Stuart Landridge sums up the ongoing debate about the future of CSS. My thoughts on this are that we should:
Let browser makers run loose with implementing new features, based on feedback from web developers.
Have a standard body (or some other group) that look at those existing features and see how they can be standardized.
Have browsers change their implementation of the features to match what the standards body proclaims.
Step 1 stops the bad idea that is to have a committee of people do innovation work. You need people both that know what web developers want and what is possible to implement. Browser makers do this.
Step 2 makes sure the wild features that browser makers have played with get usable on real sites. To be usable a certain feature needs to be usable across (a majority of) browsers, and web developers of today know this. Before reaching this stage features will be played with, but not used on production sites.
Step 3. Browser makers having implemented a feature differently from what the spec suggests need to change their implementation to match that of the standard. This step means that the standards body need to have the trust of the browser makers.
For the features of a certain browser, this means that a feature will move from "experimental", when it's played with, and to "standardized", when there is a spec and the feature conforms to it. Browsers need to specify this somehow.
I believe this is a much saner process than what is going on right now. Let's try it.