Pacific Connection(英語)

To Be or Not to Be

Now that Apple Has Bought NeXT, Be's Operating System Must Survive on its Own

Last November, Be, Inc.--a California-based startup company headed by ex-Apple veteran Jean-Louis Gassee--announced a deal for its fledgling operating system. Power Computing Corp., the largest manufacturer of Macintosh clones, agreed to offer the BeOS as an alternative to the Mac OS. Under the deal, users booting up their machines get the choice of which operating system they want to run, thereby preserving Mac compatibility while having access to an operating system that claims to be faster and take better advantage of the PowerPC processors. The deal was struck amid what the two companies called "months of secret collaboration" and was announced at the Macworld Exposition, Boston.

The deal with Power Computing gave Be considerable attention here in the U.S., and helped fuel speculation for an even larger agreement with Apple. Many speculated that Apple would purchase Be outright. Gassee himself was publicly optimistic. "Apple has set itself up so that it has to make a major decision about its operating system by early next year," he told the San Francisco Chronicle in December. "What other choices [besides Be] do they have?"

The answer to that question was to come about two weeks later when Apple purchased not Be, but NeXT, the company headed by Apple co-founder Steve Jobs and developer of the NeXTStep operating system. By all accounts, the deciding issue was money. According to later reports, Be first assumed that Apple had no choice but to go with Be's technology, and asked for a whopping $400 million. Later, the price came down to 10 million shares of Apple stock, which at the time were valued at $225 million. But by that time, Apple had decided to go with NeXT, which it purchased for the $400 million--Be's original asking price. In doing so, Apple felt it was getting a more mature, proven technology than the BeOS, which by all accounts is still not a finished end-user product. And that NextStep, with its UNIX foundation, could better compete with Windows NT on the enterprise level. And of course--that acquiring Steve Jobs would garnera lot of attention by bringing back into the fold Apple's ever-visible co-founder.

At first glance, the Apple-NeXT-Be negotiations resembled those of other recent industry deals that have pitted Netscape Navigator against Microsoft Explorer, in which one technology gets accepted while another, perhaps equally promising, is spurned. But this is not the booming Internet market. What is especially notable about the mating dance between Apple, NeXT, and Be is at least one big difference here: all of the players are especially hungry.

For Apple, revamping the Mac OS has become one long nightmare. Looking back, one can count some half dozen separate attempts to produce a successor OS, including Apple's failed collaboration with IBM. Internal to Apple, the long-promised Copeland project never bore fruit, and was declared dead after the lead engineer left Apple for AT&T. Yet without a clear OS strategy, customers and industry analysts alike have wondered whether Apple would even survive against Microsoft, whose own turbulent OS strategy with OS/2 and Windows appears single- minded by comparison

Apple's operating environment has been its single strongest differentiator--and with Windows' domination, its single largest liability. Not only have corporate purchasers, even those loyal to Apple, strayed toward the Microsoft camp, but then have home consumers, long considered an Apple marketing strength. But in the quarter ending last December Apple reported that sales of the Performa family of home computers dropped, this despite an expensive marketing campaign and cuts in the Performa's price. The Performa contributes 15 to 20 percent of Apple's revenue and the result was a 10 percent drop in sales and a predicted loss of 80 cents to $1.20 a share, versus a 4 cents a share loss predicted by Wall Street analysts. That sent Apple's stock plunging from $21 down to $18 right on the eve of the Macworld show in San Francisco, where Gilbert Amelio, Apple's chairman, president and chief executive, was set to give what would turn out to be a 2.75 hour keynote presentation.

But if Apple was hungry for a new OS, NeXT Computer Inc. was equally hungry for a hardware platform it could truly call its own. Ever since NeXT discontinued its hardware offering, NextStep has been an operating system highly praised by developers, but never embraced by a public who could make that operating system a commercial success. According to internal NeXT documents reported by the Wall Street Journal, NeXT has not made a profit on its operations for at least three years. In 1995, NeXT reported an $801,000 net loss on revenues of $46.8 million, and for the next six months of 1986 it had an $8.5 million loss on revenues of $21.7 million. While NeXT did show a $1 million profit in 1994, it came from one-time licensing fees--mostly with Sun--whose success was not repeated in subsequent months.

And even with the purchase there is understandable skepticism. Some observers doubt that Steve Jobs, perhaps the industry's most vivid personality, can work with the more cautious Amelio. And then there are questions about compatibility. Apple has already made it clear that the new Apple operating NeXT-based system--code-named Rhapsody--will not support most System 7 applications in its first release. Merging technologies is never easy, and backward compatibility will not happen until at least 1998. For obvious reasons, Apple will continue to support System 7, with two updates planned for this year. Third party emulation software was developed for System 6 and ran "fairly well," but System 7's emulation has yet to be developed.

Be: the odd man out?

So where all this leave Be? When I called the company last December, the night before the Macworld show, a spokeswoman said that Be had no comment on any negotiations with Apple and said it would have no comment on the Apple-NeXT deal whatsoever. "It doesn't concern us," she said. On one level that's true. Be has created not only the software, but a hardware specification. There is no reason, at least in theory, why the company couldn't go on to create its own hardware-software platform based on one or more PowerPC CPUs. The only catch, and it's one that plagues every developer of operating systems, including IBM with OS/2, is getting developers to create applications. The BeOS--no matter how good it is--will only be an experimental technology and its audience will not grow past the early adapters who are running it on their Macs just to see what it will do.

And just what will the BeOS do?

"There are four areas where we hope to bring something to the marketplace, and I want to emphasize that word 'hope'--we are a young company and don't yet have a mature, proven product," said Gassee. "We see the BeOS supporting high-bandwidth, media- rich Web servers; Image processing, Photoshop type applications, digital audio, and digital video."

The sexiest feature of the BeOS is built-in symmetric multiprocessing, with the ability to move threads over more than one processor, depending on system loads. Be says the BeOS supports anything from a uniprocessor on up through a four processor system. By contrast, NeXT is only promising MP performance.

Obviously this is a specialized box--one that Apple may never make. Hence, Be--like NeXT before it--offers its own hardware. The BeBox runs two PowerPC microprocessors at 66 MHz or 133 MHz--depending on model--with the BeOS preconfigured. The company also offers a specification of the machine architecture for other manufacturers--of which, at this point, there are none. Be says that about three-fourths of its engineering effort goes toward software design, the remainder toward designing hardware.

Other BeOS features:

  • Multithreaded operation: Be claims that the BeOS will divide an application into multiple threads on its own--even if the programmer doesn't explicitly do so.

  • Object oriented design: Like all new operating systems, the BeOS claims object orientation.

  • Handles designs for realtime media and communications. Be says that BeOS is customized for realtime, high bandwidth data types, including audio and video.

  • Compactness: Be claims the operating system's API is highly compact compared to those of other operating systems.

    At Macworld, Be announced that it had begun shipping developers a version of the BeOS designed to run on Power Macintosh-based hardware. This version came in addition to the BeOS DR8 development kitintroduced last December. That kit supports the dual processor BeBox. Differences between the Power Mac and BeBox versions of the OS involve drivers at the hardware abstraction layer, which Be says are transparent to developers.

    Be also demonstrated a handful of BeOS-compatible software packages. The most visible development tool to date is from Metroworks, which has introduced a Be-native version of its CodeWarrior integrated development environment. Other packages include a word processor and drawing package, both from Lorrienne Software, as well as a Web server, presentation development tools and audio design system.

    BeOS Users

    So far, so good. But will anyone really buy this technology? Judging from the participants at Be's Usenet newsgroup ( appears that a sizable group of users are from the university community. One of them is Erik Steinmetz, who is working on a doctorate in artificial intelligence at the University of Minnesota. He's running the BeOS on a Macintosh 7600.

    Steinmetz is interested in the BeOS for a program he is developing to read and record games of Go in the SGF (smart game format). His interst stems from living in Japan for three years, where he came close to becoming an amateur sho- dan. "I've been back in the States for almost four years, and have hardly played more than once a month, and so dream of someday joining the ranks of computer Go programmers," Steinmetz wrote in an e-mail exchange. (Steinmetz says that his Go program would not be used for his Ph.D. thesis because, according to his advisor, game research is out of favor.)

    "I realize that this is an incredibly esoteric application to write, but I bought my BeBox to have fun programming, not to make money. Because of this, of course, I am not disturbed by the fact that Apple bought NeXT instead of Be. I think that if I ever write a program to play (instead of just record) Go, the inherent multiprocessor ability of the BeBox will be quite an advantage. Just today, Be announced a version running on four processors. I also *greatly* appreciate the Be API, for it makes writing applications fun and easy. One can concentrate on one's own code and ideas instead of spending too much time writing the code to get the menus to work, or trying to learn obtuse frameworks to do this for you."

    Steinmetz also believes that the BeOS's ability to create new threads, which can be quickly assigned to different processors, could also be an advantage. And he likes the fact that the OS was built as an object oriented system from the ground up--though, having never worked with NextStep, he couldn't compare the two.

    While academic developers like Steinmetz can show the potential of the BeOS, it will take the services of commercial developers to make the operating system truly viable. No OS can stand alone. They need development tools, and ultimately, applications. With the Apple deal gone, Be must now bank on there being a market for high-performance multimedia applications. If such applications are commercially viable, they will, by definition, look radically different from today's multimedia titles, running on faster processors, or multiple processors, and dealing with multiple streams of digital data. They will have to look as different as the Apple Lisa and Macintosh computers did when they were first hatched onto a skeptical market.

    An Interview with Jean-Louis Gassee, CEO, Be, Inc.

    Jean-Louis Gassee was with Apple for 10 years, serving as the president of Apple's R&D and manufacturing divisions. Before that he was senior vice president of research and development and vice president of product development. He now heads Be, Inc. He is widely credited with helping turn the Macintosh into a household name. He himself is a household name in the Silicon Valley, and--in the words on one local editor--is "intellectual, funny, epigrammatic, and crude." The interview was conducted the Monday after Macworld.

    It seems to me that developing a successful new operating system is about the toughest thing you can do in the software business.
    It is indeed a high-risk, high-reward proposition. But the founders of Be were involved with several operating systems over the course of the years, including, of course, the Mac OS at Apple. It became clear to us that there was a gigantic opportunity because of two factors: the growing girth and fragility of the Mac OS, and the because the Amega was leaving the scene--not because it was a bad computer, but because Commodore killed it.
    What is the BeOS's market niche?
    Multimedia. Imagine an "f" curve--which describes most phenomenon in nature. [An f-curve describes a growth rate that is initially slow, then accelerates rapidly before tapering off.] The office market is at the upper shoulder of the curve, while the multimedia market is at the knee--where it is poised to explode. I think there's a great opportunity here. Because of the availability of computing, communications, graphics and storage, there is no area of computing that will be untouched by multimedia. Look at emerging media such as digital video cameras and DVD [Direct Video Disc]. These are going to accelerate the demand for creative multimedia applications. That's where we are focused.
    Given Apple's acquisition of NeXT, Microsoft's marketing advantages with NT, and IBM's failure with OS/2--can you really hang in?
    That's a good question. Let's go over these things one-by- one. OS/2 was a system for the office market. It had technical merits but targeted a mature market, with a leader that was extremely well trenched. NT is an enterprise server operating system--it's very good, a holy terror in the office market, designed by the gentleman who was running VMS at Digital--Dave Cutler. So NT is really a UNIX killer. But it was not designed for real-time, high-bandwidth, applications that make use of multiple streams of multimedia. Both have their merits--but in different application domains.
    NextStep is another example of an enterprise play--it has scored points in enterprise-wide, mission-critical, custom development. So clearly, this is an enterprise play--something Apple did because perennially, they have had trouble gaining a foothold in corporate America. One of the reasons is that the Mac OS is not considered a robust, enterprise-wide operating system. The thought is that NextStep offers that to Apple. NextStep is based on a UNIX kernel, Display PostScript, and custom programming tools that have been well-tested in the marketplace. But it is not a multimedia system.
    Do you envision Be-compliant applications as requiring multiprocessor boxes?
    Not necessarily. The advantage of our system is that it scales up and down very gracefully. We do demonstrations where we start with a four-processor machine-- running it on all processors, then turn the processors off one-by-one. The application gets slower, of course, but it keeps on running-- demonstrating the independence of our software from the number of processors.
    You have said in the past that you are putting about a quarter of your resources into hardware development. Will that remain true in the future, or do you want to get out of the hardware business?
    As the availability of the BeOS on the Power Computing machines ramps up, we will see how the demand for our hardware goes and will let the market place decide. If we do a good job, I expect there will be agreements with other Macintosh clone makers. We had dinner with a number of clone makers at Macworld who are very supportive of our products. The clone industry will ship a bit under a million units this year, and probably significantly over that in 1998. That, in and of itself, constitutes a fairly nice installed base.
    Will you place more emphasis on the BeBox than the Mac?
    We cannot hope to succeed if we are not anything but supportive of the Mac. We have no intention of playing proprietary games.
    Looking at the Usenet newsgroup on Be, it looks like most participants are associated with universities. Does that reflect who has come to your booth and developer sessions?
    Actually, not quite. Our most recent developer conference was attended mostly by multimedia developers. But I'm happy to see interest in universities, because this is where a lot of creative ideas originate. We also see a lot of interest in northern Europe and in Japan.
    Are you making a special effort to port the OS to Japan?
    Yes. We are committed to providing 16-bit character representation and producing a Japanese version of our OS.
    e market in Japan as any different than in the U.S.?
    I see it as a big, friendly, multimedia market--but one that must be respected. A lot of western companies take the Japanese customer for granted, and I want to make sure we are very mindful of the requirements of the Japanese consumer. We will go slow there, because we want to avoid any false steps.
    I'm a Japanophile--even before I came to the U.S. The French culture has long had a good connection with the Japanese culture. When in Japan, I see the large number of young people reading computer magazine, and I think there's a lot of interest in creative multimedia applications in Japan. I hope that *if* we do a good job, we will be welcome there.
    What would be the difference between a Be multimedia application and one running on another OS?
    The differences are speed, agility, and robustness.
    The speed difference is shockingly apparent when you run the same application on the BeOS and Mac OS side-by-side. You readily see the amount of power that is smothered by the layers and layers of software silt that inevitably accrue with an aging operating system.
    The second difference is that a multi-threaded system makes it very easy to have applications that run without interfering without each other. With the MacOS, if you click on the pulldown menu, your QuickTime playback stops. And Mac OS is not memory protected--if an application fails, it usually takes the entire system down with it.
    As a multithreaded OS, the BeOS for example would enable you to take two video cameras and paint, in realtime, the videos on a rotating cube. That's not something you can hope to do with the Mac--or with a UNIX system running on low-end hardware. You can do that of course, with an expensive Silicon Graphics machine.
    Are you implying that you will compete with SGI?
    One of our investors wanted to pull my leg by suggesting we are "the poor man's Silicon Graphics." I didn't like that characterization at first, but really, Silicon Graphics is a very good reference for high-end multimedia. If we can offer similar capabilities at a much lower price, with much more widely available hardware--why not?
  • おすすめ記事