Do you love Microsoft?

There was an interesting article a few days ago by Alex Brown (the convenor of the UK technical group on OpenXML) in which he argues that it is not possible simply to vote against OpenXML in JTC1.

Having read his article carefully and spent an unhappy hour looking (not for the first time) at the JTC1 Directives, my main conclusion is that the JTC1 Directives should be awarded some kind of prize for bad drafting.

To play this game you have to start by getting hold of a copy of the JTC1 Directives (available in several places on the web). Make sure you have 5th Edition, Version 3.0 (the latest). Pros will want to read through the 179 pages a couple of times before joining the discussion.

Let us start with 9.6 which sets out how many votes DIS29500 (aka OpenXML) needs to pass:

For a FDIS/DIS/FDAM/DAM/FDISP to be approved, the count taken by ITTF shall meet the following criteria:
• At least two-thirds of the P-members voting shall have approved;
• Not more than one-quarter of the total number of votes cast are negative.
A P-member which has given appropriate notification that it will abstain from participation in specific work items (see 3.1.2) shall not be counted as a P-member when counting votes for drafts relating to such items.
Abstentions are excluded from the count.

Ignore for the moment the possibility that votes can be cast by entities which are not P-members. Suppose there are T = Y + N + R + A voters in total, of whom Y vote Yes, N vote No, R rule themselves out with the “appropriate notification” and A others abstain. Then the first bullet apparently requires Y at least (2/3) T, and the second requires N at most (1/4) T.

Hmmm. Is that right? Maybe “voting” in the first bullet means Y + N, not T. Perhaps we should check with 9.7 which sets out how you are allowed to vote? Unfortunately, that uses the word “reply” rather than vote, and gives three kinds of “reply” (approve, disapprove with comments, or abstain). But maybe the last type is a “reply” but not a “vote”.

The two exclusions do not help much because they seem to deal with “votes counted” rather than “votes” or “votes cast”.

So we are left with two requirements: Y at least 2N and N at most T/4. That makes sense. You want a healthy majority of those voting and not too much of your membership unhappy enough to vote against.

But before feeling too pleased with ourselves we need to look at the possibility of votes by entities which are not P-members. Serious types will immediately delve into the distinction between P-members of JTC1 and P-members of SCs (sub committees, in particular SC34 which is dealing with DIS29500). But that is probably a red herring, because 9.5 provides:

The voting procedure which uses simultaneous voting (one vote per country) by the P-members fo [sic] JTC 1 and by all ISO member bodies and IEC national committees on a letter ballot is called the combined voting procedure. This procedure shall be used on FDISs, DISs, FDAMs, DAMs and FDISPs.

So we appear to have reached:
(1) voting is open to P-members of JTC1; and
(2) any members of ISO and IEC who are not P-members of JTC1.

To pass, OpenXML must get
(3) twice as many yes votes as no votes from P-members of JTC1; and
(4) not more than 1/4 of all votes cast negative (but unclear if votes cast includes abstentions).