How Just is JCP?

Standardization is a strange beast. It is veiled in a aroma of do-good but in the end it is unfortunately mainly driven by greed and fear of the players. For standardization to be accepted by a large enough audience to become relevant it must be marketed as philanthropic effort by benign companies for the greater good of mankind. To get the resources to make a specification it requires the participation of companies that are either afraid that the specification disrupts their business (fear) or see an opportunity, where the opportunity can range from a chance to disrupt the competitor's business or a genuine business case (greed).

For Java, these battles mostly take place in the JCP. The marketing of the JCP is working fairly well because most people think the JCP is an independent organization (www.jcp.ORG) where people merrily collaborate to create great Java specs for the rest if us. In reality, it is more like a medieval war zone with a king (SUN) that plays its many dukes (specification leads) against each other to stay on top.

Ok, I agree I am negatively biased. I want to be on JSR 277 because with my 8 years of OSGi experience I think I should be there. Despite help of many people, I am denied access to the discussions for very vague reasons.

However, what upset me most about this rejection was the discovery how the JCP operates.

The JCP is not a not-for-profit organization as the name seems to apply. The JCP is run by the PMOwhich are SUN employees and are not organizationally shielded in any way. Despite the org in the URL, it is just a front for Sun Microsystems Inc, it is not even a subsidiary. That is, when you sign the JSPA, you sign it with Sun Microsystems Inc, not JCP.

One sign of the secrecy is the lack of communication between JCP participants. Surprisingly, there is no mailing list for JCP. Oh yeah, you can sign up for news from the JCP, but there is no way to discuss the JCP in any mailing list. Contributions are not appreciated. A couple of years ago I started to push the PMO for such a mailing list because it looked like a simple omission. After many mails I got some excuses that there were technical problems setting up a mailing list ... Sigh.

But there are the ECs? (Excuse Committees I'm told). The JCP is governed by an EC for the embedded space (J2ME) as well as an EC for the Enterprise space (J2EE). The members of the ECs (except SUN) are chosen by the participants of the JCP. However, once they are chosen they must keep the discussions in the EC secret. Rumors are also that the EC members have no rights to raise agenda points; only SUN can set the agenda. This may explain the long-winding additions to votes on the EG pages, these votes are basically the only way to voice opinions about what is happening inside the JCP to the outside world.

But anybody can start an EG and be a spec lead? Yes, and the way that model works is very interesting. As a spec lead you become a duke and get special favors and status from the king. In the area you are granted you are now in charge. Spec leads can choose their army all by themselves. Spec leads can also set the licensing conditions for the specification which in the end creates a disastrous problem for manufacturers. Implementing a device which implements many JSRs requires an uncanny number of negotiations with all the spec lead companies; not to speak of the cost in running all the different TCKs.

Come on! There is a public review cycle? Yes, but this cycle is a one way street. There are no mailing lists where public specs can be discussed, nor is the feedback made available for a wider audience. What the spec lead decides to do with the feedback, is up to the spec lead. Also, by giving feedback you must hand over any patents you might have in the given area to the spec lead and SUN.

So what is good about the JCP? Well, grudgingly I must admit that it seems to work. There is a lot of activity (albeit many JSRs have been dead for a long time) and many JSRs have decent quality. So should it change? Yes, I think it is in the interest of the industry to make the process more open. Java has become a major force in the computing industry because it was at the right place at the right time. It has been furthered by the efforts of thousands of companies and the commitment to use it from millions of companies, universities and individuals. It has become an intrinsic part of the fabric of omputing. It is just plain wrong that a single company can exert full control over something that is so important to the community under a cloak of secrecy.

Got any complaints on the JCP? Want to defend it? Let me know!

posted by Peter @ Friday, September 16, 2005