Can Fragmentation of Open Source Still Work?
Dana Blankenhorn has posted an interesting article discussing how open source software is being divvied up among top software companies (IBM, Google, Microsoft) in a way reminiscent of Japanese keiretsus. She argues that only these three seem to have the “size, scope and ambition” to play in this space, though Sun also continues to seek such a dominant position.
According to the article, in American usage “keiretsu” has become a term describing a much looser form of business association, with one large company and a number of smaller ones beholden to it in various ways. The Mozilla Foundation’s dependence on Google would be a good example, a relationship of substantial ties between independent entities.
Years ago, I was thinking open source might end up being its own “keiretsu”. Nevertheless, I guess it was inevitable that it would instead end up fragmented and mostly beholden to big companies. Blankenhorn cites the examples of IBM-Red Hat and Microsoft-Novell as other instances of such ties.
On the other hand, if these big companies do things right, open source software can advance and still end up producing a thriving ecosystem. IBM’s relationship with the Eclipse Foundation is a prime example.
It seems that key to the whole process is how a Big Co. views the software product lifecycle. If it accepts that functionality gradually will become commoditized, it will view open source as the likely end-point for most proprietary software. Such a strategy/outlook will result in the Big Co.’s typically releasing the source after some years/decades, with the intent of building still-proprietary software and services on top of it.
IBM did that with VisualAge (now Eclipse), and has also worked hard at promoting the open-source Linux operating system. On the other hand, Microsoft seems to want to milk its Windows operating system forever, which makes it hard to play well in the open-source world.
While it may seem wasteful, companies such as Apple have shown that a steady discarding of old technology can do a lot to promote innovation. I keep waiting for more companies to follow their example.
Evidence continues to mount that “creative destruction” is indeed a key process in a healthy economy. Peter Drucker argued that companies ought to continually make way for the new by killing off old products, rather than waiting for the market to do it for them. Henry Ford’s reluctance to part with the Model T, and its nearly destroying Ford in the process, is the classic case study on this.
With software, however, there’s a difference, since many software products are foundations for other systems (software or hardware), or integral to the use/retention of valuable data produced by them. For this reason, it’s not as simple a matter to remove a program from the market. Software may be used for many different purposes by different customers. Some may be able to switch quickly to new products, but others would incur great cost.
Consequently, it seems inevitable that users are eventually going to demand some kind of protection from software vendors - or else from government regulators - that a software product’s source code be made open source if the product is abandoned. At the least users would be protected, and possibly the product might see further development by others (WordPress, successor to b2, is in some respects an example of this).
More interesting, however, are the more state-of-the-art open-source projects such as Linux and Eclipse, that promise innovation and a product that is “built right” for the future in a platform-agnostic way. Knowing that a software product will be here to stay, because the source is available, seems to be such a strong draw that many of these best-of-breed projects have been able to attract top talent to contribute, often on a volunteer basis, as well as substantial support from Big Co.’s such as IBM.
For the software industry to thrive and not just reinvent the wheel, we need strong and viable foundations to build on. If these Big Co.’s are willing to kill off the revenue streams from their old software somewhat before it dries up, their code may well retain importance, or even grow in dominance as Eclipse has. This can offer a strong foundation both for services and for additional products higher up the stack.
Moreover, it will produce an overall healthier software industry ecosystem, since the underlying code will continue to be developed, increasing the value of it and everything higher up on the stack, due both to stability and to greater innovation.

It makes you wonder who’s responsibility it is to push the envelope for advancement. Developers are bound to the limitations of the OS, but the OS is bound to keeping things status quo for developers of established software.
Could Microsoft get away with pulling a move like Apple did when the released OS X? Do they still have the technical prowess to even be able to pull it off?
Comment by Jeff Smith — November 13, 2007 @ 3:12 am
Yeah, Apple has been training its users for a long time to expect the train to leave the station. I figured for quite a while that they’d drop Classic, especially after they made it a user install in 10.4.
Now folks are figuring 10.6 (whatever kind of cat it’ll be) will drop PowerPC, but I read someone saying Apple’s said it will be 10.7. In any case, we all know it’s coming.
Microsoft has long made backward compatibility and cross-gradability from other products a key part of their strategy. This worked great as they grew, but Windows ended up with a lot of old junk in it. Just working around the 640K barrier lasted an awfully long time.
If Microsoft had gradually moved folks forward as Apple does, they wouldn’t have the messy codebase they must have (judging from how hard it was to bring out Vista). Personally, I think Apple moves too quickly. It would have been really helpful to have an AppleTalk port in the early iMacs, for example, but it’s better than not moving forward at all.
Actually, now I’m wondering if Apple’s going to pull the plug on Carbon in a few more OS releases, maybe 10.8??
Comment by Gordon R. Vaughan — November 13, 2007 @ 6:16 am