TPAC/2011/Revisiting how W3C creates standards

From W3C Wiki

Chairs:

Kai Scheppe

Marcos Caceres


We had a very positive and active discussion that captured many issues that need addressing. These discussions will continue as part of the W3C Process Community Group.

The following are notes taken on flip chart.

Minutes are available (thanks Chaals!)

Problem introduction

There are a variety of problems with the standards creation process

Goals for the session

  • Identify these problems in a brainstorming session.
  • Solution identification had to be delegated to future efforts (see below)

Identified problems

General

  • specs are too large
  • the process is too long and too complex
  • process documents do not match the development model we use
  • process itself is fairly old fashioned as evidenced by ad hoc spec development elsewhere
  • stakeholders need stability
  • there is tension and conflict between needed stability and a dynamic environment

Early phase

  • ideas that the group might have might not fit the charter, but amending the charter takes too long
  • finding intersted parties to work on a given problem
  • getting input from necessary but uninterested parties
  • in the charter the deliverables are fixed, but not the true goals
  • a large scope may inhibit participation due to IPR concerns

Mid-process

  • the draft in TR space is continuously out dated
  • within a spec it is not possible to distinguish between different types of changes
  • suitable stability points are not clearly identifyable. Need to be metrics based (see also CSS WG)

Late phase

  • many reviews happen only at LC and not before
  • LC often is far too late in the process
  • LC often is not truly the last call
  • LC is in general overloaded with communication efforts, actual comments, prop. edited CR
  • LC contains intense and lengthy communication with other groups
  • CR phase is like a 2nd review phase, breaking the intent of LC
  • a single, large document should not be the only thing we publish

IPR

  • patent protection kicks in only after REC status has been achieved
  • LC is an IPR milestone that takes long to pass

Proposed solutions

  • In order to address these issues the suggestion was made to form a community group that can find solutions for the individual points
  • create small, clean, orthogonal specs