There have been a lot of rumors about Be Inc., makers of the BeBox and BeOS, over the last few months, and I have been very skeptical of them. When I would read that BeOS is revolutionary because it has protected memory or preemptive multitasking, or that the BeBox is based on the PowerPC microprocessor and a PCI bus, I would always ask myself "Hasn't this been done already?" I was desperately in need of a demo and someone to answer my questions about why their product was so wonderful. And I got it.
BeBox with monitor
Be, Inc. "Evangelist" Scott Paterson came to Columbia University on December 4 packing a BeBox and a wealth of information to help muster support for the fledgling computer maker. Presenting to packed auditorium, he answered all of my questions, and then some.
Radical design philosophy
The Be design philosophy is brilliant. Recognizing that current personal computers are plagued by backward compatibility problems on the operating system side, and an overabundance of hardware standards and scalability constraints on the physical side, they set out to make a complete package to solve these problems. They would make a box that was only based on the most promising and the most powerful technologies, and they would create a modern operating system from the ground up that can take full advantage of the state of the art in computer science.
So the BeBox is based on the PowerPC family of CPUs, with a PCI bus, SCSI controller, and so on. Nothing more, nothing less. The result is a hardware design straightforward enough that they admit could be created by any good engineer, but for some reason doesn't exist on the market today. With a standard hardware base, they knew exactly how to build a powerful operating system around it. With no legacy software to deal with, they created a powerful 32 bit operating system with preemptive multitasking, symmetric multiprocessing, protected memory, POSIX compliance, and put it into about four megabytes of RAM. And they did it well.
The only remarkable feature of the BeBox itself, other than its clean design philosophy, is its multiprocessor motherboard. Currently sporting dual CPU motherboards, Be is soon going to launch quad CPU and ultimately eight CPU motherboards, beyond which there are some hardware limitations. The BeBox also comes with a nice slew of features available on other machines, just not standard. It comes with an ethernet card, built in MIDI support, a slew of serial and parallel ports, two game ports, and a proprietary port called the "GeekPort". The GeekPort is a mish mosh of all the simple things you ever wanted: easily accessible I/O lines, digital data acquisition hardware, etc.
Back panel of a BeBox is rich in connectors
The operating system from hell
But the truly exceptional product is BeOS, the operating system that comes with the BeBox and also performs admirably on Power Macintosh computers. With no operating system or application legacy to support, Be was able to stack BeOS to the hilt with features, and still keep it tight and fast. In the demo, it handily played a slew of QuickTime movies and audio files, and ran an interactive Live3D application, while maintaining extremely smooth performance and stunning interactivity. But this was on a dual 133MHz PowerPC 603 with 32MB of RAM. When they ran the Power Macintosh port of BeOS on an off the shelf Power Computing 120MHz PowerPC they blew me away. An A/B comparison of MacOS and BeOS on the same machine was truly a joke. But then again, and A/B comparison of MacOS and any other operating system written since 1990 is truly a joke. In all honesty though, the performance was unlike anything I imagined, but what about the features?
Symmetric multiprocessing speeds things up
One important scalability feature is the inherent support of symmetric multiprocessing. On the dual CPU demo machine, the "Evangelist" giving the presentation demonstrated the effect of disabling
and enabling the second CPU while running a computationally intensive application, on the fly! While users may never want to disable any of their processors, this clearly demonstrates how robust the multiple CPU support is in the operating system.
BeOS being put through the paces
The BeOS is also obviously optimized for the demanding task of shuffling around real time audio streams. You will just have to trust me on how awesome the demo was - I cannot do justice to it here. The window manager is also very well designed. For example it doesn't rely on applications to refresh their own windows, leading to guaranteed screen refreshes that must be seen to be believed. Imagine dragging an opaque window with a playing movie over two other playing movies with zero visible artifacts or missed frames.
New OS features give sweet relief
One nice improvement over standard operating systems is the inclusion of a system wide database. Among other things, it keeps track of all files on disk, allowing for blindingly fast file searches and the like. Having a standard database also provides a means for applications to share information gracefully. For example, if you change your boss's contact information in your personal information manager, your email client will know about it as well. This kind of information sharing between multiple software packages is almost impossible if the operating system doesn't guarantee a common ground where they can all talk.
BeOS delivers so many other features that users have been begging for that I will not be able to exhaust them here. One nice feature is dynamically loaded device drivers. No more rebooting your machine after installing new drivers. Just turn them on and they start working, and will unload on the fly automatically as well. It also provides direct, yet still protected, access to the video buffer for game developers, so you can expect to see some hard hitting action games on the market in the next year.
Great! But can they market it?
As amazing as the Be product is (and reasonably priced to boot), I fear that it is destined to a fate similar to those of Amiga and NeXT. The main reason for this unfortunate conclusion is their target audience and marketing philosophy. Be is strongly focused on the digital media market, and does not plan on competing for the home and office markets, currently dominated by Microsoft, at all. Aiming to edge into the digital content creation and web development niches will work against them in several ways. First, it obviously restricts the size of the market. Second they cannot perform nearly as well in that niche market as they hope to, for several reasons.
The media niche will not work for Be
Scott Paterson, our "Evangelist" for the evening, told us that the 1996 PC market is about 68 million computers, and that the niche they are aiming for is about 3 million computers. Of course, they do not expect to kill the market, and have set a reasonable goal of about 600 thousand computers. This sounds good for a growing 45 person company. It is huge compared to where they are now, but seems modest enough to be attainable. However, companies tend not to last too long if they do not offer sustained growth, and starting off in a niche market does not leave much room for a good growth plan in the future.
In addition, I do not think they will even be able to break into this niche market, for a few very good reasons. These include everything from timing, to software legacy, to a fundamental flaw in their product targeting.
Traditionally, Macintosh has had a corner on the media niche, and that is starting to change. There is currently a market trend where people in the area are looking to move away from Macintosh. That would be great news for Be, if they had a complete product right now. Unfortunately, it will be at least another year before their package is complete enough to pose a market threat, and that leaves a lot of time for disgruntled Mac users to buy PCs. If they were a little bit ahead of where they are now they would be much better poised to break into the market.
Lack of media software legacy will haunt them
On top of that, software legacy will be a problem. This will be frustrating for Be, because what allows them to be so great is breaking away from software legacy problems. Be is actively recruiting third parties to port their applications to BeOS. The first one everybody asks for is Adobe Photoshop, which is indeed on its way. However, a media designer needs more than Photoshop. The production of The Moment alone has required the use of Photoshop, Macromedia Freehand, Adobe PageMaker, Microsoft Front Page, and many others. A professional designer needs countless programs to function. Even if 9 out of 10 are ported to BeOS, that's not good enough. Take one piece out of the puzzle and it falls apart. If a new project comes up and you suddenly need to burn a CD, will Corel CD Creator be available?
Even a simple thing like Internet Explorer is crucial to web development. After design, the production of a web page requires incremental testing in all significant browsers, currently Navigator and Internet Explorer. If you only test in one (whichever one), it is almost guaranteed not to work in the other when you are done. Once you reach that stage you are sunk. In short, Internet content needs to be developed on the target platform to be successful.
They do not offer the market anything it really needs
Finally, if their target really is the media production niche, there are several flaws in their perception of what the market really needs. In terms of stability, the problem that I am faced with right now is not that my operating system crashes a lot, it is that Freehand 7 crashes a lot, and only Macromedia can fix that. People in the media market do not need an operating system any more stable than Windows NT, even Windows 95 is stable enough for this market to be honest.
In terms of speed, the end user needs the power to play multiple media streams without glitches more than the content developer does. It is not that important to be able to play four movies simultaneously during development, because people usually develop them one at a time. In fact, smooth performance may be more mission critical on the target computer than it is on the production computer. During production, the designer has the luxury of only working on one media object at any instant in time, waiting half a second for a screen redraw, or performing calculations that can't quite be done in real time. The viewer has none of these luxuries.
Of course, high end media production is calculation intensive and deals with large data streams, so a fast computer with optimized data paths is preferable, but it is not the be all and end all. If anything, it is the final display of the media product that is most critical, because it must occur in real time. If the product is to be mass distributed, this means that the target will most likely be a "Wintel" machine, a Mac, or some Unix derivative, not a BeBox.
Also, Internet content development, a large portion of Be's targeted market, doesn't even require that much power. To render 72 dpi images, edit five megabyte video clips, and make an animated GIF, current technology already works fine. Again, in this field it is much more important to have the robust software support that comes with a conventional PC than it is to have the extra speed that comes with a BeBox.
Be also expects that a small portion of the PC video gaming market will be captured, but this is almost a complete fallacy because there will be almost no major video games for BeOS. Video game retailing for the PC is a market based on extremely thin profit margins and large market fluctuations. A developer needs to know that his game can reach as large an audience or possible. Hence, most computer games are made for Microsoft operating systems or for Macintoshes. Be only hopes to sell a few hundred thousand units, and those mostly to professional content developers. If there are already tens of millions of Microsoft and Apple machines in the home, why would anyone go to the trouble and expense of developing a major game for BeOS?
Hope vs. Destiny
I make these predictions with great sadness, because I, like many other computer users, would simply love to have a BeBox on my desk, assuming it could run all the software I need. I wish them the best of luck. The BeBox and the BeOS combine to make the most robust and powerful PC ever, period. That is not even up for discussion.
However, despite all the buzz about Be, I fear that they are destined to become the subject of a great many stories similar to those surrounding the Beta video format and OS/2.