Do you love Microsoft?

So have you all been paying attention? If 12 P-members and 25 other ISO members vote yes, and 2 P-members and 10 other ISO members vote no, does OpenXML pass the fast-track ballot?

Well, attentive readers of Parts 1-3 should answer No. It meets the 2/3 of P-members voting test 12/14 exceeds two-thirds, and it meets the 1/4 of all voters test 12/49 is less than a quarter. But it fails the quorum test - less than half of the 32 P-members of JTC1 voted.

However, that is the wrong answer. We have to consider Ballot Resolution Meetings. Again we have to deal with the absolutely appalling drafting of the JTC1 Directives. They provide:

13.5 … the SC Secretariat … shall make plans for the handling of the ballot results through the formation of a ballot resolution group … [and shall] schedule a ballot resolution group meeting to consider any comments on the DIS … appoint a Project Editor for the DIS … [who] shall be responsible for producing the final DIS text in case of acceptance … In some cases the establishment of a ballot resolution group is unnecessary and the SC Secretariat can assign the task directly to the Project Editor.

13.6 … the SC Secretariat shall distribute [the ballot results and any comments] to the SC NBs, to any NBs having voted that are not members of the SC and to the proposer … The NBs shall be requested to consider the comments and to form opinions on their acceptability … Comments received after the normal voting period will not be taken into account …

13.7 NBs of the relevant SC shall appoint to the ballot resolution group one or more representatives … NBs having voted negatively, whether or not a NB of the relevant SC, have a duty to delegate a representative to the ballot resolution meeting.
…
13.9 If, after the deliberations of this ballot resolution group, the requirements of 9.6 [the 2/3 and 1/4 criteria] are met, the Project Editor shall prepare the revised DIS …

13.10 If it is impossible to agree to a text meeting the above requirements, the proposal has failed and the procedure has terminated.

There is no indication of when a ballot resolution group is unnecessary. One view is that it is unnecessary if there are no comments. Another view is that it is unnecessary if there are already enough yes votes.

Note that there is a misconception about yes votes. The accounts that have come out about the USA’s technical group suggest that one of the options it was considering was a “approve with comments”. But that is not a valid vote under 9.6. You can only vote “approve” or “disapprove with comments”. Of course, whether an “approve with commments” would be treated as a spoiled ballot, an “approve” or a “disapprove with comments” is anyone’s guess.

The next problem is that it is not directly spelt out what the BRM is entitled to do. If you look at 13.5 and 13.6 it is apparently asked to form opinions on the acceptability of comments. So suppose country X has voted “Yes”, and country Y has voted “No, because A should be changed to B”. Suppose a majority of the BRM agree it would be good to change A to B. Presumably Y is then expected to change its vote to “Yes”. It is unclear if it is obliged to, presumably not. Similarly, X might presumably change its vote to “No” because it preferred A. Equally it is unclear whether the proposer (Ecma in our case) has any say in the matter. Suppose Ecma is dead against B, does that make any difference (assuming others at the BRM disagree)? Presumably not, since it is clearly not given any vote or veto.

However, the much more important point is who votes on the results of the BRM deliberations. It is not spelt out, but the clear implication is that the votes are conducted at a BRM meeting, and hence that you have to make sure you are there when the vote takes place. In other bodies it is notorious that people call for votes when their opponents are temporarily absent, so presumably any NB that wants to be sure of voting needs to have someone attend all the meetings.

But can any ISO member turn up to the BRM? The apparent implication of 13.6 (a part not quoted above) “SC Secretariat shall also send notification of the [BRM]… to SC NBs, to any NBs having voted that are not members of the SC and to the proposer”, is that only those parties are entitled to turn up (and vote). So the 25 P-members of SC34 are apparently entitled to attend and vote whether or not they voted during the current ballot period, but other ISO members are apparently only entitled to attend and vote at the BRM if they voted (as opposed to abstained) during the current ballot period.

There is apparently no time limit. The SC Secretariat must distribute the comments etc within 2 months of the close of the ballot (according to the SC Secretariat paper 0779, but I cannot at the moment find this requirement in the Directives), the date for the BRM cannot be sooner than 2 1/2 months after this circulation (13.12). Within 1 month of the end of the BRM, the SC Secretariat shall distribute the final text in case of acceptance (13.12). But in principle the BRM could apparently go on (with adjournments) for months or even years.

So the practical upshot appears to be that the BRM is one big haggle. Assuming a reasonable number of comments to provide a pretext, everyone is free to vote whatever way they want, changing their vote as often as they want. Presumably the “convener” (appointed by the SC Secretariat per 13.5) tries to see which comments could be taken into account to push the number of “Yes” votes up to acceptance.

Given the sheer volume of comments, Project Editor would appear to have a Herculean task completing his work in a month if any number of them are accepted. But it is hard to see that an overrun would invalidate the process.