xml-dev - Widening Circles

[Jonathan.Robie@SoftwareAG-USA.com]

Title: Widening Circles

When I have an idea, I have to work it out for myself to the point that I have some sample code, a few diagrams, and can convince myself that I think it makes sense. No point in getting feedback until it makes sense to me, and I'm busy putting together a proposal I can share with a few people.

Then I share it with one or two people for whom I have deep respect, and with whom I think I can work well over the long haul, since these people are likely to be my close partners for the lifetime of the idea. I always prefer people who think differently from me, and bring their own ideas to bear. No point in getting feedback from a wider audience until I understand what these people would say, and I'm busy interacting with them and integrating their suggestions.

Then I start looking for a wider audience, perhaps six or seven people who I think are influential and knowledgeable. No point in getting feedback from a wider audience until I understand what these people would say, and I'm busy interacting with them and integrating their suggestions.

If I am in a Working Group, this is the time that I submit a proposal to the Working Group. In every large group I have been involved in, whether an informal mailing list or the W3C, a small number of people, certainly less than ten, offer most of the useful opinions, and I again find myself involved interacting with a small group of people.

The Working Group then takes time to come to consensus that the proposal is good (or dismisses it). At that point, we have something we can publish.

In my experience, most work gets done by small groups of people. This inherently leaves most people out of most of the process. A small group will emerge in any setting, whether a public mailing list, an industry consortium, a formal standards body, or some other kind of group. I like the fact that we have several different kinds of groups using different processes - W3C, IETF, ISO, OASIS - this allows us to gain experience with different approaches, and see how well they work to (1) create de facto or de juris standards, and (2) create industry buy-in. But I suspect that, in each of these settings, a few people will do most of the work, and their work will largely determine the character of the result. In fact, I think this is probably a Very Good Thing, because the more diffuse the design work becomes, the more likely it is to become a Design By Committee.

I like the W3C requirement to publicly publish Working Drafts and solicit feedback every three months, but I would hate to require more interaction with the public than this, because there is simply way too much work to do, interaction takes time, and the best designs are often created by a small number of people. Naturally, public review is important, but it should be sought at appropriate times. Every three months is a good number for me.

Jonathan