by the Electronic Visualization Lab the University of Illinois-Chicago for the National Center for Supercomputer Applications. All three exisUng CAVE s (which cost a cool $1 million each)Ñat the NCSA, Argonne National Laboratory, and ARPA, in Washington, DCÑwill be networked together this spring to create "televirtual" experiences that can be shared simultaneously. It will be interesting to see if any GIS applicaffons were submitted and qualified for testing in this environment.

Without a doubt, although HMD's may improve over time and offer stiff competiffon, for the moment, shared immersive environments are the future of virtual worlds: in them, communicaffon of experience and sharing of ideas and knowledge is much enhanced.

For More Informabon...

Where can GIS professionals go to learn more about virtual worlds? The best place for news, comments, and fabulous archives is the USENET newsgroup, "sci.virtualworlds", maintained by Toni Emerson, "cybrarian" for the Human Interface Technology Lab (HIT Lab) at the University of Washington. (There are similar conferences and forums on America Online, BIX, CompuServe, eWorld, and GEnie.) Toni also maintains the HIT Lab Web site, http://www.hitl.washington.edu/.

Another important source of informaffon is the industry journal of record, VR World, published bimonthly by Mecklermedia Inc.; back issues and some current arffcles are available on the WWW, at http:/lwww.mecklerweb.com/. The two commercial industry monthlies, both of excellent quality, are CyberEdge Journal in the U.S. and VR News in the UK. VRASP, an association of "students and professionals," publishes Pix-elations!, an occasional journal showing up on many bookstore shelves. The technical journal of record, Presence: The Journal of Virtual Environments and Teleoperators, is published bimonthly by the MIT Press. Conferences are almost too many to identifyÑand certainly too many to attendÑbut among the best are the two industry-oriented Mecklermedia VR Expo's held in San Jose, CA (May), and Boston (December) each year. Others include the European VR Conference, hosted by the Fraunhofer Institute in Stuttgart earlier this year; the VRAIS, the IEEE's technical conference, held at the University of North Carolina last March; the Singapore VR conference hosted by the Institute of Systems Science at the National University in the summer; and a new conference, the International Conference on Spatial Multimedia and Virtual Reality, which will take place for the first time at the New University of Lisbon in October 1995.

About the Author

Bob Jacobson is the president and CEO of Worldesign Inc., a Seattle-based virtual worlds design studio and software development house he founded in 1992. He formerly was a co-founder and associate director of the HIT Lab at the University of Washington. Thanks to Brian Matuschak for the inspiration and occasional necessary nudge to get this article written.