The Standards Blog

Detailed IEC Voting Results on OOXML Appeals

OpenDocument and OOXML

As I reported yesterday, the OOXML appeals brought by Brazil, India, South Africa and Venezuela have been rejected by the Technical Management Board (TMB) and Standardization Management Board (SMB) of ISO and IEC, respectively.  I have now received the actual voting results for the IEC vote, and an indecipherable screenshot of the ISO votes.  I'll hope to add the ISO votes later on when I get more comprehensible information, but in the meantime, here are the IEC results. 

In each case, the questions included in the ballot were the same:

a) Not to process the appeal any further

b) To process one or more of the appeals, which would require setting up of a conciliation panel

For each appeal, a board member could vote "yes," "no" or "abstain," with only the yes and no votes counting for purposes of determining whether or not the vote would pass, and a two-thirds vote being required for passage (the "Conclusion" at the end of the voting results cites clause 10.4, Rules of Procedure, as the reference for these rules). 

The inclusion of two questions in the ballot is (to me, at least) rather curious, in that each asks the same question in a different way, but with a "yes" vote having the opposite import.  In the first vote, it would take two thirds of those voting and not abstaining to stop an individual appeal from proceeding, while in the second it would require a two thirds vote to allow the appeal to proceed.  I confess that I don't understand the nuance that is being addressed here, and perhaps one of the old ISO hands that sometimes reads this blog can illuminate this for us (although one wonders whether at least some of those that voted may have been confused as well, given the results). 

Those voting on the IEC SMB were Australia, Brazil, Canada, China, Germany, Denmark, Estonia, France, Great Britain, Italy, Japan, Korea, Netherlands, Sweden and the United States (the full IEC membership can be found here).  The results were as follows:

Brazil appeal: 8 yes, 4 no, 3 abstain (the votes in support of hearing the appeal were cast by Brazil, Canada, Great Britain, and Korea).  Accordingly, on a percentage basis, the vote was 67 - 33% against  this appeal meriting further review.

India appeal:  10 yes, 2 no, 3 abstain (the votes in support were cast by Canada and Great Britain; Brazil voted against this appeal because the remedy requested by India was different than the one that it desired).  The vote here was therefore 83 - 17 against further review

Venezuela appeal:  9 yes, 3 no and 3 abstain (the votes in support were cast by Brazil, Great Britain and Korea; Canada abstained this time).  The vote here was therefore 75 - 25%  against further action.

South African appeal:  9 yes, 3 no and 3 abstain (the voting was the same as in the Venezuela vote).  Again, the vote was 75 - 25% against further passage.

One has to wonder whether various SMB members understood the difference between the two votes differently, because the votes on the second question came out so differently relative to the first.  Instead, some votes switch, while others translate from an active vote to an abstention, or visa-versa (those that switched votes in one or more cases were Canada, China, Korea and the Netherlands) .  I've "translated" the first votes and added them in parentheses below to make this more obvious:

Brazil appeal: 4 yes, 10 no, 1 abstain (4 yes, 8 no, 3 abstain).  Accordingly, the vote was 29 - 71% in favor of advancing the appeal (compared to 33% in the prior vote)

India appeal:  11 yes, 3 no, 1 abstain (2 yes, 10 no, 3 abstain), or 21% in favor, 79% not in favor (compared to 17% in the prior vote)

Venezuela appeal:  9 yes, 3 no and 3 abstain (3 yes, 9 no, 3 abstain), or 25% in favor, 75% not in favor (the same as in the prior vote)

South African appeal: 5 yes, 9 no and 1 abstain (3 yes, 9 no, 3 abstain), or 36% in favor, 64% not in favor (compared to 25% in the prior vote)

Eight out of the 15 members entitled to vote, 8 included brief comments on their ballot.  Sweden, which did not vote in favor of further prosecuting any of the appeals, noted that the same sort of confusion over process could arise again, and accordingly stated that it:

..recommends that a review of the Directives of ISO/IEC JCT [sic] 1, especially those governing fast-track standardisation will take place.  This has been proposed earlier and has become more important now.  The aim should be to get the JTC 1 directives better aligned to the ISO/IEC Directives.

How to interpret all this?  Here's the best that I can do: 

  • As many as 5 out of 14 and as few as 2 out of 12 voting up or down thought than an individual appeal should receive further attention
  • One third of those on the Board thought that there were serious enough problems with the OOXML process to merit an appeal in at least one case (and two thirds, of course, did not)
  • At least one additional members took advantage of the comments to recommend that the OOXML process demonstrated the need for procedural reform - a sentiment that has been widely echoed by other members
  • Ironically (at least) the ballot issued to resolve the dispute appears to be as inscrutable and difficult to deal with as the process that was being appealed.  In a comment to its vote, the Australian representative, while voting along with the Secretaries General recommendation, added, "The above voting set up does not clearly reflect the intention of such vote."

So, while the appeals are likely dead, I would certainly hope that the cause of reform will proceed.

For further blog entries on ODF and OOXML, click here

sign up for a free subscription to Standards Today today!

Comments

Permalink
Without having looked at the rules for voting, if a vote requires a 2/3rds majority to pass, then the choice of question (e.g., either being in the affirmative or in the negative) basically pre-biases the result because it makes it much harder to return one result than the other. But asking in both forms makes it roughly equal. If they had only asked a single question, that would have caused one of the sides to complain, surely?

If the votes had come out contradictory, that would have caused an obvious problem to me (but would have been logically valid), but given the questions are almost (but not quite) diametrically opposed, that's not really a likely result.

Permalink
Anon,

I agree that asking the question twice nominally balances things out superficially, but I'm not sure that was the reason - or the result. 

To begin with, from the biasing point of view, note that the vote was accompanied by a joint recommendation from the Secretaries General to deny the appeals, and the first vote is set up to support that recommendation.  Passing that by for now, if the true reason was not to bias the result, then the second vote would simply have been stated the same as the first, with the minimum number of word changes possible.  Like changing:  "Not to process the appeal any further," to something like "To process the appeal further."  So it seems to me that there is something else going on that escapes me.

Regardless of the intent, it seems to me that there was at least confusion on what the purpose was, given the fact that not only did the vote not come out symmetrically in any precise fashion, but it did not come out symmetrically against single appeals, and National Bodies did not seem to apply a consistent reasoning in the way in which they changed their votes, either.  To be fair, I doubt that any of this affected the outcome.  It's more a symptom of a process that seems out of touch with best practices, and that seems to be immune to effective improvement.

As noted in the main text above, this is sadly consistent with the (non-appealable) process decisions made by those in charge of the OOXML process  from beginning to end.

What a surprise.

  -  Andy

Permalink
For the results of the second Brazil appeal the text "the vote was 29 - 71% in favor of advancing the vote" confused me.  Should it read "in favor of advancing the appeal" or perhaps "in favor of dismissing the appeal"?

"Venezuela appeal: ... (the came as in the prior vote)" should perhaps be "the same as in the prior vote".

- Stuart

Permalink
I don't follow. Surely 'a' and 'b' aren't independent questions but are the two possible responses to a single question?

Also why should an appeal need a 2/3 majority to be allowed to be considered? This allows any cabal of 1/3 of the voters to ignore problems which the majority may recognize.

Also why should an appeal need a 2/3 majority to be allowed to be considered? This allows any cabal of 1/3 of the voters to ignore problems which the majority may recognize.

Well... no.

Note that this is an appeal of something regarding which the voting has already taken place. That is, it is a request to overturn (or consider overturning) the will of "the majority" as expressed in that vote. Given that it is a request to ignore the majority, a somewhat higher standard (such as 2/3 majority of the board) is quite reasonable.

Greg Byshenk

There should be no vote by the majority on whether to question the majority's decision or process for arriving at that decision. Any appeal which is even coherent should be examined for legitimacy without the ability of the majority to quash such an appeal.
This vote is nothing more than asking the majority to determine whether they should allow questioning of their previous voting process. No one should be surprised that it turned out as it did.

First, not quite. This is not a "vote by the majority on whether to question the majority's decision", but an evaluation of the merits of the appeals regarding a different vote of a different group, the SMB, that is constituted to evaluate such appeals.

Second, while I understand your concern, I don't see that there is any solution to the problem. Any more or less democratic system is open to abuse of minorities by the majority, and any "solution" enables abuse of the majority by minorities.

Third, your argument here seems to be almost exactly the opposite of the one to which I responded. The original claim was that the process "allows any cabal of 1/3 of the voters to ignore problems which the majority may recognize" (that is, abuse of the majority by a minority), while you are arguing the opposite (that is, abuse of a minority by the majority).

Greg Byshenk

In reply to by Anonymous

Permalink
First, not quite. This is not a "vote by the majority on whether to question the majority's decision", but an evaluation of the merits of the appeals regarding a different vote of a different group, the SMB, that is constituted to evaluate such appeals.
The list of voters that Mr. Updegrove provides looks extremely similar to the list of voters participating in the process which the four appellants question.

Even granting that "looks extremely similar" is somewhat imprecise, the claim seems odd. Do these two lists in fact "look[] extremely similar"?

Australia Australia
Azerbaijan Brazil
Belgium Canada
Canada China
China Germany
Cote-d'Ivore Denmark
Cyprus Estonia
Czech Republic France
Denmark Italy
Ecuador Japan
Finland Korea
France Netherlands
Germany Sweden
India United Kingdom
Iran USA
Ireland
Italy
Japan
Jamaica
Kazakhstan
Kenya
Lebanon
Malaysia
Malta
Netherlands
New Zealand
Norway
Pakistan
Saudi Arabia
Singapore
Slovenia
South Africa
South Korea
Spain
Switzerland
Trinidad and Tobago
Turkey
United Kingdom
USA
Uruguay
Venezuela
Yes, there is some shared membership, but the two lists look quite different, at least to me. To be sure, both lists are made up of ISO members, and they are "similar" in that respect -- but that is something that is hardly to be avoided.

In short, as I wrote previously, "This is not a 'vote by the majority on whether to question the majority's decision', but an evaluation of the merits of the appeals regarding a different vote of a different group, the SMB, that is constituted to evaluate such appeals." Certainly there are other ways in which an appeals process could function, as surveyed by Rob Weir, and certainly it could be argued that some of these other ways are superior to the ISO appeals process. But, even if one accepts that the ISO appeals process is flawed, that does not make it a "vote by the majority on whether to question the majority's decision", which it is not.

Further, Rob Weir's view of the "best practices" for appeals procedures (which I agree with, for the most part) does not serve to solve the problem noted previously. If decisions are to be reached by some majority system, then the majority can (ultimately) run roughshod over the minority, regardless of what appeal process might occur prior to the final majority vote. On the other hand, if the appeals process can carry  on endlessly, then the minority can endlessly block the will of the majority.

Greg Byshenk

 

In reply to by Anonymous

Permalink

In short, as I wrote previously, "This is not a 'vote by the majority on whether to question the majority's decision', but an evaluation of the merits of the appeals regarding a different vote of a different group, the SMB, that is constituted to evaluate such appeals."

You list 15. 2/3 of that is 10. Of those 15, 6 voted to approve OOXML by way of the processes being appealed. 3 of the 15 voted to disapprove. Looks like a majority of those considering whether to even give the appeals further evaluation found nothing wrong with the voting process being appealed. 4 of the 15 abstained from voting on OOXML, 1 did not vote, and 1 withdrew due to irregularities.

Any way I slice it, the majority decided that the majority did nothing wrong.

In reply to by Anonymous

Permalink
Well Greg...I think your post is missleading. The real problem here is overlap of roles and not what NBs that voted on if process the appeals. What has happened are that personal of the NBs are involved in a vote for approval of OOXML, unfortunately the process is handled so badly that four NBs decide to appeal.

Any standards organization with intrigity would at this point require that evaluation if the appeals to be heard must be performed by skilled people that did not participate in the vote that are appealed. They could for instance bring in expertice from some other technical area that ISO handles. Problem is that ISO makes no such demands but instead let the same people who handled the OOXML vote make the decission.

What it boils down to is that if the NB personal have clean hands there is no reason for them to try to stop an investigation that will clear their name. Yet if there are irregularities that they are involved in they have every reason to try to stop the appeals bacause continued investigation will reveal what they have done.

In short, as I wrote previously, "This is not a 'vote by the majority on whether to question the majority's decision', but an evaluation of the merits of the appeals regarding a different vote of a different group, the SMB, that is constituted to evaluate such appeals."
[...]Any way you slice it, the majority decided that the majority did nothing wrong.

Only if you engage in some perverse wordplay based on an ambiguous use of the term 'the majority'.  The problem is that, as suggested previously, your two instances of 'the majority' do not refer to the same thing. As an illustration: if we suppose that there were an entirely independent appeals board, which board voted (by majority) to deny the appeal, then one could similarly claim that "the majority decided that the majority did nothing wrong". But, also similarly, this would be misleading, because the two instances of 'the majority' are not the same.

The real issue, as the second Anonymous points out, is the overlapping of roles and interests between the two groups. Generally, when evaluating such things as appeals, one prefers that whoever is doing the evaluation be someone uninvolved with the original decision. And this need not have anything to do with anyone having done anything "wrong"; it can be just that those who favour a proposal will naturally want it to go forward, and will view any appeal through that filter, while those who oppose the proposal will naturally want it to be stopped, and will view any appeal through that filter.

I will add, for the second Anonymous, that I do not believe that anything I've written is misleading. Indeed, I have noted that the appeals process under discussion is at least arguably an inferior one. The only point that I have tried to make here is that certain comments -- e.g.: it "allows any cabal of 1/3 of the voters to ignore problems" or "the majority decided that the majority did nothing wrong" -- misrepresent the situation.

Greg Byshenk

Permalink
It's interesting that they would accept such results:
   a) Not to process the appeal: India - 10 Yes, 2 No
   b) Process one or more of the appeals: India - 11 Yes, 3 No

So answering the first vote, it looks like 2/3 majority does not want the India appeal to proceed.  However, answering the second vote, 2/3 wants the India appeal to proceed.  Without reconciling the voting, the people in charge counting the votes can simply take the vote they prefer and ignore everything else.

RAS

The bottom line is, the people in charge are crooks. I don't know why we keep wasting our time paying attention to this rotten process!

Whether OOXML gets adopted by the world or not is not that important. What is important is to make sure ODF is better and it will definitely be the genuine standard that will furnish what is desired from a standard.

ISO has to be revamped or abolished.

Permalink
RAS,

I apologize, and realize that the way that I wrote this up would only be clear to me.  The first number is in favor, the second number not in favor.  I've now edited this to make it more clear.

  -  Andy

I took it at face value the numbers you presented were the original, unmodified versions of the actual votes that were passed.  With the two questions, the first one being:

   a) Not to process the appeal any further.

A vote of yes would mean "don't process", a vote of no would mean "proceed and process".  It's understood that question was asked for each of the four separate appeals.

The second question is:

   b) To process one or more of the appeals, which would require setting up of a conciliation panel.

A vote of yes would mean "process", a vote of no would mean "don't process".  This question is also understood to be asked for each of the four separate appeals.

So long as the above understanding is correct, we have a cross-logic situation currently occurring in what you provided.

   1) The numbers presented are the actual votes that occurred.

   2) The numbers presented in quotes are the actual votes that occurred.

If the numbers are correct, we have:

   To answer the first question for India, 10 yes, 2 no - majority says "don't process",  the second question for India, 11 yes, 3 no - majority says "process".

The question would be: how do they get "interpreted" into being counted against?

However, if the numbers in quotes are correct, we have:

   To answer the first question for India, 10 yes, 2 no - majority says "don't process", the second question for India, 2 yes, 10 no - majority says "don't process".

Which would raise the question: How did you move from 1 abstain in the response to the second question on India to 3 abstain?  Where did the 1 abstain come from?

If they had been reversed due to a cross-translation, wasn't that translation applied to all four?  How come the numbers reflect a conflict?  For example, two of them have the numbers reversed on yes in the second answer compared with the other two:

Brazil: First 8-4, Second 4-10 (4-8) -> Seem to reflect same voting (4-10 vs 4-8, both weighted in the same direction)

India: First 10-2, Second 11-3 (2-10) -> Conflict between quoted and non-quoted compared to Brazil (11-3 vs 2-10, conflict in vote weight direction)

Venezuela: First 9-3, Second 9-3 (3-9) -> Conflict between quoted and non-quoted compared to Brazil

South African: First 9-3, Second 5-9 (3-9) -> Seems to reflect same voting as Brazil

RAS

It seems that the Secretaries-General managed to lose everybody: the TMB and SMB members, Andy, you, me and everybody else.  Perhaps was it their goal ?

But, as their ballot "clearly" stated that the TMB/SMB members should vote "on EITHER a) OR b) for each of the four appeals" (emphasis is theirs), and as most members voted on BOTH a) AND b), shouldn't the dual votes be declared as invalid ?

Methinks the vote should be started again, with a clear and unambiguous question, like:
"The members of the TMB/SMB are invited to indicate if they want to process further
- the ABNT appeal                       [Yes] [No] [Abstain]
- the BIS appeal                           [Yes] [No] [Abstain]
- the FONDONORMA appeal     [Yes] [No] [Abstain]
- the SABS appeal                       [Yes] [No] [Abstain]"

The ballot should also clearly state how much votes are required for the ballot to pass.  It seems that the rules are different in IEC (2/3 of the votes) and in ISO (1/2 of the votes ?).

Or better than a vote, the TMB/SMB boards could meet, discuss and reach a consensus as recommended in the JTC1 Directives :
"1.2 General Provisions
These Directives are inspired by the principle that the objective in the development of International Standards should be the achievement of consensus between those concerned rather than a decision based on counting votes. "

Luc Bollen

The preview area showed the table very nicely but the post didn't.
Let's try again with html
<pre>Vote                                                                   County of Appeal Yes   No     Abs     For
        a) Not to process the appeal any further      Brazil                       8      4          3      66.67%
                                                                                      India                      10      2           3     83.33%
                                                                                      Venezuela              9       3          3     75.00%
                                                                                      South African         9       3          3     75.00%           
       b) To process one or more of the appeals   Brazil                       4     10          1     28.57%
                                                                                      India                      11       3          1     78.57%
                                                                                      Venezuela              9        3         3     75.00%
                                                                                      South African          5        9         1    35.71%
</pre>

Andy,
  I am an engineer and I don't appreciate verbiage too much. Tables were created to make things a lot easier to differentiate and compare. I am sorry your are a lawyer. :)
 Here is a table form of your figures that would make things a lot easier to comprehend by everyone.

Vote County of Appeal Yes No Abs For a) Not to process the appeal any further Brazil 8 4 3 66.67%   India 10 2 3 83.33%   Venezuela 9 3 3 75.00%   South African 9 3 3 75.00%             b) To process one or more of the appeals Brazil 4 10 1 28.57%   India 11 3 1 78.57%   Venezuela 9 3 3 75.00%   South African 5 9 1 35.71%

Permalink
"According to the ISO/IEC rules, DIS 29500 can now proceed to publication as an ISO/IEC International Standard. This is expected to take place within the next few weeks on completion of final processing of the document, and subject to no further appeals against the decision."

I also first read this as saying that the decision is not subject to further appeals, but on second reading, seeing that it does not say "and is subject to" and that the rules do provide for further appeals, I think they meant what would have been more clearly worded as

"Subject to no further appeals against the decision, this is expected to take place within the next few weeks on completion of final processing of the document."