The Standards Blog

The ISO/IEC Voting Process on OOXML Explained (and What Happens Next)

OpenDocument and OOXML
I've had several questions from journalists today who want to know how the votes will be counted on the OOXML vote, and what the distinction is between the influence of P (Participating) as compared to O (Observer) members of ISO. I had promised in a prior blog post to explain the rather complicated rules, and with voting now in its final stages, that time is now. While I'm at it, I'll also explain in greater detail why the surge in membership in SC 34 matters, and what will happen between now and the OOXML Ballot Resolution Meeting (BRM) in Geneva, Switzerland in February 2008.  

First, here's the flow chart for how it will be determined whether the vote to approve OOXML has passed:

[Updated:  The following explanation is presented in linear fashion for clarity.  As Alex Brown points out in the comments to this post, each step is of equal importance, and are applied simultaneously.  Alex is helping to maintain a detailed explanation at the subsection of the OOXML Wikipedia entry that addresses the ISO process.  Alex will be the chair of the BRM (in ISO-speak, the "Convenor") - a difficult job indeed.]
 
Step one: Count all P votes cast, and determine whether at least 50% of P members have voted
 
If yes, then proceed to next step
 
If no, then the resolution fails
 
Step two: Subtract all abstentions (in other words, throw them out)
 
Step three: Determine whether 2/3's of the remaining P votes are to approve
 
If No, then the resolution fails
 
If yes, then go to step four
 
Step four: Determine whether 25% or more of the total votes cast by both P and O members are "No" votes
 
If yes, then the resolution fails
 
If no, then the resolution passes
 
From this, it becomes clear that P votes are far more important than O votes, and therefore why the recent upsurge in P members – from 30 to 41 this year – is so significant. We will see what the actual votes will be next week, but my expectation at this point is that the votes cast by those that were P members on January 1 of 2007 will be heavily negative, and that OOXML would have failed decisively absent the recent upgrades. 
 
From the above, you can also see that an abstention is the next best thing to a "yes," since it helps make the 50% test, and also avoids another vote in the no category, which can't exceed 25% of all votes cast. As a result, a conversion of the Swedish vote from a "yes" to an abstention, due to irregularities, is still helpful to the OOXML cause. That's because if the vote would have been "no" without the sudden increase in membership (and it appears that it would have been), the increase still succeeded in taking a no vote off the list – a tactical victory.
 
So now let's turn to what happens after the votes are cast.
 
Step five: The ballots and comments will be sent to Ken Holman, the Chair of SC 34 (Canada currently holds the Secretariat roll for SC 34, but recently announced that it wishes to relinquish that role. As a result, Canada will later be replaced by Japan). The votes and comments will also be sent to Ecma and to all JTC1 members
 
Step six: Ecma and Microsoft have until January 14 to come up with their suggestions for how to resolve the comments. Those suggestions can be to ignore or to address any given comment, and if the latter, how to address it. Note that while Ecma and Microsoft will have four and a half months to make their recommendations, the JTC1 members will only have about six weeks to review those recommendations before the BRM convenes on February 25.
 
Step seven: SC 34 reviews the suggestions, and forms its opinion on what should be done with each comment. One task would be to determine whether some issues are "irresolvable."
 
If yes, then cancel the BRM, and the process ends
 
If no, then consider how to resolve issues
 
Step eight: The BRM meeting is held. If the initial vote was to approve, then the process completes with the agreement on which comments should be accommodated. If the initial vote was no, then the question is whether enough comments can be resolved to the satisfaction of those that voted no.
 
Step nine: If necessary, revote. And note: if there is a revote, it will be of all P members at that time. So keep an eye on whether there are further upgrades of O members to P membership, even after the September 2 vote closes.
 
That leaves one last topic: why does it matter that SC 34 has more than doubled in size this year?
 
Here's why: if a National Body votes no, it has an obligation to attend the BRM. On the other hand, if it votes "yes," then its role is basically done. But – if it joins SC 34, then it has an obligation to participate in the process of addressing the comments. If SC engages in meetings before the BRM, it has the obligation to participate in them. And at the BRM itself, it also has an obligation to help resolve the comments. Since the resolution is by "consensus," then the more proponents of OOXML are members of SC 34, the greater the pressure to resolve comments in a manner favorable to Microsoft – if it so happens that Microsoft is pulling the strings of all of these new SC 34 members.
 

It will be interesting indeed to see how the votes come in next week, and how those votes map against those that have upgraded their status, and to those that have conceived a recent urge to be part of SC 34. I will report on, and analyze, those results as soon as they are available.

For further blog entries on ODF and OOXML, click here

subscribe to the free Consortium Standards Bulletin

Comments

Permalink
If SC engages in meetings before the BRM, it has the obligation to participate in them. And at the BRM itself, it also has an obligation to help resolve the comments.

Obligation is one thing, and enforcing that obligation is something else.

I don't see any mechanism for actually enforcing that "obligation," such as barring non-participating members from voting.  In a normal standards setting this wouldn't matter, but we've already seen quite a few instances where an NB forms a WG to deal with DIS-29500, and the WG holds a meeting the same day to approve it without discussing the content at all.  What happens if:
  • MS (via ECMA) makes no substantive changes to DIS-29500
  • 33 P members play possum until the BRM
  • At the BRM, all 33 vote to approve DIS-29500 more or less as it stands today.
In other words, a straight "party line" vote; my way or the highway.

Again, I'm well aware that in a normal standards process, failure to participate weakens a party's persuasive authority.  However, if the votes are there for a coup de main, then this isn't really a "normal standards process" any more, is it?

I expect you're right about there being no teeth to the "obligation" part, but would note that it's not infrequent in standard settings to deny the right to speak or vote if you haven't attended, say, at least three out of the last four meetings, so that you don't get dragged down rat holes or otherwise waste time.  Since a committee works in more of an ad hoc fashion in this setting, though, I expect that there would be no standing rules to that effect here.

That said, what I was really highlighting here isn't the obligation, but the opportunity for the newbies to act just as you suggest.  The OOXML process at the NB level has been awash with examples of people pulling fast ones with rules, from the chairpersons position, and with innocent chairs being steamrolled by stacked rooms.  Having all these new people involved in the comment resolution process, acting together, can have a profound influence over a bunch of people who are not nearly so motivated, or who may independently be under pressure to bend.

So my point here is the potential for a lot of people who have already joined just to vote "yes" who would normally be pretty much out of the loop after that to do the same thing over again, using the opportunity to exercise the "obligation." 

  -  Andy

 

Not to worry. There's no way anyone would know that was they case unless they understood the rather Byzantine system. One reason I did this post was because even if you read the rules, they aren't all that easy to parse through. Hopefully this will provide an easier roadmap for folks trying to follow, and understand, the process. - Andy

Permalink
"Note that while Ecma and Microsoft will have four and a half months to make their recommendations, the JTC1 members will only have about six weeks to review those recommendations before the BRM convenes on February 25."

That four-and-a-half months is not wasted time. During that time, NBs (their experts, delegates, working groups etc) can discuss and familiarize themselves with the other NB's comments, and work towards provisional national opinions on them.  It may be that the NB experts think "Oh, we agree with that one" or "Freedonia's proposed solution to this problem is much more practical than ours" or "Laputu's comment is crazy"  or "Utopia's point is against the goals and scope of the standard" or "Mordor  has scrutinized this more than us, and they think this is not so important, so a partial fix is OK". So during this time an NBs emphasis can shift on which issues are critical to be resolved and which issues there can be a compromise (or, even, which issue the NB was mistaken in). 

Rick Jelliffe

Your point is perfectly true, and I'm sure that NB's will be using that time for exactly these purposes.  Yesterday was a really busy day (that was my second post, and I did four interviews besides more normal work).  I typed this one in the car while someone else was driving, so it's probably not my best and most thorough piece of work.

Also, I admit that I was focusing on the half full glass here rather than the opposite.  Many people thought that Microsoft's and Ecma's manner of addressing the comments made during the Contradictions phase was not as productive as it could have been.  If one assumes that Ecma will be cooperative with Microsoft, and that Microsoft will want to minimize the number of accommodations to comments, then they will have four and a half months to discuss strategy and craft rationales for why this comment and that comment should not require action. 

The NB's that made the comments, or agree with the comments, will then only have six weeks to come up with their rationales for why they think this point and that point actually do require action.  Given how many hundreds of comments will need to be dealt with, that's not a lot of time.

And to be fair, given that addressing some of these comments would presumably be very difficult for Microsoft (e.g., replacing codes and specifications of Microsoft's with existing codes and standards), it would be logical for Microsoft to want to do as little violence to its already-existing and released product that is based on the specification in its current form.

In any event, with another standard with fewer comments and a less highly charged atmosphere, the distinction would make much less difference.

  -  Andy

Does this not also give Microsoft another 4+ months to 'buy off' or 'extort' additional countries ?&nbsp; I'm thinking high-level diplomats being told by Gates/Ballmer/etc that MS will move thousands of jobs out of the country if they don't vote the 'MS way'.&nbsp; Also, while this does allow extra time for MS to 'persuade' its flunkies/partners to participate in another grass-roots movement like we saw in Sweden, this also allows time for some of the MS letters/communiques to 'leak' to the general internet.<br /> <br /> This is not to say that all countries are 'buyable' (though it looks that way to me - especially after reading about the 'vote' in Poland), but that I question how much technical vs political and monetary 'negotiation' will be going on during the next 4 months.<br /> <br /> All-in-all, this is looking worse and worse for the ISO.&nbsp; I've already lost all respect for the ECMA - they are clearly in Microsoft's pocket.

Permalink

I get the feeling that what we are seeing here is similar to what happened with email -- a system was created without realizing that people would misuse it for commercial advantage.  So we got junk email years ago, and now we are about to get a junk standard.  Spamdard?

I hope the standards-making process is better-suited to repair than the email system has been.

I wonder whether this is the first time that someone has misused or attempted to misuse the standard-making process.  If it has been tried before, it would be interesting to look back and see what came of that.

Hmmm. Spamdards.  Perhaps if Microsoft does get OOXML approved, we should make a practice of always referring to it as the OOXML Spamdard and let the natural tendency people have to avoid looking ridiculous work to minimize the number of official adoptions of the spamdard.

First of all, I like your "Spamdard" term. Regarding whether this is typical behavior: the best analogy would be to weather. There are all kinds of gradations between a beautiful day and a Hundred Year Storm. This is a hundred year storm. 

But there are lots of cloudy days, showers and lesser storms in standard setting. Where everyone is experienced (of course, not always the case), some of it is even just the way things are done, and no harm done - it's like politics in any government, where it may look bad from the outside, but it's a way people have evolved for working together. And, some industry niches are much more rough and tumble than others.

Most experienced people would say that this type of thing only happens at this scale once every five or ten years. Because of that, people are likely to shrug their shoulders and say that it's not worth reforming the system to stop something that happens so infrequently.

This time, I think that there's going to be a more determined effort to straighten things out so that it doesn't happen again. - Andy

Permalink
Andy,

I think there are some things in your description of the process which need correction/clarification:

For Steps 1-4, it's not clear (though the JTC 1 Directives aren't too helpful on this point) whether the votes are even counted at the end of the letter ballot: the important test is whether the ISO arithmetic holds at the end of the process, not this mid-way point. Consider, for example, if every country votes "no with comments" with only trivial technical comments, in this case it would be clear that the comments would be ultimately resolvable and the process would continue.

For Step 7, I'm interested you state SC34 has been tasked with reviewing Ecma's proposed changes. Where did you get this from?

Steps 8-9. The ballot resolution meeting operates ultimately by voting, not consensus, and for DIS 29500 to proceed the vote at this meeting (potentially of all those who voted in the letter ballot) must meet the usual ISO arithmetic. I've been trying to keep the description of the BRM process accurate on Wikipedia.

- Alex Brown.

It's a pretty confusing process, and I'm using the flow chart to try and make it as clear as possible, rather than trying to make it slavishly follow the Directives point by point (for example, it would be as useful, from an explanation point of view, if I had used Step Four as Step One). 

On points 8 and 9, the consensus I was referring to was more in the nature of addressing the comments submitted with votes one by one, which is what can take days.  Also (on point 9), I didn't go into as much detail on the different alternatives as I could have.  For example, by "revote," that would only apply if the initial vote of yes had failed - then you'd need the revote.  If it had been yes, then if the comments were significant enough to require a BRM, there would be no need for a revote - just consensus decisions on the comments.

Thanks for the pointer to the Wikipedia entry - I should have thought to look to see if there was an explanation there, and to point to it if so.

  -  Andy

I'll defer to Alex on this if he wants to comment, but I believe that the role of SC34 is to help prepare for the meeting.  At the BRM itself, those at the meeting will be in control.  Of course, if you cross compare the two committees, and particularly at the P level, there will be quite a bit of overlap.

Another thing to watch for at the BRM is that normally those that vote "yes" don't show up.  Here, clearly those that voted yes with comments would have an interest in showing up to defend their positions.  If they voted yes with comments and don't show up, that would be curious.

  -  Andy

@Alex,
I see you have been appointed as the convernor of the BRM. Can you tell us a bit more about your role? Is "converner" like the chair of the meeting? Do you get a casting vote? What can you do if you feel the meeting is full of people who don't know what they are talking about but want to just vote to ignore every technical issue raised? What can you do if the timescale for the meeting (1 week I believe) does not allow time to cover the full agenda of comments raised?