No matter which publications you read, it's difficult to avoid chatter about the information superhighway, cyberspace, and/or the Internet. An article on yachting in the Caribbean in your Sunday paper, for instance, might be wedged next to one on the Internet ftp (San Jose Mercury News "Travel Section," June 5, 1994). Or if you were to take a vow of spirituality and subscribe to the New Age magazine Magical Blend, you'd find, next to an article on cancer-healing herbs, a piece on the information highway (June 1994). And if you think you can avoid the metaphors and find solace with a burger and fries, don't go to San Francisco's Icon Bar & Grill, which has a TCP/IP link to the Internet.
Puns aside, there continue to be newsworthy developments for those of us interested in software technologies and new opportunities, particularly given the ongoing stagnation and consolidation of the mainstream desktop-software market. The basic news here is that, for many in the industry, the focus has shifted from the pie-in-the-sky data highway, down to the pragmatic reality of its Neanderthal precursor, the messy but very real Internet.
Before examining some of these new opportunities, let's take a moment to dispel the more egregious hype about the information superhighway. Recently, San Francisco area newspapers had front-page stories on Pacific Bell's "ground-breaking ceremony for the information superhighway," a $16 billion project to connect residential homes with video-capable coax. Shown also on TV news broadcasts were scenes of a PacBell construction crew unrolling yards of coax onto a suburban street, as the mayor and other politicians looked on approvingly. Local residents were interviewed and were generally supportive, although it's likely that few could describe the end result. PacBell's news release states that this seven-year effort to rewire California will "help customers in their daily lives, learning, shopping, playing." PacBell ads trumpet: "The communications highway, it's real, California!" But try to order an ISDN link in San Francisco and you'll be told the central switch doesn't support it. As for that $16 billion, a Wall Street Journal analysis showed that, based on PacBell's capital spending in recent years, the company would have spent $14 billion over the same period anyway.
And PacBell has plenty of company when it comes to inflating these numbers. In the April 1994 issue of DDJ, Jonathan Erickson noted that Southwestern Bell was also using construction of the information highway as an excuse for proposing changes in their rate structure. Similar grand announcements about mammoth projects have been made by Bell Atlantic, MCI, Time-Warner, Ameritech, and others, resulting in a total investment of $100 billion. Yet the Wall Street Journal's John Keller concludes that "most of that total involves basic capital spending that the companies would be investing, highway or not. Only a small portion of it will go for new interactive gear--and that part isn't nearly enough." (WSJ, May 18, 1994).
So why the grand pronouncements? Analyst Mark Stahlman of New Media Associates explains:
It is strongly in the interest of both cable and telephone companies to exaggerate their plans for interactive systems deployment. After all, they are constantly asking for regulatory relief. Moreover, it is strongly in both their interest to exaggerate the competitive threat posed by the other guy. [Wired, March 1994]
This explanation jibes with recent rhetoric by cable and telephone companies. Brian Roberts, president of Comcast, the nation's fourth largest cable company, writes: "The Federal Communications Commission's micromanagement of cable rates will sap the industry's incentive to invest in new technologies_to keep moving forward and creating jobs. I hope these new rules will be rethought, and soon." (WSJ, May 19, 1994).
Likewise, Phil Quigley, chairman and CEO of Pacific Telesis, has written:
The linchpin of federal telephone regulation, the 1982 consent decree that broke up the Bell System in 1984, is strangling the lively competition it was meant to protect. In the process, that decree is stalling the delivery of new products to consumers and propping up the artificially high price of long-distance telephone service_. It's time to let the regional Bell companies become full-service providers, offering local and long-distance service as well as interactive video. It's time to let cable companies offer telephone service. And it's time to let long-distance companies into the local telephone market.
According to Richard Brown, vice-chairman of Ameritech (also known as Illinois Bell), we are at the point in history when we will "shape and deploy the most powerful technology in the history of human civilization." This, he adds, is "a job that will require full competition at every part of the existing network." (Digital Media newsletter, July 26, 1993).
The end result of these Herculean efforts will be what is known as the "digital convergence," in which everything gets digitized and compressed, and the boundaries between voice, video, image, and textual data disappear into the infamous melange of 500 channels. As the saying goes, it's all just bits. At first glance, the convergence of different data types into a homogenous bitstream seems to imply competition and conflict between the industries that currently deliver the data types. But the conflict between the long-distance carriers and the local telephone and cable companies is as genuine as TV wrestling.
Rather than real competition, the goal is to establish media monopolies for the new millennium. Much of the talk between these corporate giants has been of mergers, acquisitions, and joint ventures (by one count, more than 300 alliances) rather than competition. Intel's Andy Grove has remarked that greed is the central motivating factor behind the digital convergence. So it's no surprise that, when the Senate Commerce Committee recently considered a telecommunications bill that would let regional telephone companies offer cable-TV services to their customers, the Committee faced surprisingly harsh opposition from those very companies. USWest chairman Richard McCormick told the Committee, "If the bill passes in its current form, we will stop building the information highway within our service territory" (WSJ, May 5, 1994). Likewise, the president of the National Cable Television Association asked the committee to go slow in relaxing restrictions on the "Baby Bells."
This situation has led one observer to comment: "Cable and telephone operators seem prepared to cut their noses off to spite their faces_both contingents continue to pressure Congress to pass bills that will favor them and cut the other provider out of as much of the digital data delivery business as possible" (Fred Dawson, Digital Media, May 16, 1994). There's more to this story, of course, and there are some complex issues here, but one's eyes rapidly glaze over. For the moment, it seems that the business opportunity of the 21st century has come to a legal standstill.
But even if the regulatory issues were resolved, there's reason to be skeptical. Mark Stahlman emphatically states:
It won't work. All networks aren't created equal. Bandwidth must be configured to be useful, and the switched topology of voice telephony is fundamentally different from the broadband/broadcast character of cable television. These systems have developed and continue to evolve to address very different business needs. They will not converge. There is neither the business nor the technical basis for them to converge.
And technical issues aside, a look at history shows that it is not the old, large, established companies that prevail when a radical shift in technology occurs. The growth of the auto industry in the early 1900s was a result of many small startups--new barnyard-based manufacturers that, for better or worse, brought us into the age of asphalt-and-concrete highways. (Barns were to turn-of-the-century entrepreneurs what garages are today.) Our present-day car culture is not a result of successful mergers between turn-of-the-century railroads, horse ranchers, and buggy manufacturers.
One thing you can be assured of is that the winner of the coax/copper/fiber wars will need the assistance of computer manufacturers and software companies. The major players in the computer industry realize this, of course, and are jockeying for position to provide the smarts for the underlying infrastructure. Microsoft, Oracle and Sybase have all announced video servers, positioning them as key components of the information highway. Microsoft's entry, known as Tiger, is being developed by its Advanced Consumer Technology (ACT) division at Redmond, with 500 employees and a yearly budget of $100 million. Around Redmond, the information highway is called ITV(interactive television).
In keeping with its tradition of trying to win all battles on all fronts, Microsoft's strategy does not restrict it just to the video server. Reportedly, ACT has set up a mock suburban house on the Redmond campus, complete with kitchen and appliances, to test interoperability between various household components. Those of us who have joked about "Windows for Cuisinarts" may eventually find that this is no joke.
All this is yet to be proven. Nathan Myhrvold, senior VP of technology at Microsoft, says:
At the heart of Microsoft's ITV solution is an object-oriented distributed operating system to support the entire network. This means that Microsoft's ITV operating-system software will be located in the set-top device in the home, at the network operator's central facility, and in media servers throughout the network. This approach is different from a client-server model, which takes a less integrated approach to intelligence across the network. We will incorporate revolutionary operating-system management features such as rate-guaranteed kernel design, which guarantees the fastest possible network response time, delivering all consumer requests for time-sensitive data with no disruption in service. [TechEd '94 Conference keynote]
Those who are trying to run Windows NT in performance-critical applications and who are now waiting for Daytona--the evolutionary update to NT reportedly going through an unexpected second beta round--are not holding their breath for this "revolutionary" OS technology.
Also caught up in the future is Paul Allen, who was Bill Gates's original partner in founding Microsoft. Allen's new venture is a company called Starwave, which will be one of the companies providing content--in this case, information for "serious sports fans"--for an upcoming online service from Ziff-Davis. Starwave is awash in highway-oriented metaphors, as evidenced by this recent recruitment ad:
We're taking the fast lane of the digital superhighway. In the year 2000, will the digital highway be a dingy strip mall of psychic advisor infomercials and sitcom reruns?_ We think not. Starwave is a member of the Paul Allen family of multimedia companies, specifically building content for the consumer information networks of the near future. We are a team of artists, writers, engineers, and Hollywood types building the next generation of entertainment products.
In the case of Allen's company, Asymetrix, its Toolbook software was initially a product whose concept was better than its execution (as with his Portland Trailblazers
professional basketball team). One hopes this will not be the case with Starwave.
In contrast to these future-oriented and somewhat vaporous ventures, there is a recent Bay Area startup, Mosaic Communications Corp. (MCC), started by Jim Clark, founder and former chairman of Silicon Graphics. In the pre-startup stage, Clark explored a range of ideas having to do with the information superhighway. One possibility was creating an online network for Nintendo. Reportedly, Clark's partner Marc Andreessen, suggested: "Maybe the Internet is already the information superhighway." The notion is simple: Why wait for the uncertain future when you can make money in the here-and-now?
Andreessen, the 22-year-old principal author of the widely distributed NCSA Mosaic program, says "the Internet is now in the middle of a transition to a full consumer network_.This is the opportunity for us to come in below the radar screen. By the time the switched broadband network comes into being, we will already be there." (San Jose Mercury News, May 16, 1994). NCSA Mosaic is a tool for browsing the Worldwide Web, a distributed hypertext system previously discussed in Dr. Dobb's Developer Update (February 1994). NCSA Mosaic was developed by Andreessen and his team at the National Center for Supercomputing Applications, a non-profit research organization affiliated with the University of
Illinois. NCSA is known for popular, high-quality, freely distributable software, such as Telnet, Image, PalEdit, XDataSlice, AudibleImage, and Collage. But the popularity of its Mosaic browser--which comes in Windows, Macintosh, and X Window flavors--is so large it is now known as the Internet's "killer app." NCSA has tallied 250,000 downloads of the software, and estimates more than 1 million Mosaic users on the Internet.
The Internet is now estimated to have 15 million users on 1.5 million machines, and is growing at 10 percent per month--or one user every two minutes. InterNIC, the organization that allocates Internet addresses, assigned names for 14,000 new subnetworks in one month early this year. The Internet Society estimates that 55 percent of new users are commercial rather than academic or nonprofit. Yet as fast as the Internet is growing, Mosaic usage is growing faster still. Network traffic attributed to Mosaic grew 500 percent in the last year.
Mosaic combines a multimedia front end--capable of delivering data types such as graphics, sound, image, and any other data type for which a viewer exists--with a multiprotocol back end. The back end is based on the libwww code library from CERN, the European physics-research facility that originally defined the WorldWide Web protocols. The back end handles with Web's HTTP protocol and can also do NNTP, Gopher, WAIS, FTP, and local file access. The code library also parses files in the Web's hypertext format, an SGML subset called HTML. CERN recently placed this code library, along with the HTTP and HTML protocols, in the public domain. NCSA Mosaic, which is only 9000 lines of code, does a masterful job of integrating these diverse components. The result is a powerful tool for taming the massive, unruly tangle of information now available on the Internet. Mosaic is not the only browser for the Web, there are many others (such as Chimera, Cello, Viola, Lynx, Emacs-w3). But, compared to the other graduate-student efforts, it is the best looking and best supported--and runs on multiple platforms.
The development team at NCSA recently suffered a heavy blow, when many of its key members left. Marc Andreessen, who was reportedly making $6.85 an hour, is now VP of technology for Mosaic Communications Corp. According to a recent Netnews posting by Andreessen, he hired "virtually the entire core Mosaic technology development team--Eric Bina, Chris Houck, Rob McCool, Jon Mittelhauser, and Aleks Totic--as well as Lou Montulli, author of Lynx (a popular character-mode browser from the University of Kansas)." In addition to this team, all of whom are in their early 20s, "we have a core set of killer ex-SGI people, mostly in their 30s" (Open Systems Today, June 6, 1994).
One who takes exception to this is Chris Wilson, another ex-NCSA programmer and now product development lead at Spry Inc. of Seattle. In a recent posting (comp.infosystems, May 10, 1994), he writes:
I'd like to take public offense at this. First of all, considering that the above does not include me, it is NOT virtually the entire core Mosaic development team' from NCSA. As I am one of the two people in that group who were full-time employees for any length of time, that is incorrect. There are also still people left at NCSA working on Mosaic.
Public bickering aside, there is no doubt that Mosaic and its descendants will be the dominant Internet-access tool over the next few years. Eight companies have announced commercial licenses of NCSA Mosaic: EPRI, InfoSeek, Quarterdeck, Quadralay, SCO, Spry, and Ubique. O'Reilly and Associates, publisher of the well-received Whole Internet User's Guide and Catalog, by Ed Krol (see "Programmer's Bookshelf" DDJ, February 1993) has announced "Internet-in-a-Box," a software package that includes Mosaic software enhanced by Chris Wilson and his team at Spry (the company numbers 60 people already). Given that Krol's book has sold 250,000 copies, the prospects for Internet-in-a-Box are bright indeed.
Many other entrepreneurs are springing up to take advantage of the Internet's rapid growth. One by one, they are addressing known problems and fixing them. One of the biggest problems with Internet usage is the lack of a user interface. This is now being addressed by the ventures mentioned earlier, and by over 30 books oriented to the beginning user.
A second problem with the Internet is obtaining access. As you probably know, there is a difference between being able to send e-mail to an Internet address (which you can do from MCI, CompuServe, Prodigy, America Online, and perhaps your company's LAN e-mail gateway) and really being "on the Net"--that is, being IP-connected, and therefore able to run telnet, irc, ftp, gopher, archie, veronica, Mosaic, and so on, from your local machine.
This distinction is not clear to many people. When America OnLine and Prodigy installed limited Internet support, there was a barrage of messages from confused users. In the case of Prodigy, which only supports e-mail to Internet, one user wrote: "I do not have a clue about this Internet stuff. Do I have it or what? If I have Prodigy am I in the Internet?_ Do I need a phone number or something?" (Boardwatch, June 1994). In response, many access providers have come out with a range of ways to get connected, spanning the range from a 9600-bps SLIP connection (for $20/month)
all the way to a 1.544 Mbps T-1 line ($2,000/month, not including phone-company charges). The access providers and phone companies seem to be the only ones making real money at the moment. One company, Netcom, has grown from a small Bay Area provider, to one with 40,000 users (most of them dial-up connections) nationwide in the last two years.
Other than selling access tools or services, it is still unclear how to make money on the Internet. The technology infrastructure may be in place, but the business infrastructure is still being built (not to mention the cultural sea change that is required). This is being addressed by another group of other ventures. One of the oldest is Clarinet, which is about the only company to make money selling content. Clarinet provides a Netnews-style feed that includes material from Reuters, Newsbytes, and other wire services. The price is per-user and ranges from 40 cents to $3 per month. Netcom pays over $15,000 a month to provide this service to its users.
Although other ventures, such as QuoteCom, are getting into the business of selling content, this goes against the prevailing Internet culture, which encourages information to be free. A different approach is being followed by Randy Adams, a programmer who developed the LaserTalk utility for debugging PostScript programs several years ago, and then sold his company, Emerald City Software, to Adobe in 1990. Adams has transformed himself into a mass marketer by founding the Internet Shopping Network (ISN), which sells physical goods rather than content. ISN uses the Internet as a means by which customers can browse its electronic catalog, which has 22,000 computer products from 1,000 different manufacturers. Using a Web browser such as Mosaic, customers can order these goods electronically. Presently, about 100 people a day are doing so. Advertising and startup costs for ISN are minimal.
Former musician and designer Jeannie Novak of Kaleidospace has come up with another angle to the problem of making money. Her two-person company sells virtual space to artists who want to sell their creations over the Internet. Using Mosaic, you can view a catalog of an artist's paintings, hear portions of their music, or watch clips of performances or films. Rents on the Kaleidospace server start at $50 per month. Novak says:
By using the Internet, small companies can develop a virtual presence as rich, connected, and powerful as the largest corporation. It was possible for Kaleidospace to create an Internet site, potentially visible to 20 million users_ for about $30,000_. There's little distinction between company size outside the Internet, mainly because there's little difference in the quality of their electronic presence. [Computer Currents, May 1994]
The Kaleidospace visuals are certainly striking and attractive, the result of many 3 a.m. sessions struggling with the rather primitive HTML authoring tools.
For developers, it comes down to leveraging the existing electronic infrastructure here today on the Internet, or waiting for a grandiose future built on vagueness and promise. Some have the luxury to wait, most do not.