jd's Journal: Closed standards

With the introduction of PCI-X 2.0, the world has yet another closed standard. Only individuals who are employed by paid-up member organizations of PCI-SIG are able to read that specification. What does this mean?

Well, the first thing it means is that it's a lot harder to get Open Source/Free software that takes advantage of the new hardware. After all, member organizations may well write Open Source/Free drivers, but how complete are they? Do they make best use of the capabilities? Are there capabilities that aren't implemented? We don't - and won't - know. There's no way to know.

PCI-X 2.0 is only the latest in a long line of such TecSpec Substance Abuse. ISO and CCITT (now ITU) are notorious for limiting access to specifications, Microsoft claims not to have any, and SCO just sues those who might have them.

As well as programming, I design. But I can't design if I have nothing to design with! Without a spec, I cannot do anything for or with closed standards for software or hardware.

It is no coincidence that there are no Open Source X.400 e-mail servers. X.400 is certainly a more comprehensive standard than SMTP, and it certainly supported attachments a long time before SMTP servers/clients did. It's a very powerful system. It's also a very dead system. Powerful is irrelevent, when nobody can use it. All the power in the world attains nothing, if it's locked away, out of reach of those who need or want it. People go for the alternatives.

Intel tried to lock up the specs for their processors and chipsets, so rival companies produced incompatiable versions. Those rival companies are bashing huge chunks out of Intel's market, and Intel can't claim that back -- their system is (by definition) equally incompatiable with those of their rivals.

The danger for PCI-X is this -- if a rival standard is developed, that's about as fast, preferably cheaper to implement and - above all - open, PCI-X will be as dead as PS/2. Which, incidently, was also a closed standard. And expensive.

Thus, I throw open a challange. There's no "prize", other than bragging rights (and the possibility of world domination). Sorry. The challange is to produce a fully-Open specification for a bus that competes with HyperTunnel and PCI-X on performance and capability, while being cheaper than either to implement.

It doesn't matter if this spec is never implemented. What matters is whether the mere possibility of a threat to the closed model can do for hardware what the IETF, Linux, *BSD, the FSF, et al, have done for software. Be present enough to terrify the dinosaurs, force them to move their mind-sets 250 million years forwards into the modern-day era.

Standards with out implementations?

[Sloppy]

It doesn't matter if this spec is never implemented. What matters is whether the mere possibility of a threat to the closed model can do for hardware what the IETF, Linux, *BSD, the FSF, et al, have done for software.

Of course it matters if it's implemented, because deployment of that implementation is what creates the threat to the closed model. Microsoft, for example, isn't scared of IETF standards; they are afraid of their customers using competing products that adhere to those standards.

What good i