• Add a Comment
  • Edit
  • More Actions v
  • Quarantine this Entry

Comments (8)

1 localhost commented Permalink

Good post Bob. I've been reading an interesting book by David Megginson called "Imperfect XML" that has a great chapter on standards - advantages and disadvantages of standards; the different standards bodies, de facto vs. de jure standards - you'd enjoy it, and so might your readers.Looking forward to your insight on standards!

2 localhost commented Permalink

Congratulations on a new position Bob. I believe that now you have a unique opportunity to define when standards can be a hindrance to an innovation pipeline. Ultimately a degree to which a standard can overreach with its sphere of influence determines both a level of innovation and the level of competion in a technology formalized by the standard. As an example, consider VoiceXML. First I must warn about how difficult it must be for someone who is an outsider to the speech community to discern the subsets of the specification that are almost universally accepted practices from the remainder suitable for further research and innovation. However, if one accepts a premise that a standard might be trying to define subjects that are simply not understood enough to be cast in a stone of a standard, one can see how the faulty standard may start to dictate directions of development of these subjects in a way where all future efforts must be spent on fixing areas of the standard that in really cannot be fixed without starting from ground up. This issue really goes back to the question of vendor/de facto standards vs the open standards. An open standard with an overreaching sphere of influence is actually _worse_ than an equivalent standard from a vendor because it can declare any competing attempts to improve upon its decisions as "non-standard". In contrast, a de facto standard faced with a more elegant solution coming for either an open standard or another vendor will have minimal chances of survival. Witness the IE vs Firefox battle playing out right now.So please, tell us how we can put a stop to frivolous open standards.

3 localhost commented Permalink

I think standards can be beneficial, but there seems to be a mania to standardize everything prematurely. Good standards seem to emerge from the "de facto" camp. Competing technologies gain or lose strength as users find out what works and what does not. New features emerge as users request them, or as competitors innovate and steal from each other. Eventually, a standard is born. Most of the TCP/IP protocols emerged this way, warts and all.Bad standards seem to come from standards bodies, where a committee representing competing companies decides to sit down and codify the solution to a problem. Each representative tries to insert things that give his/her company a competitive advantage, while the other members try to prevent such insertions. The committee tries to build consensus by adding obscure features to satisfy dissenters. The result typically is a mish-mash of compromises that over-complicates the solution in a way that satisfies no one. Examples include XML Schema and CORBA.Even worse standards emerge when a single company sits down and writes a "standard" promoting its proprietary tachnology. Witness Enterprise JavaBeans and SNA.I also look forward to reading your thoughts on standards.

4 localhost commented Permalink

Jim,Are you arguing that good and bad standards only come about in the ways you mentioned? For any given standard, you might say "look we can simplify it to be X". I might say "look, we can simpify it to be Y." Who says whether either one of us is right? What we each refuse to support the other's exact formulation? Is it ok to live with the two versions if we sacrifice interoperability (that's the WHOLE point)? Should we settle it via fistfight? (I don't think so.) So let's say we compromise and produce XY. That fellow over there with his version W thinks we're both fools because we didn't produce the simplest possible thing. And so on and so on.This points to why WS-I is so important. It basically says: we'll produce the best possible standards in the orgs that we can, given compromise and negotiation. Then after a while we'll revisit them based on actual industry experience and say what subset or common practices we should. Ideally that becomes what most people use and it influences future versions of the standard. All of the standards creating methods have their pros and cons. Those depend on things like the subject, the geography, laws, and the participants. I can think of several efforts that would have turned out differently (and better, to my mind) if certain particants were not present. We need to be practical. Your thoughts?Bob Sutor

5 localhost commented Permalink

Bob, the mission statement from WS-I "we'll produce the best possible standards in the orgs that we can, given compromise and negotiation" is a fine goal but the industry needs to improve on that by having a well defined set of metrics that help judge if something needs to be standardized in the first place.I'm in agreement with Jim that standardizing prematurely is what leads to a "bad" standard.

6 localhost commented Permalink

If you wait too long, how do you dislodge de facto standards if others prefer a more unified industry approach? I expect that "timing is everything" but I'm not sure how we ultimately get to a unified approach if we let too many flowers bloom, as they say. I'm trying to play devil's advocate here. This whole standards creation process is very complicated and fraught with peril.Bob

7 localhost commented Permalink

Bob,I am arguing many standards are written in committee meetings before anyone has a chance to figure out how real people will use them, and how they will work in practice.If I say "standardize on X", and you say "standardize on Y", that is a prime indicator the standard is not ready for prime time. Let the market sort it out before we standardize on anything. If a standard is bad, it can create more problems than it solves. You end up with a standard "XY" that everyone hates and only uses until a better alternative emerges.Look at XML Schema. This is a complicated and horribly-designed mess. RelaxNG is starting to take hold as an alternative. Not only is it simpler, it is a functional superset of Schema.If WS-I is willing to revisit and simplify standards, that sounds good. I don't think that is politically possible, though. Once a company has paid a large sum to join a standards body, how do you simplify them out of a standard?

8 localhost commented Permalink

In regard to waiting too long before dislodging de facto standards, I think history has been favorable to the open standards although some battles take longer than others.I agree that trying to "time" creation of a standard is futile. A position of "Oh, we'll just see when the technology from the vendors becomes mature" will not give potential customers of a young industry the peace of mind they expect for their investment. Its only rational for them to want to know what will happen with a technology two to five years down the road before they fork over their cash.That's the question that the standards bodies need to address. I suspect the existing standards bodies need to be enhanced to define not just the processes of making standards (design, draft, internal review, public review, etc) but a sequence in which various aspects of a technology are formalized. The sequence should focus on ensuring that an investor in a technology has both a clear roadmap and a back out plan. One possibility is that such a sequence begins with data exchange standards and culminates in a full featured programming model.

Add a Comment Add a Comment