The Poison Pill: the INCITS non-finding on Open XML

by Rick Jelliffe

I'd long ago told my Microsoft contact that I thought ultimately ANSI will abstain on the Open XML vote at ISO, due to an inability to achieve consensus, so I was quite surprised that they only missed out by maybe one vote in the end in the V1 technical committee. From emails from my friends on both sides of the table, it seems discussions at the V1 committee meeting became, if not acrimonious, then whatever the step beyond "robust" is. It will be interesting to see what INCITS/ANSI does to proceed, but it is delightful to see that the opponents of Open XML have started protesting that they really wanted Open XML all along!

(NOTE: I am withdrawing the rest of the blog, for now, while I think about whether it just perpetuates tit-for-tat sniping. Consequently, some of the comments no longer have their context. Apologies. The missing parts express surprise at Sun's recent statement of support for Open XML, bring up what anti-trust means in the context of standards, points out the impracticality of adding thousands of pages of binary mapping documentation to the Open XML standard, looks at the bad logic used to justify it, brings up the notion of a "poison pill" which is a trick where an impossible-to-fulfill clause is added knowing it will cause refusal, and also thinks about voting procedures when there are multiple choices.)

18 Comments

marc
2007-07-18 17:31:11
"I'd long ago told my Microsoft contact that I thought ultimately ANSI will abstain on the Open XML vote at ISO, due to an inability to achieve consensus, so I was quite surprised that they only missed out by maybe one vote in the end in the V1 technical committee."


correction: two votes, no one vote ( 15yes / 25 total = 0.6 < 0.66..., 16yes / 25 total = 0.64 < 0.66, 17yes / 25 total = 0.68 > 0.66...).


and regarding your surprise: haven't you anticipated the flooding of members at V1?



Rick Jelliffe
2007-07-18 20:33:55
marc: But participation is good. People join because they have an interest and leave when they have none.


We may snear at single-issue voters, rather than long-termers, but that does not make their vote unimportant or misguided or in any way wrong: we want more of their participation not less.


And it is one thing to join because you want to support a new standard that you can use, and entirely another because you want to prevent a standard that your competitors would like. The latter has anti-trust and bad-faith-participation risks.


The idea that there is something improper in people who need a standard participating in order to get it, while there is something proper about people who don't need a standard participating in order to block it in favour of their commercial interests, is complete rubbish. The former is what the process is set up to acheive, the latter is an abuse.


Committees typically have procedures so that new members cannot vote immediately, until they find their feet, get a grounding in the procedures and issue, and perhaps so that the old alpha males get to establish their dominance over new members of the pack. (I sometimes laugh to myself that Lynne Price, who really loves dogs, should be involved in the standards process since it is sometimes viewable in such canine terms.)


In some standards bodies voting accreditation is based on time, in some it is based on the number of meetings: in the case of V1 where they had multiple meetings this worked in favour of people who were interested but only could spare the time to attend the last critical meetings. Now newcomers may cause frustration to old-timers, but being an old-timer is no guarantee that you won't waste time asking procedural issues or mistaking people's answers: a bore just becomes a bore about more arcane matters, for example.


New members will always be disruptive, not always in a negative way, because they don't necessarily buy into previous decisions of a committee. Indeed, it is when interested parties think a committee is likely to go in the wrong direction that they will join. So long-running committees tend to be a little schizophrenic in their later stages: voting against things they had previously agreed on. (Indeed, some committees have a rule where it takes a simple majority to agree on an issue in the first place but an absolute majority to alter this: there are various justifications for it, but it has some bad effects sometimes.)

dave
2007-07-19 02:10:52
What do you think of this summary prepared for the UK government?


http://www.xmlopen.org/ooxml-wiki/resources/ooxml-in-ten-points-for-apig.pdf


It seems to echo Sun's line that having OOXML documented would be a very good thing, but that the current proposal falls very short of achieving that aim.


The notes from their deliberations are rather 'robust' as well:


http://www.xmlopen.org/ooxml-wiki/index.php/Main_Page


e.g.


  • There is a startling amount of errors in examples. This was supported by the automated testing of Inigo Surguy (cf http://surguy.net/articles/ooxml-validation-and-technical-review.xml)
  • It seems that there was no close technical and editorial examination of the documents by ECMA.
  • OOXML should't have been fast-tracked without such an examination.

I'm also intrigued by their apparent consensus that voting "Yes, with comments" allows the comments to be ignored, since they apparently can only be mere editorial comments in that case.


In other words anyone voting yes, is voting to fast-track (!) the actual content of the document as-is. Given the nature and number of errors that appears to be an appalling abuse of the standards process and I personally would be questioning the motivations of anyone voting yes, rather than those wishing to see a certain level of quality in ISO standards.

Rick Jelliffe
2007-07-19 03:47:01
Dave: There are different requirements for the kinds of comments that can be given at different stages of a standard. For the final vote on a normal standard, "Yes" comments of a technical nature will be put in the maintenance in-tray.


I'll have to look up (again...sigh) to see how Yes with comments are processed for "Enquiry drafts" (which is what the fast-track draft is AFAIK.)


The errors in examples is an interesting point. There is no requirement in ISO documents that examples should be well-formed or valid: leaving out tags, adding elipses, omitting attributes: all those things are legitimate editorial decisions. Now it may certainly be true that non-well-formed or invalid examples may be confusing, but that is a different issue to their being wrong. Of course, examples that are not well-formed because of typos such as delimiter errors or mismatched names or incorrect names is an entirely different issue: my reaction to the Surgey material was that it was good information with incorrect conclusions, because it doesn't cope with elision or tag omission.


Unfortunately my Linux system doesn't read that PDF file from the URL: I'll have to look at it on brand X. I'm going to re-install a shiny new Linux tonight anyway.


As for robust discussion, the UK (like Japan) has a very activist tradition of producing lots of comments: I expect they will find it congenitally difficult to accept that in a fast-track spec they are primarily documenting an existing technology rather than inventing and pottering one.

Rick Jelliffe
2007-07-19 07:08:21
Dave: OK I while I am installing my new Linux, I have sneaked to a collegue's computer and looked at the UK PDF.


I think it is very good and a fair overview. I particularly like points 6 and 7, because even though MS and the ODF vendors get all the attention in the gladatorial editorials, the most important stakeholders to consider are indeed the UK plc (who need it) and the general 'national interest' (where it is not particularly significant in the scheme of things for the UK, like most standards: ODF probably scores a little more here, one would hope. In India, for example, it is probably much more in the national interest, because their software industry is comparatively important for foreign income.)


Dr Alex Brown (who sometimes comments here) has some recent good, calm blog items with more detail on ISO procedure and so on. In a recent one he commented that we have very little experience in some of these new procedures. I think one thing that is particularly disconcerting is that the Fast-track fast-track compresses and mixes up the order that things are done in, compared to the normal procedure. For example, though fixes are made against the enquiry draft, the ballot resolution meeting votes on editorial changes not the final text: the editor then goes ahead and makes the changes and the ISO/IEC ITTF then checks that the changes are adequately made (of course, National Bodies would be able to bring up problems individually, outside the fast-track process.)


(This lack of experience shows itself perhaps also in the lack of appropriate early feedback or general SC34 position statements on the Ecma drafts before the draft was submitted. I would expect that any subsequent Fast-Track drafts will have a slew of much more definite instructions for acceptability from SC34 informal channels.)

Rick Jelliffe
2007-07-20 00:24:13
Update: Well it looks like I was wrong.


From this report it seems that the INCITS board has decided to recommend that ANSI votes "Yes with Comments". That allows all the comments to progress to the ballot resolution meeting stage at ISO (if it gets that far) but removes the apparant necessity of comments I think are impossible to fulfill. I would have preferred a "No with comments" with no poison pill (as I consder it) but "Yes with comments" is at least progress.


I had email from a committee person, who mentioned that the thing I label as a poison pill (Which is the last ietm on Page 6 of the PDF recommendation from the V1 committee) to say that the subject at issue there was in fact a particular clause concerning a facility to encapsulate existing binary documents in OPC packages to allow import and conversion (re-export of those binaries are banned.) However, I don't see that from the text in the PDF.


However, in fairness, I should take my own comments from a couple of months ago to heart: that the standards process is not like a court of law but like a formalized conversation aimed at agreement. So even if the formal meeting comments were uniluninating on the details of this issue, they form part of a larger conversation (i.e. they act perhaps more like formalized talking points rather than contractual commitments.)

Inigo
2007-07-21 13:07:57
"Of course, examples that are not well-formed because of typos such as delimiter errors or mismatched names or incorrect names is an entirely different issue: my reaction to the Surgey material was that it was good information with incorrect conclusions, because it doesn't cope with elision or tag omission."


(Surguy, not Surgey)


Actually, I attempted to account for elision, and tried to remove all the examples from the list that failed to validate due to missing content but had ellipses present. I left in those examples that omitted content but didn't signal this with ellipses, since I believe those to be confusing (that the majority of examples that omit content do use ellipses seems to indicate that the editors of the spec agree with me).


My only definite conclusion is that there are a large number of errors in examples in the specification that should be fixed; which I think is unarguable.


My personal belief is that beyond a certain point, the accumulation of editorial errors, errors in examples, etc. mounts up sufficiently high to indicate that a specification has not been through sufficiently rigorous technical review in previous standards bodies to be suitable for fast-tracking. However, the lack of clarity around the whole fast-track process is such that I don't expect to find any ISO guidance that explicitly endorses or rejects my belief.

len
2007-07-27 07:14:50
Rick:


I am disturbed by Bob Sutor's statement that IBM is being locked out of the Portugal meetings. I can understand events where because of early interest by one alliance and ignoring it by another, a committee winds up being composed of a lop-sided majority to one side of an issue, but lock outs I don't understand. Being an expert in the processes, how does a chair justify such an action formally and is there any recourse?


len

Rick Jelliffe
2007-07-27 08:05:57
Len: Every national body will have different participation rules and every venue will have different capacity rules. In the initial case, the non-MS chairman from the Portugese government said that the room was at capacity, so the late-comers were excluded. I have no way of judging what the facts were and I don't know what the participation rules and venue capacity rules were.


People can certainly be excluded for unruly behaviour. This sometimes happens when you get an obsessed nutter, or when someone gets hot under the collar, or when someone wants to force their opinion despite the rules, or when someone refuses to abide by the chairman's directions. Sometimes this even results in the whole committee being suspended, as happened in Malaysia, which was a response to supposed rudeness between participants. Things were heated enough that the chairman "Angrily states that only by delicacy they are not sending people away"


Of course, at this point, the culture and personalities come into play as well, of course.


Furthermore, of course, it is not impossible that a party who has been determined to be not acting in good faith in participating in the standards process could be excluded; this is certainly true at the ISO committee level for experts, as the Directives make clear. (I don't understand how, for example, someone who has a "NOOXML" sticker on their website can be acting in good faith as an invited expert at SC34. ODF's Patrick Durusau is a great example of an expert who does not like or want Open XML but still had managed to act in good faith.)


Now I don't know the details here apart from the various reports. But if I stood up at a meeting and shouted at the chairman, refusing to follow his instruction, and then made any kind of public accusations (rather than persuing it directly to the higher authorities first) I would expect to be banned by that chairman. You cannot have that kind of grandstanding in a standards committee.


Note that Sutor references Groklaw. Stephen McGibbon was the MS expert at the meeting, and he has his comments. The comments degenerate into MS v Sun tit for tat, but it is worth reading the Gary Edwards article (complete dynamite) on that page.


I'll certainly add the Sutor comment to by Bribery Watch page. It will be fun to see if any committee manages to escape being accused of some terrible procedural or member irregularity from this.

Rick Jelliffe
2007-07-27 08:32:25
Len: I should also add that many national bodies have decided to take a strong line on bad behaviour: the South African body made a point of saying they did not want to be subject to a big lobbying effort, the Malaysian committee was suspended, and if Portugal has decided to take the same hard line, it may be that IBM got caught out and has been sidelined.


(I note that the Sutor article mentions communications with Portugese, so it sounds like it is an official exclusion, but Sutor does not state the reasons they gave and trivializes it with the chair comment. It will be interesting to read the reasons from the NB. We simply don't know until then: I would personally be delighted if NBs were taking a hard line, regardless of the venom they may be inviting if they do it to the wrong side.)


But, as I said, I only know about the details what I read and what I think about the sources.


On a completely unrelated matter, I am preparing a thesaurus. Len, you say reading reading Sutor's entry has disturbed you. Which word most closely matches your disturbance? A. Fear B. Uncertainty C. Doubt

Rick Jelliffe
2007-07-27 08:58:46
Len: Reading the report of the open source guy more, it seems that Item 7 restricts the committee to 20 members.


RB: Atacks RS claiming lack of respect. Recommends ONS to only accept at most 20 elements in TC


So a new person could come out if an member withdrew. In other words, it is a jury-style committee. That is probably the basis of the exclusion. It is indeed odd and unusual, but I don't know what the Portuguese committee rules are.


It seems that they also have a requirement for a certain balance and number of representatives from industry, government and academia too, which is quite unusual too for people from English-speaking countries. (I wouldn't be surprised for, say, Chinese to reserve some seats for academics, and I know that other standards groups sometime actively seek out academics for ad hoc balance.)

len
2007-07-27 11:10:20
Long answer, Rick. Apologies.


It would be "uncertainty". I've no doubt this is hard ball game nor any fear of the outcome because little will change as a result except that MS will or won't be excluded from some contracts. There is very little chance that will actually happen given the size of the base. The law of gravity can't be repealed.


On the other hand, a clean game does concern me. Specifications and standards have made possible an unprecendented state of integration when compared to the world you and I started in. We don't want to wreck this by having too hard a ball or a cracked bat. We really don't. It undoes the most valuable work of our generation which after all is not the technology of the web, XML, and so on, but the way of creating these through cooperative effort across jurisdictions, company boundaries, competitive boundaries, and international boundaries. The day that happens will be a very bad day. See India/Pakistan.


I don't want to use the phrase 'fair and balanced' for purely national reasons (you'd have to be in the States), but I've been attempting to figure out some rationale for the unfolding events. I spend a long lonely time as you have arguing the case for ISO and that was rewarding in the example of X3D (why it is still the only real 3D on the Web standard regardless; ISO standards last through the lean times) after watching SGML eviscerated (results were OK but the process left a bad taste). Now I am getting feedback from west coasters that the rap on ISO these days is they are a 'good old boys club'. There is an emerging effort to gut newly won respect for ISO to enable investors in some markets to regain the upper hand in sequestering intellectual property. It is a multifront effort and I won't go into my paranoias here but like all such, the forces that converge don't have to be planned, just aligned to get the derivative effects.


The handling of the document standard wars (a much older war than the newbies realize) is something of a public test case for the efficacy of ISO in contrast to or in partnership with the consortia. I've been writing and saying that the partnerships tend to work well because the consortia can handle inner wheel specifications at the pace required and ISO can handle the outer wheel 'grinds slow but small' standards process. It isn't a perfect solution but I've seen it work even when things got hot in the inner wheel.


I asked a question of Sutor that hasn't been answered regards the charges that parties have been stacking committees. I know by experience that when a standards effort is announced regardless of the initating source, who shows up first determines the stack. There are no processes or initiatives that don't have a community of interest that bifurcates along the lines of the various vendor keiretsus. That means when some company or individual has an interest in a product market and the standards that affect it, they have to go to the meetings, sign up for the lists, bear the costs and accept the participation agreements. So I have asked if this is a case where the standard effort for OOXML got underway, was summarily dismissed because of the source, then by the time the competitive interests realized it really was going forward, the rooms were full, the committees were staffed and the opposition was left to suck hind tit. If so this is not a case of stacking but bad strategy, or 'late to the dance and no one to dance with, too bad, so sad, your dad'.


'Die OOXML' efforts are just bumper sticker politics. The web is unsurprisingly easy to inflame given the demographics and a long history of wicked plays on the parts of the pioneers (why I was such a PIA about it in the XML WG days: we set an example many choose to imitate and we pay for it now with discomfort).


On the other hand, charges of bribery or stacking are very serious charges and have to be investigated. If the outcome is the charges are false, industry representatives are left facing legal actions should the accused choose to pursue it. If true, then committee chairs need to be summarily dismissed, wg's reorganized, work withdrawn, and so forth. These are both expensive consequences so professional players should be more prudent with this kind of public rhetoric.


In short, the interests here are not explicitly that OOXML or ODF win. IMO, both lose in the long run. My interests are that international standards efforts have the appropriate respect and be carried out with the required decency and fairness. Given the old phrase, "Truth, justice and the American Way", of late my countrymen have been waking up to the fact that without the first two, the third is impossible and preposterous to claim. I assert one might substitute other processes for that last term and the lesson and the outcomes are identical.

Rick Jelliffe
2007-07-27 14:59:57
Len: "I know by experience that when a standards effort is announced regardless of the initiating source, who shows up first determines the stack. There are no processes or initiatives that don't have a community of interest that bifurcates along the lines of the various vendor keiretsus." Very interesting. I certainly have seen this too; even if it is not a universal rule it is a common event.


Yes, specific charges should be investigated. But the person making the charge needs to provide the details to make their case known: in the case of a committee procedure, they need to reference the committee rules or general administrative law principles that have been violated. If there is bribery, they need to show what evidence they have that money or value was paid to a public official.


But allegations made without the benefit of information about the specific laws, customs, procedures don't provide enough information to say anything about charges, especially in a foreign country. Apart from that the whole thing is regrettable but not surprising. My support for Open XML is not because I think MicroSoft is great: they have joined my team (pro standards) rather than me joining their team, IYKWIM. I know what my prejudices lead me to think, but they are just prejudices.

len
2007-07-27 17:23:39
Wise. You are right, it is not universal but common.


The standards will come and go, the technologies will churn. If the base technologies churn for disruption instead of evolution, the business tiers that depend on the base destabilize at an unsustainable rate. Non-linearity is a self-referencing function. Disruption as a goal fragments markets by pitting system level against system level. It doesn't evolve them. It creates economic spikes for local advantage. Unless the spikes couple, they are noise. If they couple, they amplify and clarity. AM/FM.


Standards are signal. To increase consumability, play well with others to set tastes, but when payment on delivery is the first matter of business, the second is testing conformance to signal so they will sign the check.


len

Rick Jelliffe
2007-07-28 05:31:49
Len: As quoted in my Bribery Watch page, it seems that Rui Seabra is now clarifying that Sun and IBM were denied *seats* on the committee (not chairs in the room) by the standards body, because they already had their quota of representatives. The denial seems to have happened because the committee was organized in a hurry and Sun and IBM did not respond in the time limits.


However, when the meeting was held, all the people who turned up were allowed to join the discussion (and I hope not any voting apart from straw polls) which included some newcomers, but this did not include Sun and IBM people who were not there.


However, and this is the supposedly sinister thing, the Portuguese did not choose to hold the meeting in a room with fewer or exactly the same number of chairs as the expected number of participants, but used a room which had a few extra chairs! What monstrousness is this! Only trust a committee where people have to sit on each others laps, I say!

len
2007-07-29 11:22:19
That sounds right. Late to the party, and no place to park, wired for paranoia and willing to make loud noise but without the evidence to make the charges stick.


This one will go down as a FUBAR but not a scandal. It seems that some are just learning to play the standards game and didn't get the memo about constant vigilance, ie., ready to join a new group and keeping a bag packed ready to go.


Not a high minded strategy, but the inevitable side effect of taking bare knuckle competition for business into the standards arena.

Rick Jelliffe
2007-07-29 20:19:51
Len: Well, if it were in USA, I would hope that Sun would get a seat in that circumstance. I think that realistically two weeks is the minimum amount of time for notice for forming a national evaluation committee if it is a hot subject where people are already teed up, but four weeks would be a more reasonable minimum time otherwise. I don't see that it would particularly be kowtowing to pressure to allow one or two highly competent and interested local participants who narrowly missed a deadline if that is in fact the case.


I don't see much value in flocks of parrots however.


However, I am not so sure that there is so much imperative for a non-US country to include participation by foreign corporations, frankly. Inviting along the champion (Ecma), its editorial representatives and its technical experts (MS) is an obviously necessary thing, but whether *any* foreign companies should be voting is another thing. I know it would be difficult/impossible to do in practice, but I tend to think that *only* local businesses and actual stakeholders should vote: not foreign companies, not people who don't need the standard. Some people complain about stacking of meetings by "proxies" (which seems to mean someone who disagrees with you), but I would tend to think that *only* proxies should be allowed.


A national standards committee should be listening primarily to the local people who want to use a standard: system developers, governments, and so. The opinions of foreign corporations and their rivals are interesting, and the opinions of people who don't want to use a standard are interesting, but for backgrounding and better judgment on decisions and for better community engagement, but ultimately are icing on the cake *if not irrelevant* as far as deliberations at a national level IMHO.

len
2007-07-30 05:58:17
48 hours did seem to be a very short notice, but I don't know the why what or who of that circumstance. Still it seems that the charges flying around the web don't have much merit given the available evidence. I'm not involved with this one but it appears from here that the outcome will be two standards for document formats once the dust settles and they can take their place in the dusty stack in the corner because ultimately, the majority of us are tool users not tool builders.


It is the rage to disrupt that troubles me. I can see very clearly from where I sit and what I manage how a world of never-quite-gets-out-of-beta foundations can make it a lot harder for those of us who build on foundations to prosper. Silly Valley needs to get rid of the trendy consultancy thinking forced on the industry by the investment experts who know everything about a spreadsheet and nothing about software design and production. We're exposed to a lot more danger from that trend than from having two largely unread office document standards. Standards are supposed to be the answer to that problem, but now they are subverted by it. But that's a topic for a different blog.