Standards Best Practices Blog

I should make a clarification. Yesterday I mentioned a blog article from Andy Updegrove quoting the South African position on PAS submissions. This position was actually published last February, and likely doesn’t have anything to do with the current OOXML issue, as OOXML was not submitted by Ecma to JTC1 under the PAS process but rather through their Class A Liaison; Ecma is not a PAS Submitter (see the list here). The South African position is still good, I believe; perhaps they could say the same thing about submissions under the liaison process.

Members of ISO committees, including JTC1, the joint committee with IEC, are national bodies such as ANSI, BSI, DIN, AFNOR, etc. JTC1 has created two different means by which an organization other than a national body can submit completed work for approval as an international standard. One is the Fast Track Processing allowed for Class A Liaison members, and the other the PAS (Publicly Available Specification) Transposition process. Both are described in the ISO/IEC JTC1 Directives (their process document), document N8122, sections 13 and 14, respectively. The document is available at www.jtc1.org.

Before submitting work to JTC1, the organization must become certified as a Class A Liaison or PAS Submitter by making an application to JTC1. Qualifications to become a Class A Liaison are described in N8122 Section 3.3.4.2 as

Organisations which make an effective contribution to and participate actively in the work of JTC 1 or its SCs for most of the questions dealt with by the committee.

Then Section 13.1 says that

Any P-member of JTC 1 or organisation in Category A liaison with JTC 1 may propose that an existing standard (or amendment with the approval of the responsible SC) from any source be submitted without modification directly for vote as a DIS (or DAM). The criteria for proposing an existing standard for the fast-track procedure is a matter for each proposer to decide.

While the process to become qualified as a PAS submitter requires an evaluation of the organization and its process, etc., I don’t see anything required to become a Class A Liaison beyond an overlap in topics, which is how I read the description in 3.3.4.2 above. I’ve heard from numerous people within JTC1 that they consider PAS a much more difficult status to achieve, which is perhaps why there is currently only four of them (Free Standards Group, OASIS, OMG, and WS-I).

So, the JTC1 process doesn’t say anything at all about the process – or lack thereof – under which a specification must have been created and approved before it is submitted by the Liaison to a JTC1 SC “without modification directly for vote as a” Draft International Standard. So anything, whether it was created under an open, balanced, consensus-based process or not, can become a DIS without modification. It is up to the national body members of the JTC1 SC to decide if the quality of the specification overrides any lack in the process under which it was created and approved.

There is a required one-month comment period to identify contradictions, which for OOXML we had a couple months ago, and then a five-month ballot period, which is nearing its end. After approval as a DIS by the JTC1 SC, the specification is sent to ISO/IEC for publication as an International Standard.

Last week I was a bit critical of INCITS’s committee process that allowed what appears to be stacking of the committee in advance of a major vote. We’ve also seen reports from at least one national body where a weak process or – worse – bad enforcement of the process has made one wonder about how standards are created and approved, specifically Portugal. And Rob Weir points out that Ecma is pretty blatant in advertising that it approves submissions without revisions, allowing them to be submitted to JTC1 essentially without review.

But at least our friends in South Africa seem to be catching on to what is going on, and see the misuse of the PAS process as A Bad Thing. I’m starting to feel like the PAS process is broken, frankly, which is a shame. I think that the basic concept behind the PAS process, where work done outside the ISO/IEC JTC1 process can be submitted to JTC1 for approval, is a good one, and allows international recognition of work done by a variety of organizations. But as I said in February, “It’s not so much a bad process as a process with certain weaknesses.” JTC1 needs to take a look at revising the PAS process and be much stricter about what it allows to be submitted. Rubber stamping of work created in a closed, commercial environment is not compatible with an open, consensus-based standards process.

I’ve quite encouraged by Andy Updegrove’s report this morning that the INCITS Executive Board has defined the mechanism (an ad hoc process, I suppose) they will use to deal with the OOXML ballot and produce a U.S. recommendation to JTC1. I know most of the Executive Board members, as well as the V1 chair and the folks at JTC1 SC34, and know them to be professional, competent people. I’ll trust them to do the best they can with the process they have.

Even the best of organizations have weaknesses in their committee and approval processes. Not every possible future situation or controversy can be anticipated. Organizations learn over time, and most constantly revise their processes based on these new situations. We’ll just hope that all of the organizations involved in this recent controversy will, once things settle down, take a look at how to prevent the next one from being so messy.

I’m trying, generally, to keep out of the ODF/OOXML fray. While I have my personal opinions on the matter, in this blog I’m focusing on best practices for standards setting. I’ll leave the OOXML discussions to Andy Updegrove, Bob Sutor, and others.

This morning I read Rob Weir’s account of the voting at INCITS, the ANSI-accredited organization tasked with recommending the U.S. national position on ballots for JTC1. My reaction was the same as after reading the account, last January, of the JTC1 approval process, which didn’t define a key term — bad process.

In the past few weeks, according to Rob’s account, Microsoft and its allies have been stacking the V1 committee at INCITS. Where there were previously seven members of V1, six of whom have now voted against OOXML, within the last couple of months the number has grown to 26, with the 19 new members voting in favour. I can understand the interest that Microsoft has in getting OOXML approved as an international standard, and that it will push the rules as far as it can in order to accomplish this. I will fault, instead, the process of a standards organization that allows this stacking to happen.

A technical process should allow any interested party to join in the work of developing and approving a standard. That’s part of being open, and openness is required by the ANSI Essential Requirements as well being a basic principal required by the World Trade Organization’s policy on Technical Barriers to Trade, the U.S. Standards Strategy, the U.S. OMB Circular A-119, etc.

But a committee process shouldn’t necessarily allow anyone to come and go as they please, and give the right to vote to anyone who walks in off the street at the last minute. Allowing public inspection of a committee’s activities is important, and public comment should be encouraged. But participants should not be allowed to vote until they have been participating long enough to understand the issues and the technology behind the specification, and the process should especially guard against last-minute packing of the room.

Luckily this committee vote was only a preliminary step. We’ll wait to see how the further approval steps at the U.S. and JTC1 levels proceed over the next couple of months.

What’s the best way to conduct a virtual standards meeting? Given the cost in time and money, to say nothing of the hassles, of travel, most standards committees hold at least some, or even most or all, of their meetings virtually.

Over on Bob Sutor’s blog he wonders about using a virtual world such as Second Life to aid in the standards setting process. I’m not keen on the idea as I don’t think that watching a screen full of cartoonish avatars adds anything over and above the advantages of teleconferencing, instant messaging, file sharing, collaborative editing, virtual whiteboarding, electronic balloting, etc. that are already gained using existing tools.

A decade ago video conferencing promised to improve the quality of our virtual meetings, but it seemed that we always spent the first twenty minutes of each meeting trying to get the damn thing to work right. Perhaps there was some benefit from it once we got it working, in that you could get some small glimpse of body language, and you could tell who was in the room or when somebody left, but the added complexity, at least for me, was always a distraction from the purpose of the meeting.

For those committees who do conduct virtual meetings, keeping in mind the physical meeting metaphor generally makes sense. People are either in the “room” or not, based on whether they are dialed in to the teleconference. Instant messaging replaces the passed notes of the whispered private conversations taking place around the table or the back of the room. Passing around a paper document to the people in the room is replaced by uploading a document that everyone can access. Ballots of various levels of formality and complexity (from “when should we break for lunch?” all the way up to approving a specification) can take place using electronic balloting systems. And so forth.

But at some level this physical meeting metaphor breaks down. This usually happens at about the point when some poor soul from e.g. Australia or Japan, attending at what for them is the middle of the night, falls asleep and starts snoring on the phone. (Don’t laugh; it happens.) The committee then decides that discussions will start happening asynchronously over email, and that meetings will no longer be held.

Other than the snoring problem, are there other issues with virtual meetings that need to be compensated for or addressed? I’d like to hear comments from any readers who have run across issues not solved by sticking to the physical meeting metaphor.

When I’m consulting for clients helping them set up or improve their organizational structure, policies, and processes, one of my favorite topics is the technical process. This document is the “rules of the game” that specifies how the organization’s standards work is done: how are committee or work groups formed, how they operate, and how is work advanced through the approval process. As I suggested in a previous blog entry, it is important that these rules be codified and the document be made available to the participants. It’s also important that there be some sort of referee or process administrator whose responsibility is to make sure that the  rules are being followed.

While there is no “one size fits all” technical process for all standards organizations, there are a few basic topics that need to be covered, and some standard questions to be answered. Generally each step in the process needs to be structured as What is being done, How it is done, and By Whom. The following should be defined in the process:

• How open is the work? Is balance among participants a goal, and how is it achieved?
• What level of coordination of work is done within the organization (i.e. between all committees or work groups)?
• How is a new committee or work group formed? Who proposes the formation, and who approves it?
• What will the committee do? What is its charter?
• How is the committee leadership selected?
• Who can join the committee, and when? What participation and voting rights do members have?
• How is existing work accepted into the committee? (This is an area of overlap with the organization’s IPR policy.)
• How are the committee’s deliverables developed? How are they reviewed? Who can review the work and for how long? How are comments handled?
• What levels of approval are required both inside and outside of the committee? Who votes, and how?
• How is the work distributed or published?
• How is the work maintained and revised in the long term?
• How are committees or work groups closed?
• How do participants and other members make appeals, and to whom?

The answers to these questions will vary considerably, of course, depending upon the nature and makeup of the standards organizations, but these are the sorts of things that need to be defined so that the standards work can proceed smoothly and without controversy.