Thinking the Unthinkable

If Red Hat Purchased Be: A Thought Experiment

Scot Hacker, 12/99

A few weeks ago, Bloomberg Financial published a story rumoring an imminent purchase of Be, Inc. by Red Hat Software. Although the rumors had still not been substantiated as of this writing, their effect on the community of BeOS and Linux users (not to mention Be's stock price) was immediate. Stories appeared simultaneously on BeNews and, where the respective user communities hashed out possible ramifications. Of course, everyone knew they were quibbling over rumors, but nevermind that. The prospect raises so many issues that people apparently couldn't help themselves.

Let's pretend -- in a purely hypothetical mode -- that Red Hat did buy Be. What would it mean for Be? What would it mean for Linux? What would it mean for users? The answer to those questions, of course, depend on knowing what, exactly, Red Hat planned to do with BeOS. I'll present two imaginary scenarios followed by imaginary repercussions, were those events to come to pass.

Scenario One: Red Hat Pursues Twin Markets

In this scenario, which I consider to be the most feasible of the two, Red Hat intends to do nothing with BeOS other than to simply own, distribute, and market it as a second business. Be and its engineering staff remain intact, while Red Hat gains the opportunity to diversify their coverage of the alt.os market. One can imagine how difficult it already must be to explain to potential Red Hat customers why they should choose Red Hat over other Linux distributions -- heck, even geeks have trouble adequately describing the differences between the many available distros. In other words, it's not easy for Red Hat to distinguish themselves from the competition.

With BeOS under their belt, Red Hat would be able to offer a product which stands little chance of being confused with Linux (except by the occassional clueless journalist.) More importantly, Red Hat would be able to approach two entirely different segments of the operating system market. Despite the concerted efforts of thousands of programmers to bring a server-optimized OS into line with the needs of desktop users, Linux is still leagues behind BeOS in terms of ease of installation, configuration, and use, not to mention Be's hallmark performance: fluid responsiveness even when under heavy media loads. Be is making inroads in the desktop space, and the age-old chant of many people interested by both systems remains "Linux on the server, Be on the desktop." For Red Hat to re-interpret that slogan as a business proposition would be a brilliant move for the company.

The source code to BeOS would stay closed, the Be staff would continue working on their directed, consistent vision of what an OS can be, and little would change. However, Be would get to piggy back on the fact that Red Hat is a household word. Be's technology, Red Hat's fame and marketing muscle. As long as Red Hat kept their hands off the technology itself, this would be an intriguing scenario -- at least from a business standpoint. As one poster pointed out in the Slashdot discussions (and I paraphrase), "Red Hat has a killer market but not much of a product, while Be has a killer product but not much of a market. Seems like a marriage made in heaven."

Scenario Two: Red Hat Merges the Two Products

Before you get any big ideas, let's be very clear about one thing: Be licenses a lot of technology from various companies, and therefore can't simply open BeOS to the world -- they would never be allowed to. In addition, the process of merging two totally alien codebases would be heinously complex -- probably harder than starting from scratch in many respects. For the purposes of this discussion then, we'll hypothesize that Red Hat purchases Be with the intention of opening the source to system technologies Be has engineered themselves. The intent, presumably, would be to either A) create a sort of BeOS/Linux hybrid combining some of the best features of each, or B) to somehow extract Be's most attractive technologies and incorporate them into a highly unusual Linux distribution. This is really slippery territory, but what the heck -- we're only pretending here.

It's worth noting that the two systems are both mostly POSIX-compliant, and both use ELF binaries on the x86 side, created via the gcc compiler. This does not mean that binaries created for one can be run on the other, but that we at least have some common foundations.

It should come as little surprise that BeOS and Linux users feel very differently about the prospect of code-base integration. At one extreme you have Linux users salivating at the prospect of being able to integrate some of Be's core technologies into Linux. At the other extreme are BeOS users feeling passionately that this would be the worst thing ever to happen to BeOS. There don't appear to be a whole lot of people with opinions situated between these extremes. Let's take a look at these two perspectives.

You Will Be Assimilated

There are thousands of existing Linux users who look upon BeOS admiringly, but whose religious beliefs include proscriptions against using closed source software. "I'll try it when and if Be opens the source code" is a common refrain in Slashdot discussions. Because BeOS is POSIX-compliant and includes a bash shell, it provides access to hundreds of open source command-line applications, most of which originate with GNU. There exists a Linux contingency that feels that Be isn't playing fair because it includes GNU tools without contributing source back into the pool. This position doesn't wash, because there are plenty of open source software efforts for BeOS (such as Apache) which do merge changes back into the source tree, and Be makes available source code to quite a few bundled applications, not to mention full header files and freely published APIs. Of course, BeOS source to the ported GNU tools is also available. In any case, the position of some open source advocates is that if Red Hat purchased Be and subsequently opened Be's source, Be would get what's coming to them. Some actually appear to feel morally entitled to source access.

Whether or not such a case can be made, there's another Linux subculture bent on world domination via Borg-like assimilation tactics (nevermind the irony of the fact that this self-same group complains of Borg-like tactics from Microsoft even while some Linux distros borrow or emulate increasing numbers of Windows and MacOS appearance, behavior, and functionality traits). Some members of this group actively advocate opening BeOS source code not for the sake of making BeOS better, but to pillage and plunder its treasures to improve Linux.

Without passing value judgements on that line of thinking, the fact remains that such assimilation would undoubtedly occurr if Be/RedHat were to open Be's source. It would not be preventable -- Linux would awkwardly assimilate the best of BeOS, and in so doing (assuming such a thing could be done successfully), eliminate any reason for BeOS to exist. In other words, the purchase would ultimately eliminate some competitive technology from the playing field. It would destroy that which it emulates. Sorry, but I just don't believe Red Hat operates at that level, even if some of the more extreme open source advocates do.

BeOS simply could not withstand the buffering and high winds of the open source world and still remain BeOS. Opening Be's source code would result in BeOS becoming something else -- something that's not BeOS and minus some of Be's key advantages. This may come as a shock to some people, but most members of the BeOS user and developer communities feel that Be's amazing technologies are only made possible by the fact that BeOS is closed source.

Bungle in the Jungle, Take II

Several years ago, BeOS users faced a similar paradox as Apple contemplated purchasing Be and merging its technology into the next version of MacOS. On one hand, BeOS would have gained instant notoriety, and MacOS might actually have become an OS worth taking seriously. On the other hand, Apple had a long history of bungling and emasculating key technologies and intiatives (remember, this was five years ago). Be users voted "no confidence" to the prospect of Apple being able to acquire BeOS without totally screwing it up.

But this problem wasn't and isn't Apple-specific. The point is that Be has a vision. They're a small, quick, tightly focused company concentrated on bringing high performance to low-cost computers without code bloat or the burden of dealing with backwards compatibility. Integrating BeOS into any system would destroy the vision, while tangling that ultra-clean API up into an outdated architecture (all other operating systems have outdated architectures, from where I sit).

The prospect of BeOS retaining its amazing responsiveness, the automatic multiprocessor awareness of its applications, its finely grained multithreadedness, its 64-bit, fully journaled, database-like filesystem, its elegant integration of a simple, elegant GUI with the power of Unix, etc. etc... All of the stuff that makes BeOS great is predicated on the maintenance of the vision of lightweight, modern power. It's difficult -- if not impossible -- for me to imagine that vision being retained through the violent uprooting of the operating system from its home turf and into the context of an entirely different system, under the control of a much larger company without the long history of the Be vision behind it.

Never Say Never

Even if a hybrid BeOS/Linux system were technically feasible, which it isn't, I wouldn't trust anyone but Be with the crown jewels. But as impossible as it is for me to imagine scenario #2 above, mergers and acquisitions take place in this industry practically every day. Stranger things have happened. What makes sense for technology doesn't always make sense for business, and vice versa. Who knows what strange turns the next few years will hold, how the market will respond, or what operating system the technoscenti will be using ten years from now. All I know is that the Red Hat speculation so far seems to be idle, and that if Red Hat did purchase Be, it would almost certainly have to be for marketing and distribution reasons, not technology reasons.

My Final Column

By now, you've hopefully heard that SSC, publishers of the esteemed Linux Journal, will be begin publishing the printed Be Magazine in April, 2000 (keep an eye on BeNews for subscription details). Because I have been selected as Editor-in-Chief of Be Magazine, I must regretfully give up my position as a Byte columnist due to conflicts of interest. The good news is, The Be View will remain a monthly feature in this space, though it will be written by someone else.

It's been a pleasure writing about BeOS for CMP, and I'd like to thank them for hosting this column. Meanwhile, keep an eye on the O'Reilly and Associates web site for availability of my latest book "MP3: The Definitive Guide," to be published in February. And of course, The BeOS Bible is still available from Peachpit Press.

Thanks for reading!

BeView Content Archives