The Software Revolution Part 1Amy Cortese
Imagine that to read this magazine, you needed a special program for the text, another program to view the photographs, and yet another to look at the charts. Or, to watch your favorite TV shows, you needed a CBS viewer to watch David Letterman and an NBC viewer to watch Seinfeld. Oh, yes--you'd also wind up buying a new, "upgraded" viewer every year or two. And to make sure that viewer does its best, you'd buy a new TV, too.
You wouldn't stand for it. But this is precisely what computer users have been doing for years. At the root of this situation is the way that software is created, distributed, and consumed. Programs such as spreadsheets or word processing packages are written for a particular type of hardware and operating system--so your Windows Excel spreadsheet won't work on your Macintosh. And even if they're written for the same operating system, programs from different software makers won't work easily together. Worse, programs from the same suppliers don't necessarily work easily together because data must be arranged in a particular way for each: Your spreadsheet package can't deal with the text from your word processing package, and vice versa.
One way software makers have gotten around this is by creating "bloatware"--ever-fatter packages that throw in dozens of new features. With each upgrade, the customer has less need to look elsewhere. Today, the most popular business software comes in "suites," collections of programs sold as deeply discounted bundles surrounded by gobs of new code to make it appear that the programs actually mesh with one another.
This may not be the best solution for computer users who just want a better way to get at the information they need. But it's great for PC and software makers. The escalating demands of bloatware drive sales of ever-more-powerful computers, creating an unholy alliance between software and hardware makers. This fall, for example, millions of consumers who want Microsoft Corp.'s Windows 95 operating system will spend billions to trade up to PCs using Intel Corp.'s speedy Pentium chips.
Why do customers go through this? They have few alternatives. The "Wintel" standard (PCs that run Microsoft Windows and use Intel processors) represents 80% of PCs sold, providing a standard "platform" of hardware and operating-system software that hundreds of PC makers sell. That keeps prices low and ensures that there will be thousands of programs available. Wintel is also a huge improvement over the bad old days when customers were locked in to IBM or Digital Equipment Corp. machines because they used software that ran only on those brands. But even with Wintel, computer buyers are locked in to the platform: They get advances only when Microsoft and Intel deliver them, and they pay whatever Microsoft and Intel determine they should.
What if this cycle could be broken? What if you could find the information you're after without having to take into account what kind of program to use, what computer it runs on, or what kind of format it's in? What if software as we know it were to disappear? What if software could be reinvented?
Well, that's exactly what's beginning to happen--and at a pace nobody in the computer industry anticipated. The force that's shaking the foundations of the old software business is the Internet and its graphical subnetwork known as the World Wide Web. First, the Internet's TCP/IP communications standards made it possible for tens of millions of computers using different operating systems and applications programs to "talk" with one another--whether they're on a local-area network or positioned at opposite ends of the globe. Then, the Web's HTML (hypertext markup language) gave all these computers a lingua franca for displaying information in graphical "pages."
FALLING INFORMATION WALLS. In the two years since the Web and the Mosaic program for viewing its pages emerged from research labs, the Web has turned into a huge virtual disk drive. It's crammed with every possible form of information--from online magazines to digitized film archives to radio programs--all available at the click of your mouse on a blue "hyperlink." The same Web document is accessible to anyone with an HTML browser, whether it's on a Unix workstation, a Windows PC, or a Macintosh. Says Rusty Rahm, president of StarNine Systems, a maker of Web software for Apple computers: "On the Internet, nobody knows you're a Mac."
Suddenly, the barriers that have kept information from flowing between different brands of computers and software have begun to crumble. The next step will be critical: using the Web not only to make the same information available to all wired machines but to let them share the same programs. If that can be done, there will be a basic shift in the software business no less seismic than the fall of the Berlin Wall. It, too, portends a new world order and vast new opportunities.
The new software order will reflect the character of the Internet itself. Barriers to entry will be minimal: Anybody with a network link can play. Costs of goods sold will be the price of sending some bits down the wire. Bloatware blockbusters costing hundreds of dollars may give way to software on demand--snippets of new programming that come across the Net like E-mail. In short, the Internet, says Eric Schmidt, chief technology officer at Sun Microsystems Inc., "enables the deconstruction and the construction of a new economic model for the software industry."
WINTEL'S DISCONTENT? What of the old model? In the face of the Web--a truly universal standard--the prevailing Wintel standard doesn't look so unshakable. And the huge profits that Microsoft and Intel get by setting the standards don't look so safe. Since September, the hardware business has been buzzing about low-cost "Internet appliances" that will challenge Intel PCs.
Now, Microsoft is in the hot seat. On Nov. 16, Rick G. Sherlund, the influential Goldman, Sachs & Co. analyst, dropped Microsoft from the firm's recommended list, citing slower sales growth in conventional PC software and warning of the rising challenge to the giant from Internet software suppliers. The Web, he noted, "is a serious threat to Microsoft's ability to set standards for important parts of the industry."
The news of Sherlund's downgrade sent Microsoft shares tumbling 7%, to 87 3/8 in two sessions--and boosted Internet stocks. In an ironic twist, the already white-hot shares of Netscape Communications Corp. passed 110 1/2, making company founder James H. Clark the newest software billionaire.
Despite this dramatic reversal on Wall Street, William H. Gates III, the first software billionaire, is not exactly on his way to oblivion. All the Internet software upstarts combined are a mere speck compared with his $6 billion empire. And now, his organization has gotten the Internet religion--big-time. Gates has made it clear to all his troops that the Net and the Web are now Microsoft's highest priority.
All Microsoft's products, from Windows to Word to Excel, are being extended to embrace the Net. Gates's new book, The Road Ahead, is largely about the Net. And at Comdex, the giant computer-industry show held in mid-November, Gates used his keynote speech to sketch out ways in which Microsoft plans to build a new set of Net software on top of its existing programs. "The Internet is nothing but good news for software," he said. "When you get low-cost communications, you want more software to help people share information and collaborate. It means the PC is a more relevant device to a broader set of people."
Gates's challenges may be more structural and managerial than technical. Microsoft finds itself in a position similar to that of IBM a decade ago, when the PC revolution began threatening the mainframe business: Even as he tries to move into the new market, he can't afford to let up on efforts to stay on top of the old industry. Operating systems and packages such as Microsoft Office are where all the revenues are now--and the source of the 85% gross margins that support the massive organization. What's more, Gates concedes that having the lead in PC software may not land him near the top of the new order. "We're in a business where no amount of success guarantees future success," he says.
Meanwhile, thousands of programmers and computer designers are racing to prove that on the Web, there's a better way--a chance to end software lock-in forever and perhaps make bloatware obsolete. The breakthrough technology? It could be a programming language from Sun Microsystems called Java.
Once a computer--no matter what brand--is equipped with Java "client" software, it can run any Java application that comes across the Net. Click on a hyperlink to your bank to check loan rates and you'll get the data as well as a Java "applet," a special application program to calculate what your monthly payments would be for different amounts and different loan lengths. Soon, little Java programs may hit your PC the way E-mail does. "The Web turned the Internet into a giant disk drive," says Mark Pesce, a San Francisco creator of Web software and a Java enthusiast. "Java turns the Internet into a giant processor."
JAVA'S MASSIVE BUZZ. Java embodies two key attributes of the new software business. It is designed specifically for the Net, and it is an object-oriented programming tool, based on the object language C++. Object technology turns conventional software on its head. Since the earliest days of computing, programs and data have been rigidly separated. The financial information you want to assess and the program that does the number-crunching are completely independent, yet useless without one another. With object technology, information and programming are merged. These "objects" behave the way objects in the real world would: One called "checkbook" might tally up your checking account, for example.
Object programming is ideally suited for the Net era--when computers will function more as multimedia communications devices and less as glorified calculators and typewriters. With objects, information comes to the forefront, and the program that places it on your screen recedes. "The result of this is that software takes its place as a means rather than an end," says Clifford J. Reeves, director of IBM workgroup architecture.
While Java is generating a massive buzz in the computer industry, it is just one of many new programming schemes that have suddenly become practical on the Web. Steven P. Jobs's NeXT Computer Inc., for instance, is reorienting its object-programming software for the Net. "The two most exciting things right now in software are objects and the Web," says Jobs. IBM and Apple Computer Inc. have been promoting a distributed-object scheme called OpenDoc. Apple is now working on technology called CyberDog that lets developers link OpenDoc objects to the Internet. And separately, IBM is developing a framework for embedding different types of objects in Web documents.
There will be Java rivals, too. While considering licensing Java, IBM is working on its own "Web-executable" language. It's a combination of a new programming language called Bart and LotusScript, the development software that will ship with an upcoming release of Notes, IBM's Lotus subsidiary's groupware program. Netscape--which will bundle Java into the next release of its Navigator browser--has its own language, called LiveScript. It's based on Java, but easier to use. And Microsoft says it will have a Java equivalent in its Visual Basic language--once that is adapted to the Net.
From tiny one-person shops to startups staffed by prominent Silicon Valley executives, the Net is teeming with new Web software companies. Most wouldn't have a prayer in the conventional software market. But the Net represents a green field where no software maker dominates. "The Internet changes everything," says J. Neil Weintraut, managing director for technology research at Hambrecht & Quist Inc.
Start with something as simple as distribution. Today, software makers spend huge sums to crank out disks, put them in appealing packages, and move them through wholesalers to store shelves. Then, they spend millions more on advertising to get customers to go into the stores and buy. On the Web, a customer simply clicks a few onscreen buttons, and the software comes back across the line. When there is a new version of software, "you turn on your network computer and it will show up. There is no store to go to. There is no installation," says Ellison. Today, it can take anywhere from several minutes to several hours to download a program, but high-speed communications will make electronic distribution more practical.
The way you pay for software will change, too. Instead of paying a one-time license fee, which entitles you to perpetual use of a program, you may pay for software the way you pay for a magazine. For example, software companies may offer subscriptions that, for an annual fee, entitle you to unlimited usage and the latest updates to a program or group of programs. At the same time, all sorts of network services--stock quotes, digital photo archives, bill-paying services, and the like will be delivered online, along with software to use them. "What you're really selling is a service fielded over the Internet--the software comes for free," says John Landry, a former Lotus executive and now an IBM consultant.
Eventually, all we'll think about is the content--we'll assume the software is there. "The lines between content and applications are blurring," says Marc Andreessen, the 24-year-old programmer who helped dream up Mosaic, the original Web-browsing software, and is now vice-president for technology at Netscape. Helping the process will be Java and other Web-programming systems that make it possible to ship "interactive content" across the Net. An example might be an electronic catalog that includes a Java applet for ordering. When you hit the "buy" button, the program will execute on your computer--then vanish when the transaction is complete. Scott McNealy, Sun's CEO, calls it disposable software.
Bloatware, on the other hand, is just about the opposite: It's big, it's not cheap, and it requires a long-term commitment. And because of these characteristics, the biggest companies are almost guaranteed to win the bloatware game. After an investment of thousands of programmer-years of effort, Microsoft or Lotus or WordPerfect brings out a new, bigger release--loaded with features that most customers will never use, such as math equation editing. The latest Microsoft Office suite, for example, needs 55 megabytes of disk space and a Pentium-class computer to run at peak form. And installing these hefty programs has become a drawn-out affair.
For corporations, this has helped push the annual cost of supporting a PC user to about $8,000, according to consultants Gartner Group Inc. Increasingly, corporations are weighing the value of upgrading software every two years or so. This year, many corporations are taking a go-slow approach to Windows 95 because of the huge costs of upgrading. "Customers can't afford to be on this treadmill of bigger, better, faster," says Mark A. Tebbe, president of Lante Corp., a systems integration company in Chicago. Adds Thomas P. Kehler, CEO of Internet software maker Connect Inc.: "The software industry has been arrogant."
The Web may be the method to stop this madness. Instead of waiting two years for the next massive update to your favorite office suite, you may get the latest features instantly off the Net. Or, instead of buying a program you may only use occasionally, you may be able to rent it. The code will come across the Net and will be usable for a specified period. Think of it as just-in-time software.
The same concept may also extend to operating systems--with serious consequences for how hardware is designed. If schemes such as Java catch on, computers might no longer need ever-bigger operating systems--or the expensive Intel chips they use today. Low-cost Internet appliances, perhaps based on video-game players, could get by on simple, streamlined operating systems that call out to the Net for additional features as needed. "I see the PC going through this era of complexity and back to simplicity," says Ronan D. McGrath, chief information officer for Canadian National Railway Co. in Montreal.
It's an idea that has a lot of appeal in Silicon Valley--not least because it has the potential to undo the dominance of the Wintel standard. One of the most fervent supporters of the idea is Oracle CEO Lawrence J. Ellison. Oracle has developed an operating system that takes up just 1 megabyte of main memory, Ellison claims, vs. as much as 8 megabytes for Windows 95. He says the program will be given away by Internet services providers and built in to computers and consumer-electronics gadgets by next summer. After that, says Ellison, it's downhill for the PC. "I really think that Windows 95 marked the zenith of the personal-computer industry."
"A DARWINIAN DEVICE." Ellison is also a big backer of Java, which is built into the planned Oracle operating system. He says the new setup will have special appeal for corporations that want to cut PC support costs. Improvements to the software will simply show up across the Net. "We're talking about enormous savings of not only dollars but time," he says.
Predictably, Intel CEO Andrew S. Grove dismisses the notion that masses of customers will give up their Pentium PCs for cheap Internet cruisers. In fact, he says, those PCs helped make the Net so popular. "Where the hell was the Internet before there were 30 million multimedia PCs out there?" he demands. The PC, he adds, is "a Darwinian device" that will adapt, as it always has. Indeed, even if somebody can produce a $500 I-way Yugo, there may not be many consumers willing to give up all the other things a $1,500 PC can do to buy it.
Grove, whose management mantra is to stay "paranoid," may be better prepared for a fight than his Wintel partner. A quick switch to a software market of cheap network applets could quickly turn Microsoft's size advantage into a millstone. Its sprawling 18,000-employee organization may be the most potent software-development force on the planet, but it is also the world's biggest bloatware factory--which needs megahits, not lots of cheap applets, to survive.