IBM's Open Letter to Sun
The OSS is growing up and becoming hungrier, they now demand Java to become open source. The question is, can a non organization like the OSS movement handle standards? I fail to believe this ...
Today there are two reasons for a developer to work on open source. First is the itch, where a developer has a need. This was the main driving force behind initial Linux and many other projects. These projects tend to solve the direct problem but often do not get the finishing touch. Usability is often low and too many projects are stalled in an undefined state. Sourceforge is looking more and more like the aftermath of a first world war battlefield. Even Apache, with their hot projects like the Apache and Tomcat web servers have too many projects that seem to be undirected.
From an itching developer point of view, this is not a problem. There are lots of interesting snippets and gems to be found in these repositories, and for the technically inclined, very useful applications. However, to cross the chasm to the end user, this is clearly not going to make it. Raymond's appeal to the OSS community will be in vain because making their projects work for aunt Tillie is not where there itch is, the developers probably can't stand the company of aunt Tillie!
The real interesting part of the OSS movement is when it is supported by real money. Succesful OSS projects are invariably backed by commercial companies. Good examples are Eclipse that is run by IBM and Linux which is heavily sponsored by the the bubble investors on NASDAQ (thank you!) through Red Hat and others.
Money is a requirement to build real products, there is just too much tedious work to do to make it just fun. From a business point of view, OSS can make sense because it allows companies to focus their energy on their core business instead of developing the same horizontal software as everybody else. It is easier to stand on the shoulders of giants if you want to look further. SCO's claim that it OSS threatens a $280 billion market is completely ridicuoulous because it assumes that the market is finite and static. Money saved through OSS, will be spent on fullfilling more specific needs and innovation in other areas (not necessarily software though). Trying to hold down progress through locking mechanisms, as SCO seems to want, is in the end a threat for the society, like having a fireman (the one shoveling the coal in steam engine) on electric trains in England during the eighties. The good point of commercially supported OSS is that it will muster innovation in different areas because much of the mundane work can be shared.
Control is an issue, however, with commercial players. For examples, looking at Eclipse, it is clear that this is an IBM project, just like Netbeans and JCP are SUN projects. These companies have an obligation to their shareholders to make a profit on their ventures. It is obvious, and seems fair, that they use these OSS projects for their own profit and are not filantropic. As long as this game is played around implementations I see nothing wrong with it. If I use Netbeans, I can switch to Eclipse without too much pain. The game becomes very different when it starts to involve standards. Standards are different than implementations. Standards are much more difficult to handle through a vague movement like the OSS community. Standards require clearly defined organizations that evolve the standards through an open, well defined, process. A process that can balance the interests of different parties. Standards also require careful evolution and backward compatibility, something that is, to use an understatement, not the strength of OSS.
Take for example Java. What would happen if it was placed on Sourceforge? Who would be able to control the evolution of the language and its libraries? The value of Java is not in the class libraries, not in the VM and neither in the syntax. The value is that it is a de-facto standard that is evolving through a well defined process in the JCP. The Java de-facto standard allows me to write an application that can run on any platform supported by Java. Standards are useful because they enable cooperation.
Unfortunately, SUN could not resist the temptation to remain in control, the JCP is not a non-profit organization as their domain name seems to imply, but is a part of SUN. In the early Java days SUN had discussions with ECMA but in 1999 they withdrew their request.
However, asking SUN to move Java to Sourceforge is asking for much and should not be necessary. Asking SUN to make JCP an independent, not for profit organization is more realistic. Java became what is today because many companies, (not in the least IBM, Motorola, BEA, and academics), donated significant efforts toward it. Many organizations (including my own) comitted to Java because of the expectation that it was an alternative to the monopolistic practices of Microsoft. SUN's past record with the Network File System (NFS) gave that idea legitimacy. With NFS they made an open standard and were willing to compete on better products. If, however, SUN wants to be a Microsoft clone, then I'd rather be bitten by the real thing. Microsoft's leverage is much higher and they have much better developer support.
Concluding, I do not think SUN should give Java to the OSS because the OSS should be perfectly capable of building one themselves. What we need is an open standard so the OSS can genuinly call it Java. I do think, SUN should move the evolution of the Java environment to an open organization where the environment can evolve in an open and democratic process.
Peter Kriens
posted by Peter @ Friday, February 27, 2004
