Process2018
Process document reference
The official W3C Process is documented here:
The current editor's draft is here:
View a diff between current Process and the editor's draft:
Issues Summary
The list of issues that were considered or are being worked on for Process2018 are tagged in Github with the label process2018candidate (some of them are now marked closed, as the Process CG believes it has completed action on them).
This is a mix of substantive and editorial issues. Editorial issues, and editor's actions to improve editorial quality outside the issues, are not detailed here. A summary of the motivation and action taken for substantive issues follows.
The follow-on questions for the few issues that have them are separately listed after the summary of changes.
Completely or Partially Addressed issues
#5 improved errata management
This concerns the vexing question of the maintenance of specifications no longer associated with a Working Group. The entire process and patent policy assumes a Working Group behind a specification. What happens when we realize that actual field implementations don't match the Recommendation, or we get report of an obvious bug in the Recommendation, but there is no associated actively-chartered Working Group?
We introduce a new term and process: Amended Recommendation.
As questions of practice, we will seek a contributor license grant (e.g. via the member Submission process or other appropriate channel) when bug fixes and minor changes are made to a Recommendation without an active working group. Such an Amended Recommendation will be published in a way that clearly identifies the Recommendation that was amended, and what changes were made. We will also try not to get into this state in future, and ensure that active Recommendations (implemented and in use) have a maintenance Working Group home, even if that Working Group is mostly dormant and only addressing bugs and error corrections.
The process says that these will progress, shepherded by the team, following the normal progression for a Candidate Recommendation to Recommendation, i.e. wide review requests, AC review, and so on.
#13 What is a "Memorandum of Understanding”?
We realized that we used the term of art in legal circles "Memorandum of Understanding", without always meaning what that implies; we inserted a clarifying definition of what we mean by it in this context (no change in process).
#30 Add reference to Code of Ethics and Professional Conduct (CEPC) in the Process Document
We inserted a requirement on all participants in W3C to abide by the Code of Ethics and Professional Conduct.
#23 Director can dismiss a AB or TAG participant without giving a cause?
The circumstances under which an AB or TAG became vacated were not clear, nor was it clearly stated the circumstances under which the Director can remove someone from any group. We clarified that the Director may remove anyone in any activity "for cause", where violations of the Membership Agreement, the Process (and hence the Code of Ethics and Professional Conduct) and applicable laws are included in the possible causes. ('For cause' means that the Director has to give specific justification.)
#36 Do we need a superseded status for older specs?
We realized that the process defined last year for Obsolete Recommendations (ones we no longer 'recommend' for new adoption) could also be used for superseded specifications, and these edits make that clear. The process (AC review etc.) is identical; only the labeling changes. We also made editorial changes here: #57 Definition of obsolete
#33 Re-craft and re-insert the 60-day delay between charter review and Call for Participation
In the Process 2017 work PSIG had requested a sentence that gave an automatic minimum 60-day ensured interval between a call for charter review and a call for participation. The AC ultimately removed the PSIG’s proposed language amidst concerns about operational problems. This process document has similar language, but with two important differences:
- the interval is only on-request; it's granted only if any AC representative asks for it (and we hope it will be 'rare'); we hope that this keeps the disruption to a minimum;
- it applies to any Charter with new deliverables, not just inherited deliverables. Legal departments are increasingly finding themselves scrambling to conduct a meaningful legal analysis for new charters and charters with new deliverables. Allowing the additional time between the call for review and the call for participation would enable more members to complete their patent analysis and participate from the beginning of a working group.
#34 Process2014 introduced an AC ballot for CR transitions and now we have a ballot open for a year
In 2014 the process was revised such that the Proposed Recommendation ballot was opened when Candidate Recommendation was passed, so that AC Representatives could raise concerns or objections much earlier in the process than the PR transition (when it's rather late to make changes). Unfortunately, chairs find it hard to estimate when the ballot should close, and at the time of posing (at CR) the question "should this proceed to Recommendation?" is not answerable as the specification may still have issues, may lack implementations, and so on. Some on the AC find these long ballots confusing, and some prefer that the list of open ballots represent action items that they should be acting on.
We are revising the process so that the formal ballot will happen, as before, at PR. However, notification of proposed CR transitions will be sent to the AC so that any early objections or concerns can be raised at that time.
#48 The appeals process doesn't actually describe how an appeal finishes
We have just enjoyed the first-ever appeal at the W3C. We realized that though the process defines how an appeal is started, it does not define how it is conducted. This change matches what was presumed for the the appeal that has just happened: ask the question of AC Representatives whether they support or reject a specific Decision of the Director, and assess the result by simple majority.
#52 Changes to TAG makeup
The Advisory Board and Technical Architecture Group realized that the process defined that the Director is normally Chair of the TAG, but recently there have been several co-chairs. The process also asked that Director appointments to the TAG precede elections, but we'd like to allow the Director to balance the TAG by appointing after election. This change adds an elected seat and removes the statement that the Director is normally chair. It also grants Tim Berners-Lee a seat on the TAG for life, even if he were to step down as Director.
This edit also makes clear that the TAG and AB have a Team Contact.
The TAG was involved in the discussion on this change.
#62 Can the AB/TAG chair ask for a special election in advance of a known upcoming vacancy?
In some cases (e.g. someone is retiring, or their employer will not be renewing membership) TAG or AB participants can state that they are resigning as of a known future date. The current process doesn't permit holding the election until after that date, forcing a gap in service. We now permit the election to be held in anticipation of a known future irrevocable resignation.
#84 Public Working Draft or Working Draft?
Since all Working Drafts are published on /TR and are thus publicly visible, the redundant word 'Public' was removed before 'Working Draft' except in the case of the well-known term First Public Working Draft.
Follow-on questions for community discussion
#48 The appeals process doesn't actually describe how an appeal finishes
There are questions that some on the process community group would like to have considered by the AB, AC, and community.
Can we address any of these immediately? Should we hold off any formal process change here until we have debated these? Should we adopt the 'vanilla' process and then look to refine? Should we let the dust of the recent appeal settle before we have this discussion? Guidance is sought from the community.
The questions that have been raised:
- Overturning a decision of the Director is a fairly major step. In some standards bodies and other voting bodies, such an over-ride would require more than a 50% simple majority. Should we have a threshold greater than 50% to over-ride a Director's decision (a so-called super-majority requirement)? (See #76.)
- Our experience is that very few AC Representatives attend face to face meetings or vote on ballots; we often struggle to meet the minimal 5% threshold that we ask for, for charter approval. A rejection of a Director's decision is a major step, and we might prefer that those voting are in discussion, up to speed on the question, on what an appeal in general means, and so on. We used to have 'good standing' requirements for working groups; should we establish voting criteria for Appeal votes? Other bodies have such rules. For example, in INCITS one has to attend 2 out of 3 meetings, and vote on ballots, or you go into jeopardy of losing voting rights; in jeopardy you must attend the next meeting or you lose voting rights, and only regain them after attending two out of three consecutive meetings. Various possibilities exist, including:
- If the question is one in a series (e.g. chartering a working group, starting a specification, and so on), have you voted on or expressed an opinion on those?
- Have you attended a recent AC meeting?
- Have you voted (even with an explicit abstain) on a decent proportion of AC ballots over the preceding N months (or since you joined, whichever is the shorter)?
- Should you be required to have 'standing' (in the legal sense) to raise an appeal question, i.e. are you or your organization directly affected by the decision? (This is a very hard test to administer and ties up the courts sometimes.)
- Should we put more explanation behind 'extraordinary' (the circumstances under which we expect an appeal, for example that there were facts not considered or people not consulted)?
- Should the appeal demonstrate 'grounds'? Should there be a requirement to show that there were some facts, evidence, people, etc. that were not considered or consulted in forming the decision? I.e. can we restrict appeals to "ill considered" decisions, and not allow re-consideration of a decision that people simply dislike?
- Is 5% the right threshold? The purpose of the threshold is probably to ensure that we don’t open an appeal for something which very few agree is an issue — enough to ensure it’s not a personal issue but one that has some agreement from ‘a reasonable proportion’ of the AC that there is a reason to hold the vote. It’s not clear where the 5% number came from or whether it’s right.
Some have asked why there is no tallying of opposition when the 5% is being sought; should the ‘request for appeal’ require not only passing the threshold, but require also that there is more support for an appeal than opposition to holding it? The answer would seem to be that we don’t want to hold the appeal vote twice. The threshold is there answering the question “are enough of the AC concerned, that we should open the question and have the debate?”, not “is the appeal likely to succeed?”.
If the community decides on a super-majority requirement, and its size, the resulting edit to the process is trivial; similarly a consensus to change the 5% threshold to a different number is trivial to implement.
Both defining voting criteria, and implementing the tracking of who is eligible to vote, would take substantial time.
Formal definitions of 'extraordinary', 'standing', and 'grounds' may be hard to make; and if they are formal criteria to meet, we'd have to decide who decides whether they have been met or not, and it's likely that exercising such judgment will be time-consuming, will be 'conservatively' judged (i.e. err on the side of allowing the appeal) and nonetheless cause ill-will (appellants will object to an appeal being blocked, opponents will object to an appeal not being blocked). It may be better to provide guidelines that other members of the community can use to evaluate 'is this a proper appeal?', when they decide to support the request for appeal or when voting in the appeal itself (for example, enable an AC Rep to say "I do not think you have established the requested grounds for an appeal").
Deferred issues
#63 Updates to sections 5.2.2 and 2.1.3.1
The team has been permitting AC Representatives to make their decision and comments on Charter reviews available on a public forum. This change requested formalizing that, but there was concern expressed in the process community group:
- The representative might quote, and thus make public, comments of another representative that were member-only;
- The ability to discuss in public might turn into a public debate of charters, with public statements, blogging, and so on.
The process currently neither requires nor forbids this, and the group currently feels that this should be discussed, and explored as a question of practice, and not formalized at this time.
#24 Sections 3.4 and 6.2.6 have different statements about Voting rules in a Charter
We tried to clarify this, and provide sensible defaults for votes so charters had a fallback they could override if they wished; but we accidentally changed the voting criteria. The edits to 3.4 have been reverted and we'll try again next year.
#60 Clarify the voting process
In Process 2017 we moved to single-transferable vote (STV) for TAG and AB elections. We left in the statement that AC Representatives had 'one vote per available seat', which was, at best, misleading. We have not removed this statement, as some think it a desirable principle, even though it does not correspond with the way that we trialed or use STV. In particular, STV rarely resorts to using the last alternatives indicated on voting forms (i.e. in a recent 4-seat election, it's not clear if any 4th-preference rankings had any effect on the outcome). The AB wants to see if equal-preference voting can be supported (it's not supported in existing open-source tabulation software), and others want to explore 'cascaded' STV voting (seat-at-a-time) or the like. The inconsistency remains and the issue remains open.
Some are concerned that the dynamics of STV voting are such that (put negatively) more fringe candidates are elected, or (put positively) a greater diversity is achieved.
Non-addressed issues
There are other issues that are raised, but for which discussion and resolution is not yet complete, and hence they currently expected to be handled (if at all) in a future revision.