• In 2007, Second Life announced they would release their complex (and for newbies, difficult-to-use) VW viewer software to the open source community for modification and customization. As CEO Philip Rosedale says, "this extends the control Residents can have over the Second Life experience and allows a worldwide community to examine, validate and improve the software’s sophistication and capabilities.”" The Second Life platform continues to accelerate in membership. When we first began tracking it, accounts doubled from 160,000 to 330,000 accounts in four months (March to July 2006). Secondary to massive recent media exposure, the last three doublings have occurred an average of every two months, to 2.5 million accounts by Jan 2007. Over US $1M in transactions occur daily, on average, "in world."

• Web 2.0 (Participatory Web) technologies, led by innovative social networks, browsers, and search platforms, are accelerating the use of 3D and other rich media. The Participatory Web is tools and platforms that empower the user to tag, blog, comment, modify, augment, select from, rank, and talk back to the contributions of other users and the world community. Reputation-based public wikis, like Wikipedia, are pioneering examples of participatory web technology. Open APIs for tagging the web and tying it to the world, like Google Maps, are another. Rich media-enabled social networking sites like MySpace are another. Another is the open source Flock web browser, which encourages RSS aggregation, automated blog posting, photo sharing, gathering and indexing of web searches, and other participatory technologies. In Japan, companies like GaiaX have built social networking websites that allow their users, as one of many community options, to invite each other to online games and virtual worlds. This makes the social network the hub and the virtual worlds the occasional immersive experience [16]. Today's browsers are just beginning to manage 3D web capabilities (3D graphics, games, and video). Opera 9, for example, includes "widgets" that make it easy for users to organize their online games. The ability to easily incorporate YouTube and other video in leading social networking sites has really improved the stickiness of online community. Nevertheless, there is much to be done. We are very early in collaborative productivity software, like Writely (Google's online word processor). We don't have robust data interchange, rich annotation (video, etc.), or conversational search. We don't have good security, privacy, identity, or reputation. We don't have worldwide ultra-broadband or wireless connectivity, which greatly limits efficiency and scope of the collaboration space. Within 3D spaces, we don't have easy access to professional digital modelling tools, or grid computing for data rich simulations. There is a lot to be done, this is a very incomplete list.

• Head mounted displays (HMDs) and Spatially Immersive Displays (SIDs) like CAVE for virtual reality exist, but both technologies today are used only in niche markets. HMDs are likely to remain niche applications for the forseeable future (see Predictions, Tech and Science). There are many small HMD makers. eMagin makes a head-mounted, head-tracking 3D Visor with a 600 x 800p OLED display for immersive gaming for $600. Sensics uses the eMagin displays to make a very expensive panoramic HMD VR system with 2200 x 1200p per eye for military customers.

• In 2005, eBay's Internet Games category, hosted $30 million in trade for goods (virtual items and currency) that only exist in synthetic worlds [1]. Some (not all) game providers have since banned virtual asset and currency sales, driving much of this traffic to third party sites.

• Virtual tourism is in its infancy. Beginning with interactive CDs in the 1980s, virtual tourism is slowly gaining interest. Communities from Geoplace to the Virtual Terrain Project exist to promote tools for constructing the real world in interactive, 3D digital form. The attractiveness of virtual tourism systems seems today to be a complex function of hardware (speed and resolution), design (interface intutiveness), and software (story appeal and usefulness in connection to "real world" activities). As the market develops, software should increasingly become the key differentiator among competing VT systems, as we have seen with video games and online worlds. In coming years, when we have significantly faster computers, and can rapidly tour "interactive Los Angeles" or "interactive Yosemite" and sample micronarratives before deciding which of the many possible experiences we will take in a given day, and when this platform is integrated into tomorrow's browsers and today's passive, narrative-driven experiences like LA City View or the Travel Channel, virtual tourism is likely to be a very compelling activity.

• Dramatic rise of social networks and social virtual worlds. While there have been many false starts and fadeouts (Tribe, Ryze, Yahoo! 360, etc.), today's leading 2D social networks like MySpace (94 million users), Xanga (27 million users), Bebo (22 million users) and Orkut (16 million users), etc., have experienced sustained explosive growth in recent years. Most focus on the youth demographic, rapid adopters of new online behavior. MySpace is now the fourth most popular English language website, according to Alexa Internet, adding an astonishing 500,000 new users a week (Jul 2006). Automation allows these users to be supported by a company of only 500 employees. MySpace supports streaming audio and internet video, giving it limited 3D functionality and making it very popular with independent musicians and filmmakers. A number of these communities, like Cyworld (18 million users), Habbo Hotel (7 million users) and Playdo (300,000 users) also support avatars and the navigation of 2D social space, making them true "social worlds." Finally, a few social worlds, like Second Life (340,000 users) and There, offer a fully 3D immersive experience, but have the slowest acquisition rates, as the technology overhead and user effort required for 3D worlds remain higher at present.

• The state of online and virtual worlds time statistics today is poor. The BLS's American Time Use Survey (ATUS) is a major new undertaking that will give us insights into actual time use changes from year to year in American workplaces and households. The first official 2005 dataset has not yet been released, but early datasets have been analyzed in the first ATUS conference. Various non-ATUS studies have estimated that Americans spend roughly 35 hours per week either at school or at their jobs, and another 35 hours a week in leisure activities. Of this latter 35 they spend roughly 10 hours watching television. In their small annual phone survey, Harris Interactive estimated that the average American spent 9 hours online in 2006, up from 7 in 1999. Based on the movement of reading from physical to virtual space, a 1997 OC&C Strategy report estimated that by 2010 people would spend more time online than reading books, magazines, and newspapers combined. The 2006 ESA Game Player Data page claims that 50% of all Americans play video games, and that the average male game player plays 7.6 hours, and adult female game player plays 7.4 hours/week. They also claim that 44% of the most frequent video game players play online games. Other studies estimate that youth "heavy gamers", the top 25% of youth gamers, mostly male, age 14-17 years of age, spend on the order of 20-30 hours/week playing games. ESA claims that women represent about 38% of the total game playing population. In Korea, where broadband penetration is greatest, a 2000 study estimated that 80% of all youth (8-24 years) played games online at least once a month.

• The rising importance of “Network Neutrality” Policies. Lesiglation and activism that guarantees democratic access to the network. In August 2005, the FCC adopted the following guidelines regarding network neutrality:
  1. Consumers are entitled to access the lawful Internet content of their choice;
  2. Consumers are entitled to run applications and services of their choice, subject to the needs of law enforcement;
  3. Consumers are entitled to connect their choice of legal devices that do not harm the network; and
  4. Consumers are entitled to competition among network providers, application and service providers, and content providers.
Regulators have increasingly shown willingness to require broadband providers (cable, DSL, wireless) to eliminate competitive restrictions on network usage (e.g., ensuring the consumer has freedom to run free VoIP programs like Skype over telco-run broadband infrastructure, to set up Wi-Fi extensions on home networks, etc.). These emerging “network neutrality” policies bode well for U.S. consumer rights to run complex 3D applications in coming years. Even if such applications compete with existing provider offerings, are bandwidth intensive and force network upgrade, network providers have few options to restrict network usage.

• Personal cartography is emerging. People with time and patience can create annotations of their local spaces and journeys using tools like Google Maps [20], and there is a mash-up startup that revolves around people telling their personal travel stories in sharable cartographic journals. Increasingly these will maps will also have 3D (video, 3D object, and virtual world) components.

• Serious games and training simulations have been used for decades and are seeing increasing use by the military and other large organizations. Beginning with the Serious Games Initiative in 2000, the serious games community (game-based education and training occurring outside the traditional primary and secondary educational software markets) has grown steadily. SimuLearn's Virtual Leader, a corporate workforce training tool used in a few Fortune 1000 companies and developed with the budget of a normal video game, won awards in 2003. America's Army, a free first-person shooter owned and distributed by the U.S. government, has been a very effective, and controversial, global public relations and recruiting tool. Summit quote: "Training time has gone down from the amount of time new recruits have spent playing first person shooters."

• Learning/experience curve (Henderson's law). This observation, first quantified for aircraft production in 1936, and later generalized by Bruce Henderson of the Boston Consulting Group in the 1960's, notes that the time and cost to perform a broad range of technological tasks decreases exponentially in repeated trials. More generally, with each doubling in total production, unit production cost for a given level of performance or capacity falls by a predictable and generally constant percentage. While each particular technology paradigm will improve on a decaying exponential curve approaching a fixed performance limit, ongoing important problems usually elicit periodic technology shifts, allowing average capacity doubling times to remain relatively stable for long periods of time. Most classes of technology problems show experience curve price/performance improvements from 10 to 30% with each doubling of production. Most technologies at the human scale, such as transportation, eventually saturate, though they may proceed through many decades of exponential improvement, through a range of paradigms, before they do so. By comparison, technologies at the nanoscale may in theory continue their exponentiation past the limits of human systems, as they are driven by the discovery and implementation of the unique physics of the nanocosm, not human creativity, and as informational nanotechnologies become increasingly autonomous (human-independent) with time. Experience curves for ICT and nanotechnology argue that many of the computing and communication tools available only to the most powerful corporations today will be affordable to the poorest global citizens in coming years.

• Polygon count growth, motion control point growth, and the Reality Threshold (Smith). Polygon count generated by leading video game hardware doubles roughly every 2 years. Alvy Ray Smith of Microsoft/Pixar has estimated that the "reality threshold" (simulations indistinguishable from ordinary human vision) is 80 million polygons per frame, and on the order of a million motion control points on the objects within our field of view [21]. Microsoft's Xbox had a peak capacity of 2.1 million polygons per frame. Xbox 360 peaks at 8.3 million polygons (triangles) per frame (the PS3 will have only half this, yet another reason it may be a big setback for Sony). The faces of the best digital actors in Toy Story (1996) had over 100 motion control points. In Toy Story 2 (1999) they had over 1,000 motion control points. If these trends continue, the reality threshold may as close as eight years away, in 2014. While the actual transition might take twice this time (2022), it is clear that within a relatively short time our best simulated realities will be indistinguishable from physical reality. This is a development with global implications, as it promises to eliminate the current sensory opportunity cost of spending extended time in virtual environments. After this point, we may consider our best 3D online spaces to be "hyperreal" (environments with simulation capacity, and eventually entertainment, education and productivity capacity, that exceeds reality).

• Software algorithmic efficiency and simulation/representation capacity growth (Kurzweil, Ebrahimi). Algorithmic efficiency is a vast and technical topic that is dependent on automation, representation schemes, processing capacity growth, and experience curves, among other factors. Even with "software bloat," general software efficiency for many classes of problem has been quoted by Ray Kurzweil and others as doubling in price/performance efficiency every six years [35]. While still a dramatic growth rate, the doubling period is nevertheless three to four times slower than Moore's law. As Touradj Ebrahimi notes [36], algorithmic efficiency in video compression has doubled every five years for the last fifteen years. H.263 video encoding (1995) is twice as good as H.261 (1990), and H.264/MPEG-4 v10 (2003) is twice as good as MPEG-4 v1 (1998). Further developments will show if these growth rates continue to hold, but as we continually learn new ways to represent, locally store, and generate data, these growth constants seem a reasonable expectation for many classes of visual, sensory, and system simulation. Compression in portable audio (audio simulation/re-representation) capacity has grown at a roughly similar rate. First available in the Diamond Rio PMP300 in 1998, CD-quality mp3's (128kbps MPEG-1 layer 3), had a 10X compression over files in the Pulse Code Modulation (PCM) based compact disks that played on Sony's first portable CD player, the Discman D-50, in 1984. MPEG-4 AAC audio compression, available on today's iPod, offers modest additional compression over mp3. As Barry Vercoe of MIT notes, future audio standards are evolving toward a postscript-like, "structured audio" description language, which will allow audio files another 10X compression over mp3. Today's structured audio is not yet CD quality, but if the historical growth rate continues, we may see such a format used in portable players in 2012. In combination with storage growth trends, today's 60 GB, 15,000 song iPod will be capable of storing 2.4 million songs in 2014, more than the entire current song catalogs of today's leading online sites (iTunes Music Store, URGE, etc.). With regard to system simulation, virtual machines like VMWare, and such tools as software defined radio are examples of the representational capacity in software of a large and ever growing class of hardware systems. Such constant performance advances are strong evidence that not only hardware, but software will continue its accelerating ability to deliver simulated reality.

• Digital storage capacity growth (Kryder's Law) and the Lifelog. Density of information on digital storage devices has been doubling every 23 months, on average, since 1956. This observation is named after Mark Kryder, a storage research pioneer. Rapidly increasing storage capacity has enabled successive thresholds in database complexity, allowing the mass storage and manipulation of digital pictures, audio, video, HD video, etc. One development being enabled by this is the lifelog, a storage system that allows persistent global access, retrieval, and potential sharing of all past information generated by an individual user's life experience, in various data categories (email, websites visited, GPS coordinates visited, audio, video, etc.). The mass-accessible lifelog threshold for email was crossed in 2004 with the advent of Google Gmail, which began by offering 1 gigabyte of free email storage to users. Consistent with Kryder's law, Gmail adds storage behind the scenes daily. amd as of June 2006 offers 2.7 GB to their users, a growth rate well in excess of many user's email needs, including attachments. Flock, a Web2.0 browser, is developing a lifelog of all previous websites visited, and using this as a basis for context-sensitive internet search queries initiated by the user, displayed preferentially at the top of the list in the manner of Google Desktop. The No. 2 telco in Japan, KDDI Corp, has developed the Lifelog Pod, software that keeps track of every user action (photos, searches, MP3 listens, software runs) made through a cellphone or computer, extending one's searchable memory into even the minor details of their online history [64]. SLStats is a log program, activated by a watch your avatar wears in Second Life, that records simple statistics on the user's in-game activity. Passively Multiplayer is a research idea for more broadly recording and sharing user data generated in virtual worlds. Having one's life experiences auto-archived, persistently available, and only a voice query away is a threshold with major implications, including both personal empowerment and the potential for abuse of privacy, if accessible to others. We will need new social conventions for consented recording of public experiences, but we see nothing that would stop large fractions of the global population, particularly youth, from running lifelogs for personal and trusted network use as computing technology improves.

• Wireless bandwidth growth (Cooper's law). Martin Cooper observed that the spectrum efficiency of radio communication (both voice and data) has doubled every two and a half years, over 104 years, since radio waves were first used for communication. Frequency, spatial, and code division multiplexing, modulation techniques, mesh networks, and now smart antennas are examples of technologies that have kept wireless bandwidth on this growth path. Today, CDMA EVDO 1X is available from Verizon Wireless and Sprint in the U.S. for $60/month, delivering peak cellular modem speeds of 2.4 megabits/second in just 1.25 Mhz of spectrum. This presently allows rapid web browsing, very light streaming video and audio to the mobile user, and dedicated VoIP. EV-DO Revision A, allowing 3.1 megabits per second [24], are scheduled for commercial release in late 2006 [25]. In late 2007, Qualcomm will release CDMA EV-DO Revision B data modem and cellphone chips to card makers, which will provide up to 14.7 megabits per second on downlink. When these emerge in 2008 they will enable such mobile features as television, and internet browsing while making VoIP calls, and the enough extra bandwidth to support location-based streaming radio in the car (a huge new market, see: Predictions: Business and Econ) [26].

• Flat panel display (FPD) growth (Nishimura's, Kitihara's and Odawara's laws). From the U.S. Display Consortium (USDC) website: "Nishimura's Law, states that the size of [the glass] substrate used [and thus the display size available at the same cost] grows by a factor of 1.8 every 3 years, i.e., it doubles every 3.6 years, less than half the time IC wafers take to double in size (7.5 years). Kitihara's Law describes the evolution of FPD panels. In each three-year cycle, the average screen area grows 44 percent, while the power consumption required for a given function decreases by 44 percent and panel thickness and weight is reduced by one-third. The law further estimates that the number of bits needed to specify the image on the screen has increased fourfold every cycle, matching Moore's Law for ICs. Odawara's Law centers on [desktop and wall mounted FPD] panel prices, stating that each doubling in the cumulative area of flat panels produced results in a cost reduction of 22 to 23 percent [per area produced]. The cost of a full-color active matrix LCD cell-phone display, however, is expected to remain about six times higher than would be expected by applying this theory. This creates a market-entry opportunity [in the small form factor screens] for AMLCDs made with polysilicon thin-film transistors, organic light-emitting diodes (OLEDs) and electronic paper." Keeping these laws on track will require alternative technologies or new materials, or both at some point.

• Battery energy density growth, cycle life growth, and charge time reduction, and the Pervasive Computing threshold. Since batteries were commercialized 147 years ago we have seen just two doublings in energy density (a modest constant of 0.9% capacity growth per year), three doublings in cycle life (1.4% improvement/year, and six halvings in charge time (3% improvement/year). Nevertheless, after decades of research neglect, recent attempts to improve the nanostructure of solid battery technology have yielded dramatic improvements in cycle life (more than a 2X increase), and charge time (an impressive 15X reduction) in just the last few years. We can anticipate these significant gains to increase over the near term, taking us past a "pervasiveness threshold," where our portable electronics will experience no significant downtime. Figures below are for commercial cylindrical batteries, cost/cycle is in 2001 prices [27,28]:

A123 Systems, an MIT spinoff, is in phase two multimillion dollar development on nanobattery they claim is 80% lighter, has 10X longer cycle life, 5X power gain, and "dramatically faster" charge time vs. conventional batteries. Toshiba has announced a portable nanobattery, expected commercially in 2008, which recharges 80% of its power in 60 seconds. This will allow users of portable and wearable electronics to recharge while waiting, eliminating one of the major problems with current portability of the 3D web. In addition to enabling ubiquitous portable electronics, nanobatteries will eventually allow hybrid electric vehicle owners to recharge their batteries as fast as they fill their gas tanks, at fast charge stations like PosiCharge. With electricity at only 70 cents per gallon equivalent when drawn in off-peak hours (roughly twice this cost in peak hours), these may be deployed privately in coming years, once the platform matures.

• Network address density growth (Poor's Law) and the Transparent Society. In the early 1990's Robert Poor observed that spatial network address density, a measurement of digital connectivity, doubles every 12 months in urban environments [29]. This rapid doubling rate, more rapid than all the growth constants yet discussed (and not yet independently quantified) has major implications for the structure of physical reality and nature of the 3D web. Taken together with constant growth trends in miniaturization, processing, storage, bandwidth, energetics, and sensor and effector efficiency, this trend makes it reasonable to assume a coming world of incredible public and consented private data accessibility, or
"transparency" in our global digital networks, as outlined in David Brin's The Transparent Society, 1998 [30], as well as a very high degree of fidelity in our digital and visual "mirror worlds," as forecast in David Gelernter's Mirror Worlds, 1993 [31]. The next networking protocol, IPv6, already in early deployment, will move us beyond the present 4.3 billion addresses of IPv4 (allowing not even one IP address per person) to 3.4x10^38 addreses, or 50 octillion for each of the 6.5 billion people alive today, plenty enough to give every important physical item, no matter how small, its own connection to the web. This will enable a world of embedded, ubiquitous, and pervasive computing, as outlined in such books as When Things Start to Think (2000) [4], Ambient Findabilty (2005) [32], Geospatial Matters (2006) [33], and Everyware (2006) [34].

• Information technology's (IT's) share of GDP and economic growth and the emerging Metaverse Sector. In the way it is currently measured by the Economics and Statistics Administration (ESA) of the Department of Commerce, the U.S. information technology sector (IT hardware, software, and services, including communications, both business and consumer) contributed only 8% to U.S. GDP in 2003 [38]. What's more, its growth as a percentage of our total economy is an anemic 3% per year, a doubling time of 23 years [35]. But even though the IT sector's total share of the economy remains low, its marginal contribution to U.S. productivity is very high, indicating a structural transition is underway. Between 1997 and 2003, the IT sector's contribution to marginal GDP growth averaged 40%, and was as high as 80% in some recessionary years (Digital Economy 2003, Table 1.1). IT's contribution to the productivity of non-IT sector industries (entertainment, education, manufacturing, retail, tourism) is not included here, and must make the total IT figure substantially higher. This is evidence that IT, not services, is now the largest marginal driver of U.S. economic growth. Classic economic theory charts U.S. economic development from Resources, to Products, to Services sectors over time. Structural change occurs in an economy when GDP (or for the world, GWP) or employment in one sector grows to exceed another. We contend that the emerging IT-based metaverse sector, economic activity enabled and managed by our 2D and early 3D digital web (including not only traditional IT but all digital media, geospatial web, wireless and participatory web technologies), will surpass, encompass and redefine each of the three traditional economic sectors in coming decades, the way each have done during their own historical emergence. Providing plausible indicators and examples of metaverse sector development is a major goal of our roadmap. From a marginal perspective, it is clear this transition has already taken place. It may take another generation or two for the metaverse sector, however liberally defined, to grow to represent the majority of us GDP or employment, but the marginal slowdown in traditional economic sectors, as well as general economic slowdown during the transition, is already being charted by leading economists, as with The Way it Worked and Why it Won't: Structural Change and the Slowdown of U.S. Economic Growth, Gordon Bjork's 1999 analysis of the maturing U.S. economy, as it waits for the emergence of a global metaverse-enabled service and automation economy.

• Virtual world persistence. In contrast to game titles, which experience rapid turnover and obsolescence, synthetic worlds rarely go away once they achieve any kind of subscriber base. Familiarity and community creates unique inertia versus standalone games. The seven year old Ultima Online, for example, had 150,000 paying subscribers in 2006. To this day, no virtual world has ever been completely shut down unless the operating company absolutely forced it to be. Hardcore users keep the communities alive. Games like Uru Live, the never-finished online Myst successor, have been kept alive by the beta user community. Arthur C. Clarke made this general observation: "No communication technology has ever disappeared, but instead becomes increasingly less important as the technology horizon widens." Even the most primitive communication technologies (semaphore, morse code) remain used by a subculture interested in history. Increasingly, such may be the case with any virtual world that develops a passionate user base, subject of course to legal and economic hurdles.

• From Janna Anderson’s research, the key themes of any communication technology center around addressing such values as:
  - Identity, Reputation, and Trust
  - Security and Privacy
  - Connection, Persistence, and Community
  - Ease of Use, Portability, Interoperability
  - Affordability and Value
Each community of users has a threshold value for each of these that must be met before a social space will be broadly adopted.

• Major technological development will be a necessity for mass web collaboration and subsequent 3D web adoption. Many basic features of the participatory web have yet to be developed, and there is a common assumption that we'll need much better collaboration tools (artistic, scientific, commercial, and social, etc.) before the web will become the preferred social medium over today's dominant communication media, including face to face. Once that support exists, some assume that people who find it most efficient to collaborate on the web will then gravitate to 3D spaces as their technology allows, simply because 3D is more intuitive and less isolating than existing interfaces for some forms of social interaction and exploration.

• Metaverse designers commonly assume that extending a real world metaphor to virtual space is the best way to interact with computers and other people. But our fantasy worlds play fast and loose with the laws of physics, inventing new rules with less restrictiveness, and even nonfiction virtual worlds have common features, such as flying and teleportation that would be "supernatural" in physical space. So it might be more accurate to say that our best virtual virtual worlds seek to recreate the reality of physical space, yet to also exceed the limitations of that reality, in a consistent and intuitive way, with persistent rules that can be learned through play, as in physical space.

• The metaverse is a shared social space that needs 3D capacity, but there are many instances where it will not be 3D. If we try to define the metaverse in shorthand as "the 3D web," we communicate a mistaken assumption that the metaverse will be all 3D, or even 3D most of the time, to most users. But our most efficient and valuable interfaces for many types of communication will long remain 1D and 2D abstractions of a 3D world. Using a row of file cabinets in 3D would for many be a reduction of efficiency versus a navigating a 2D filing space. People are drawn to many 2D games, like Tetris or Puzzle Pirates. Not just for nostalgia, but because the simplicity is elegant, focusing, and efficient. 2D online collaboration environments, like WebEx, could continue to be more efficient than anything we can create in 3D. 3D videoconferencing, which is easily available in many work environments, is used today only for special occasions, such as meetings. It would be too distracting and disruptive to have on all the time, unless constant visual collaboration were necessary. In most cases, certainly in its early years, we will probably want to keep Metaversal 3D hidden but available. We’ll use it only if it doesn’t take too much of our system’s computing resources, and adds significant additional value to the social space. Computing resource load was a primary reason that VRML, a 1990’s 3D web protocol, was not able to scale beyond a small number of early adopters. More accurately then, we might refer to the metaverse as "the 3D enhanced web," and this roadmap is about discovering and forecasting opportunities and challenges to some of the enhancement paths.