The ISO/IEC JTC1 committee on Document Description and Processing Languages has had an interesting couple of years, what with all the controversy and new members. But for the last few months, all ballots have failed due to lack of votes. Not lack of “yes” votes, just plain lack of votes. The reason is, obviously, because most of the committee members of new National Bodies (standards organizations from participating countries) that have nominated to be active on SC34 issues are only really interested in two issues: ODF and Open XML.
No interest in anything else; no vote on anything; nothing else progresses. It is easy to castigate NBs for not living up to the obligations they signed up for, but the more productive way to approach this is that the customer is always right: SC34 needs to have some reorganization so that people interested in just one thing don’t stymie other projects. When various SC34 people were discussing this over the last week, I suggested that perhaps what was needed was a completely new SC to handle it: that would leave SC34 free to continue with “enabling technologies” rather than application technologies. A less radical proposal came up, which was merely to make a new Working Group (WG4); apparently NBs can nominate which standards or WGs they are interested in, so this can allow partition. Something will happen.
I was very amused to see this on my second favourite trash website:
Is further proof needed that these are not good-faith members? How to get rid of these MSFT stooges?
So lets actually look at the NBs that failed to vote, and compare to their vote on DIS 29500 (Open XML) as far as I can figure it out. I’ll use the ballot results on the recent NVDL draft corrigendum as an example.
The countries that voted Yes on DIS 29500 but failed to vote on NVDL DCOR = 12:
Bulgaria, Côte-d’Ivoire, Cyprus, Egypt, Kenya, Kazakhstan, Lebanon, Malta, Pakistan, Sri Lanka, Turkey, Venezuela,
The countries that voted No on DIS 29500 but failed to vote on NVDL DCOR = 9:
Brazil, China, France, India, Korea, Norway, New Zealand, Thailand, South Africa
The countries that abstained on DIS 29500 but failed to vote on NVDL DCOR = 2:
Chile, Trinidad and Tobago,
So both sides are equally slack. What did you expect the result to be? If you expected it to show that the MS stooge countries were pretty bad, while the valiant anti-OOXML forces were pretty good it shows you have drunk the Kool-aid, with all gentle respect. Some-one says something based on no objective evidence, but if it accords with what has been said enough times before, people think “That sounds about right”: but what if what was said before also had no objective evidence, that there is a chain or ripple of make-believe and demonization that merely emotifies foregone conclusions?
Actually, I think the whole way of thinking about this in blame terms is unhelpful. It is certainly very frustrating for SC34 members to be treading water: it means real money and effort wasted by those involved. May you live in interesting times! is the curse we are suffering from :-)
But ultimately this is just a problem of poor organization at the NBs by the NB bureaucrats: they should be checking that their local committees are indeed voting on the ballots before them, and they should be sending in abstain votes otherwise. It is ultimately a matter of mechanism.
So we have a confluence of three things. First, we have some new National Bodies who need to get their ballot governance procedures into shape: I think we should be highly tolerant of the antics of fledglings. Second, we have a committee structure that does not reflect the interests of members: there are some very mature and deeply involved NB who didn’t put a vote in, not just newcomers; the WGs need to be organized to reflect this division of interest. And thirdly, despite the previous two things, when the national bodies signed on to SC34 (particularly the “participating” P countries more than the Observer “O” countries) they did incur an obligation to vote, and they need to pull up their socks right away and participate: even if it means just sending in a official “abstain” on each ballot, that doesn’t derail the process.
"...a chain or ripple of make-believe and demonization that merely emotifies foregone conclusions?"
The term for that once in vogue in psychology circles is "superstitious learning" and the common example was the belief in spontaneous generation (e.g., maggots are created by rotting meat). This was my objection to very large integrated unfiltered hypermedia in the 1980s, it being a known problem of broadcast systems that amplify by repetitive linking. Probably the most noticeable example is Google Gaming. Any eigen-vector indexing system has this problem although it is assumed that as the numbers go up, the effects smooth out the errors. It doesn't work as any pollster can explain. In the words of Mark Twain, "It isn't what we know but what we know that just ain't so." Frequency, amplitude and scheduling of divisive messages cause small shifts but when fed back to high energy decisions (most effect across most domains at highest rate), it is a very effective means.
The main problem is the amount of energy required to shatter lock-in where the belief persists despite sufficient evidence to invalidate it. This is why chairs should and do have lattitude in otherwise consensus-bound processes. Of course, in a high energy decision, charges and counter charges focused on the chair result.
There isn't a fully automated process that can fix this that won't introduce either a) more opportunities for mischief (same as software bugs) or b) exhaust the system (entropy). Human respect for institutions and processes are required to guard them from abuse.
The problem has been pointed out many many times before and after the web. This is a case where for some, the bug became a feature. Caveat emptor.
Mark Stephens (aka Robert X. Cringely) made a little career out of this, without the internet -- writing up rumors about Microsoft backed by nothing but his "anonymous sources" (whom he never even partially identifies) and appeals to "known patterns of behavior" (based on his own articles).
So 9 of the 15 no voters,
12 of the 17 yes voters
and 2 of the 9 abstaining countries on the DIS 29500 vote.
Look like the countries abstaining on the dis 29500 vote were actually the more active countries...
hAl: I just wouldn't read too much into there being a strong connection between vote on OOXML and slackness. For a start, some countries vote no with comments because they want to ultimately vote yes, and other NBs vote no on the draft expecting they won't change their mind.
Also, every different NB has different rules and standards culture, and it is quite possible that the procedures and culture of one country might produce different results from a given range of opinions: we have seen that Sweden uses something like unanimity for its definition of "consensus", while the UK procedures perhaps favour a weak majority as long as issues are passed on from the NB with the ballot.
It may be that NBs who voted "abstain" would tend to be the ones where there is a good awareness of procedural options, and perhaps this indicates that their Secretary and NB teams are well up to speed. But I wouldn't even put that forward with any confidence. Every NB is different.
The only thing I am certain of is that coming up to speed in standards is a slow process: the self-righteous finger-pointing of (some of) the extreme anti-OOXML side may have been exposed by the fact their side's failures to vote is just as bad as the pro-side's failures to vote, but this is not a moral issue. Probably many NBs haven't had regularly scheduled meetings, or would be happy to pull out when DIS 29500 is done.
Some committee people on NB groups may not have even known that if they want to play standards they need to broaden their horizons: if they want to talk about XML documents they need to understand schemas, or fonts, or topic maps as well (or the other areas that SC34 has working groups) because one hand washes the other. There is an aspect of standards work that is just work: engaging in technical discussions even where their subject is not of compelling interest to you in order to move the committee agenda forward.
Rick,
I'm not that clear on the details about the difference between a subcommittee and a working group, but surely your previous comment implies that SC34 needs to be kept as unified as possible, so that the failure mode when confronted by a herd of newbies is a standstill (which can't produce anything at all) rather than an uneducated democracy (which can produce bad standards that can never be unproduced)? It seems like any attempt to satiate people's desire for laziness would necessarily be against everyone's long-term interests.
- Andrew Sayers
Andrew: A committee (TC or SC) has voting national bodies as voting members, but delegates and invited experts (inluding editors and liaison people) for discussions. It usually divides into areas of interest, which are Working Groups. Working Groups just have people as their voting members, delegates and invited experts. When an NB becomes a P country it can nominate only to be interested in certain areas, e.g. a particular Working Group. That way it doesn't hold up other Working Groups.
On the issue of newbies, I was a newbie, and while they often don't know procedure and folklore and the particular SC's bigger agenda well, their technical expertise and enthusiasm should not be considered second rate in any way. All welcome.
It is not a matter of laziness, necessarily. For example, Norway did not vote recently, yet it is one of the most active NBs in SC34 and universally respected. It just tends to be busy on its own area (WG3) and not much interested in the other WGs. That is fine when there are only a few NBs doing that, the system allows some slack; but when too many NBs do it the quorum rules kick in and it is a sign that the WG or SC structure no longer matches the interests of the members. The thing is that the SC and WGs exist to serve the members, not the other way around. So it is the job of the Secretariat to remind members, but it is also the job of the participants to suggest alternative organizational possibilities that might be workable, because the intent of the system is to bring about agreements. If the particular SCs and WGs get in the way of agreements, they needs to be re-jigged. I guess I see the current problems more as a cost of doing business rather than a crisis: the road has bumps.
And standards can actually be unproduced! In fact, every ISO standard has a five year review and may be withdrawn if it did not get uptake or has been superceded.
Rick,
Thanks, that's very informative. I still think I'm missing one or two facts though, as I can't quite get my head around the situation.
As to SCs and WGs, I think what you're saying is: there exist SC votes that are voted on by NBs, and WG votes that are voted on by individuals nominally representing their NB. NBs joining the SC are automatically added to all WGs, but can opt out. Therefore, by opting out of all WGs, an NB can disqualify itself from everything but the final vote for/against all standards of the SC. If that's accurate, why create a new WG? Also, what are the chances that new members just assumed they had to explicitly opt in to WGs they were interested in?
The "customer is always right" approach is at the same time totally sensible and totally bizarre to me, so I'd like to get your opinion on the logical extreme of the argument. Let's say the newly joined NBs formed a WG which decided that applications conforming to DIS 29500 must assert that the world was intelligently designed, and that a conforming application must include an animated assistant which uses the words "it looks like you're discussing the origin of species...". Obviously that would be distasteful on a number of levels, but would it constitute a crisis, or just a sign that the current membership is out of touch with the world's interests?
- Andrew
Andrew: The JTC1 Directives 3.12 have this:
3.1.2 A P-member may have an interest in the field of JTC 1 without having interest or competence in all of the work items which may be dealt with. In such an instance, a P-member may inform the JTC 1 Secretariat, the SC Secretariat and the ITTF at the beginning of the work, or at a later stage, that it will abstain from participation in discussion or voting on specific items. Such a position, established and recorded by JTC 1, shall entitle the P-member to be absent from meetings and to abstain from voting on the relevant FDISs.
I think this means that their vote would automatically give them a vote of abstain. Now an "item" here means a standard (or draft or technical report etc), and the idea is that the NB rather shouldn't have to say "Record Abstain for this" on each individual item, but should be able to tell the Secretariat "Record Abstain on everything from WG1" or whatever.
On your question, an SC's formal votes are always based on National Bodies, whether for ballots or at plenary sessions in meetings. In SC meetings, informal votes and straw polls can be done on any basis up to formal votes, and in fact the SCs encourage liaison with industry groups, user groups, other committees on different standards organizations, non- and semi-government agencies, etc. The more the better, as long as they are credible. For example, at SC34 we have had representatives of Maths societies, Biblical societies, SGML Users Group, and so on. Experts can be invited who are not affiliated with anyone in particular too: I was initially Australian delegate to the precursor of SC34 and now am an invited expert in my capacity as editor of the Schematron standard.
It is a very welcoming environment, though white men with beards have a quite high representation; the new blood may provide more diversity there, and Japanese Standards Association is taking over the Secretariat from Canada, so I think there will be more meetings over in our hemisphere, which will encourage participation too. But standards are one of the few places where nationality is not a bar: the editors or technologies of the ISO DSDL standard's parts come from residents of Japan (RELAX NG, NVRL), Australia/Taiwan (Schematron), Thailand/UK (RELAX NG, NVRL), UK (DSRL, NS DTD), USA (DTLL), Switzerland/Japan (CRDL) but they all have other input too (ISO Schematron folds in some Portuguese work, etc.)
As to what new members think, I am sure they think all sorts of crazy things! Just humans.
Your question is based on an analogy, because it is a technical committee not a religious one, though religious especially scholarly religious concerns are indeed treated with complete respect and interest. Every analogy breaks down. If there was a WG which did anything against ISO's requirements for neutrality and respect, the item would not even come up for national ballot, and I would expect the WG might be suspended.
But allowing the absurdity of the case for point of argument, the WG only decides what gets into the draft for the voting NBs, so it is always the NBs who decide on what is accepted, not the WG. If a WG goes to far from NB acceptability they are wasting their time. The trouble with the fast-tracked standards is that it evades the WG and goes straight to NB review and ballot, because the committee work in developing it has been done externally.
But crisis? No. Sometimes you do get runaway WGs, where there really is not the international interest in a technology. And sometimes you get one country that really hate a standard while the others are happy with it. Because the ISO standards are a library of technologies rather than a set of exclusive laws, if one set of NBs dislike a particular standard, they can get the technology they like standardized too. It is not first-come, first-served. The JTC1 annual general meeting a fortnight ago in Australia spent a time discussing this, and I am told is putting out some material to explain about the need and utility for multiple standards (e.g. for graphics) because of the recent fanciful quotation of the ISO material by anti-OOXML activists.
So the customer is right because plurality is allowed. This is why the idea that there should only be a single standard for anything is so wrong-headed: it works against plurality, and managed plurality is the benefit of standards: you might still have four different types of screw heads in dozens of standard sizes, but you don't want either just one type (ban Torx or ban Phillips head?) or one hundred types. The purpose of standards is to enable people to do things, not to prevent them.
If the Chinese wanted to bring UOF as an ISO standard, they probably would not get it at the moment because it is fairly PRC-specific. However, if UOF became popular in Singapore and HK as well as PRC (and if it became popular in Taiwan) and was worked into a form that was useful for Korea and Japan, and had some binding so that Latin names were possible (i.e. using ISO DSRL) then I don't see that it wouldn't be a suitable candidate for an international standard too.
The trouble with standards is that there are not enough to choose from!
But all this is OK because there is another side to the coin: governments and regulators making technical regulations to adopt standards. Just because something is a standard does not mean (0%) that should be mandated uncritically. I have said multiple times that I think governments should adopt HTML, PDF and ODF as the formats for public websites (and that OOXML can be allowed too as long as one of the other formats is provided) and that the drivers for ODF are different than for OOXML as a standard. People think that standards are regulations, but they are not (though there are some nations where there is indeed a closer connection for historical reasons: I was told Malaysia's standards regime grew out of the rubber industry for example.)
The talk of crisis is spin: it is intended to create fear, uncertainty and doubt. All that has happened is that a committee is being frustrated because of member slackness, disinterest and confusion, and there are procedures for sorting it out. It is just a cost for SC34 being plunged into the spotlight; once we were a little country cafe with regulars popping in for a chat and a steak, but suddenly we are Michelin-starred with all sorts of vegetarian restaurant reviewers coming in and making judgment :-)
Rick is right again.
There is an emerging effort on the surface from several domains to propagate the idea of single language unification. After all, that is what HTML is, right?
This is the wrong approach for many reasons. It creates an unhealthy ecosystem. Even where there is one language for a period of time, requirements for it diverge sufficiently to cause either so many variants that they may as well have separate titles or to eliminate areas of application that look similar on the surface but are quite different in the domain in which they operate (eg, design time vs runtime datasets).
The single language approach has a very long and public history of failures. One only has to look. XML succeeds precisely by NOT being a single language approach. SGML stalled because of having too many variations in the approach to a syntax unified system. HTML bifurcates because of too many necessary variations in the object frameworks that support it and the plethora of scripting languages that animate it.
This single language meme is stacking up to be one of the Big Idea discussions in the next year given the numbers of ostensibly separated emergence points I am seeing across various blogs and forums. Taken to extremes, it is a very bad idea. Not considered in terms of the technical reasons it is advanced, it is a dangerous idea to ignore. Not understanding that some of the motivations at some of the emergence points are not technical in nature makes it insidious.
Rick,
Thanks again for an excellent explanation. I still don't understand what's going on, but I think that's more to do with needing time to mentally adapt to the culture of the ISO cafe, rather than needing more details about today's dish of the day. In particular, it seems like you're saying that the ISO works on a principle of adding enough compexity and redundancy to muddle through, rather than building a lean and efficient system that forces a "right" conclusion to any given situation. I have a lot of respect for the robustness of that approach, but it means that one can only comment usefully if one is ignorant enough to judge the system on its output, or knowledgeable enough to understand the psychology of the system. I guess that means I'll be lurking for a while in discussions about the ISO's behaviour, but I may hassle you again next time I fall into one of those groups.
Len,
To add a little to what you've said, it seems to me that the office tools industry completely reinvents itself once every 10-20 years, for example with the introduction of calculators, computers, and the Internet. The thing that gnaws away at me when I hear people arguing that there should only be one standard is that, even if format X is provably perfect today, what happens if we throw away all the knowledge about format Y, which then turns out to be perfect when the next big thing comes along?
To be honest, I haven't really made up my mind about the issue of plurality. I do though think that the field is lumbered with enough legacy issues already, and I think it's worth looking to minimise the legacy issues we'll create this time around.
- Andrew
@Andrew:
IMO and said elsewhere, the OOXML/ODF debates are a bitter butter battle over dodo standards. Office documents based on print formats are nearing the end of their shelf life. The politics are interesting in the sense that a lot of damage is being done and contrary to what you may read, not simply by Microsoft. All of the big companies and small who dived into this for business reasons bear responsibility. It really isn't a good idea to unleash attack dogs on ISO processes. The outcome really isn't worth it. IBM will take some harsh karma on this regardless of how good they feel today. People are getting wise to it and customers are getting very tired of it. The MS customers stand to gain if OOXML is fixed and goes through. We'll see how it works out.
But the emergence of the 'single language theory' again after two decades of Internet work is surprising. This is an approach that always failed in the past, and as the saying goes, trying the same failed plan again and again expecting different results is the very definition of crazy. If a market has dozens of competitive designs, it is an immature market. If it has a handful, it is a hot competition. If there is only one and a half (one of which is dieing or trying), then it is a graybeard market about to die.
Language systems really are the closest analog to the so-called ecosystem metaphor. They do evolve like living entities. Any good biologist can tell you that cultivation of living entities requires a few elite cultivar and a wild from which to harvest new genes for breeding the cultivar. A single language proponent misses the point that without the wild, the species tends to die off and rapidly.
The ecosystem metaphor doesn't really apply to technology networks. Those tend to be more like city-state evolution where the continuing existence is based on the strength of the contracts and contract enforcement. See Rome. A more modern term is keiretsu. Some of the single language meme is coming from the new 'interoperability forum' folks in virtual worlds (led by ... surprise! IBM) who want to create a keiretsu of interlocking company deals to manage and control Web 3.0, the so-called 3D Web based on virtual worlds. It's an amazing thing to watch and not just a bit distasteful after all the hoorah IBM set in motion over ODF.
I'm a VRML/X3D builder, so I have a language preference, but it would be crazy to suggest it is the only language to use for building VW (not a good runtime format for very large connected worlds - see Forterra paper on page terrains) but a good design time language and has excellent life cycle qualities. When you look into plurality, you have to understand that each language specializes in one way or another and that specialization limits its application. X3D is an ISO standard. The new keiretsu ignore it but the authors who have suffered losing their vendors when they go out of business or are bought know that the Language IS the Platform as far as the author's right to own their content goes. Again, it isn't the whole solution though. Large runtime systems need the optimized binaries, something the XML community consistently refuses to acknowledge in some places and that damages the credibility of the XMLers in such discussions.
Horses for courses.
No matter what you manage to write about this topic, the fact still remains that SC34 functioned just fine (with 90% votes passing) before the rush of P-upgrades and now doesn't function at all. If that's completely unrelated to the P-upgrades, then how did SC34 function before and why doesn't it now?
Asbjørn: You are just trolling. Where do I say it is completely unrelated to P-upgrades? I specifically mention new NBs. It is the dumb presumption that the blame falls along voting lines with goodies and baddies that is incorrect, on the numbers. If the numbers showed something different, I wouldn't say this. But crocodile tears for SC34 by people who undoubtedly don't have the slightest wish to get their hands dirty and participate themselves are pretty bogus.
Asbjørn seems to have a valid point that you don't actually disagree with. 8 of 11 last minute P-upgrades did not vote after OOOXML came to ballot. 9 of 11 last minute P-upgrades voted for OOOXML.
They are single issue P countries, and that seems to be causing problems.
I agree that single issue P countries tend not to vote (or don't vote by definition) and that the most recent entries may be single issue countries (really, you cannot tell this on a single round of voting!) However, most of the previous entries over the last year were also single issue P countries, and there are more of them than the recent ones (11 on my count). And they have failed to vote more than the newcomers.
So it is completely silly to put any special culpability on them, that is not shared by the other non-voters. Why is a national body that has failed to vote in one round worse than a national body that has failed to vote in two or three or four rounds? The only reason that these NBs are being selected is to prop up the extraneous narrative that latecomers are bad, in order to discredit them and their votes.
But my disagreement goes further: the issue of looking at everything in terms of blame. A National Body has the right to join any SC they like, and the SC needs to try to accommodate their interests.