The Frog Race: The desire for control and how large companies interact with standards organizations

by Rick Jelliffe

I was told recently that of the 250 or so fast-tracked standards that Ecma has successfully had accepted by National Bodies at ISO/IEC, only three of them have failed. I thought it would be interesting to read up a little more on them.

Ecma (shooting the messenger)

Ecma makes standards on a wide variety of subjects, and has particularly strong involvement with the European and Japanese computer hardware industry. In a response to a comment on another item, I posted this list, which is of the current groups and chairman's affiliations, to give an idea of its scope:

  • C# (Chairman from Microsoft)

  • ECMAScript (Chairman from Mozilla)

  • Business Communications (Chairman from Siemens)

  • Near Field Communications (Chairman from Sony)

  • High Rate Short Range Wireless Communications (Chairman from Sony)

  • Environmental Design Considerations (Chairman from IBM)

  • Accoustics (Chairman from HP)

  • Electromagnetic Compatibility (Chairman from Intel)

  • Optical disks and disk cartridges (Chairman from Toshiba)

  • Universal 3D (I3D) (Chairman from Boeing)

  • Holographic Information Systems (Chairman from Fujifilm)

  • OOXML (Chairman from Microsoft)

  • XPS (Chairman from Global Graphics)

Now I knew that the C++/CLI effort had failed (for what seems good reasons to me.) But I was not so sure of other efforts.

I found this article, from 10 years ago: Sun Uses ECMA as Path to ISO Java Standardization which I will look at in more detail in a moment. But there is an interesting passage halfway down the page:
In 1996 Microsoft Corp was able to shoot down another ECMA standard, the Public Windows Initiative, at this stage, thus preventing it from becoming an ISO standard. The PWI was a Sun effort to get Windows APIs put into the public domain. ... Microsoft was able to mount a successful campaign against PWI at ISO on this issue.

What do we learn from that? That Ecma was happy to serve as a neutral forum. That Sun was happy to try to make use of the Fast-Track procedure when it suited them, for competitive reasons. That in fact IP buy-in from the critical stakeholder is necessary. And that MS has made a 179 degree turn on standards since a decade ago. (I am always amused at how often anti-OOXML material will, when it fails in a current objection, resort to decade-old material as if it were fresh and compelling. The company then was fleeing standardization; now they are participating and allowing significant changes. You do not have to trust or like them to acknowledge that.)

Control of the API

ISO standards are a very scary proposition for large companies. Many of them are not comfortable with any position other than dominance and stability. The control of the API is terribly important to them, and they regard loss of control of the API as a risk (whereas it can be a circuit-breaker and new-market enabler.) This is one reason why all the large companies try to favour the member-based boutique standards bodies: W3C, OASIS, Ecma, because there is more chance that they can establish a beachhead and make participation at those bodies unattractive or futile for their competitors. The need for stability is sometimes stronger than the need for dominance: when you see calls for "equilibrium" to be maintained in a market, you know that is a buzzword for maintaining the status quo. (And it is not always the market leader: it can be a smaller player in fear of losing their share just as much.)

It goes in cycles. The wheel turns and sooner or later the big companies are forced to deal with ISO and national bodies, and they find this lack of control very unpleasant. Sooner or later they find some reason to split back to more dominatable bodies, and they jump ship.

It is not all venal (or even venial) or negative though: for example, look at SGML: Sun's Jon Bosak (and many others) were unhappy with the way and speed that SGML maintenance was proceeding and we went to W3C as a forum for making a simple profile and addressing a lot of peripheral issues, and XML in turn became the foundation for the update of SGML. There is always an interplay between what the boutique, specialist bodies are interested in, and what the national-body-based regimes such as ISO are interested in: industry activity is actually really important, because it clarifies what the ISO groups should be doing.

The downside is that when these large, usually-US-based multinationals hop over to their boutique bodies, they have to try to justify their jump by slagging off at ISO/IEC. This is a predictable behaviour: it has happened in the past, it is happening now, and it will happen in the future. Some parts of the complaints are often reasonable, some parts are often merely self-serving, but it is not a new behaviour.

Ecma and Java

Now back to Java. Originally Sun put up Java to become an ISO standard using the PAS process (the fast-track process that ODF used) using the Open Management Group (another boutique group) as the submitter. Then Sun changed its mind and decided to submit it to become an Ecma standard (and thence to ISO on Fast-Track) because

In examining our standardization options, our primary goal always has been to preserve the industry's substantial investment in evolving and using the Java technology," said Dr. Baratz. "By paring the collaborative Java Community Process with ECMA's proven standards process, we can achieve international standardization while preserving rapid innovation and cross-platform compatibility.

According to this article Sun chose to go with Ecma, because it was flexible enough to allow maintenance to continue on through the Java community process as it stood then. Other articles suggest that one reason for Sun's reluctance to be involved at ISO was their strong desire to keep effective control. One particularly interesting aspect of the article is that it mentions the potential danger from Sun's point of view of HP, Microsoft and so on doing exactly what Sun had attempted to do with PWI: make up their own version of the standard and submit it to ISO!

Of course, what Sun was concerned about was Microsoft's attempts to destroy Java's Write-Once, Run Anywhere promise by grafting on their own graphics primitives into J++ and splitting the market. This is of course how IBM put a nail in Java's coffin for the desktop, by doing exactly the same thing with their SWT graphics library, as used in Eclipse: it is not a part of standard Java and Java applications that use it are not WORA applications.

The fight between Sun, IBM and Microsoft over their effective graphics libraries shows a couple of things that are very instructive. For a start, it shows that they all try to use standards for their own competitive purposes. It is no news: the challenge it to try to use the standards process to channel them into behaviours that benefit society and the market.

It also shows the futility of non-layered standards. The WORA spiel is really compelling, and it is something that I bought into with my company Topologi, but all systems that have to grow need to support what I call Organic Plurality. Systems with modularity in the wrong spots die but can cause problems in their death throws: it seems that with Java, the graphics interface was exactly such a spot, unfortunately for the vision. (For another aspect of this, see The Software World of 2010: Its about the Suite.)

But thirdly it shows that the big players have been involved in these kinds of standards games for years. For a while, and under the noxious impact of the MPEG group, the large companies got excited by the idea that they could use standards bodies to become revenue-generators by standardizing on Royalty-bearing technologies.

Pigs at the trough

In the middle part of this decade, there were attempts at OASIS for this, and many of us spoke out against the large companies trying to do this, and we were successful. For people with short memories, the background of this was the attempts to get non RAND-z technologies adopted for DRM proposals: the major pigs with their snouts in the trough at that time were ContentGuard (ex Xerox), Microsoft and IBM, all the usual suspects. (Readers may also be interested to note that Patrick Durusau got involved in the OASIS DRM effort, on the side of the angels: he has a very hard-headed attitude to all the large companies, and not one that endeared him to Microsoft or IBM.) By 2004, the OASIS DRM group wound up without getting this endorsement for the non-RAND-z technologies. RAND-z won!

David Berlind has quite a good article on why a non-RAND-z standards organization is a "patent shelter" and not open: it is great that OASIS has straightened up here, and I hope SC34 continues its long-standing RAND-z policy. But it is especially great that companies like Microsoft, IBM and Sun, which a few short years ago were all excessively concerned with trying to keep control and use standards as patent-shelters are behaving well now. However, just because Microsoft, IBM and Sun have little credibility in the world of standards for altruism's sake, it does not mean that they should be blocked from participating legitimately in standards. To the contrary, we need to have institutions to allow these behemoths to act as good citizens: RAND-z standardization is a great vehicle for a behemoth!

The futility of monocultures

Back to our Java story. In late 1997, SC32's Java study group had recommended that Sun should submit Java through the "more traditional" processes. Sun eventually did shift to use the Ecma route, but apparantly out of fears it would lose control. Then

.In another effort to block other companies and interests from developing Java platforms that do not meet its strict guidelines, Sun Microsystems on March 1, 2000, declined an offer from ECMA to standardise Java. ECMA, which is a standards organisation in Geneva, Switzerland, denounced Sun because the company refused the standardisation proposal. TechRepublic

Industry gossip was that Sun wanted to make their source code a normative part of the standard and they withdrew when they found it would not be possible through Ecma (or ISO or anywhere!): nice try fellows! I'd love to get some confirmation or another angle on this. But clearly the issue is one of control: integrity, interoperability are all nice side-effects. The trains always ran on time under Mussolini: we should not pretend that centralized control and monocultures do not have some benefits.

However, when we look at the way large companies act with respect to standards bodies, one very large question should arise: it is a variant on Adam Smith's aphorism (or was it G.B. Shaw) that every profession turns into a conspiracy against the public interest. If monopolistic, cartels and collusive behaviour are undesirable (I don't use "wrong" here because it carries a moral implication which distracts people from the point and lets them drink from the waters of Lethe from the sweet cup of self-righteousness) because they result in sub-optimal market operation.

So why are standards allowed: surely they are collusive, and interfere with the market?

Public policy

The traditional answer is that public policy encourages standards because and as far as they create markets. When the Torx screwdriver company got its hexagonal screwdriver heads adopted as a standard, they may have been wanting to encourage a market in screws not competitors in screwdrivers, but they were creating a market none-the-less. OASIS lawyer Andy Updegrove, who I criticize a lot for his flakey reporting and bias, has really good legal material at his website which quotes the (U.S.) Fifth Circuit Court of Appeals decision in Consolidated Metal Products v. American Petroleum Institute in 1988:

A trade association by its nature involves collective action by competitors. Nonetheless, a trade association is not by its nature a "walking conspiracy", its every denial of some benefit amounting to an unreasonable restraint of trade. In particular, it has long been recognized that the establishment and monitoring of trade standards is a legitimate and beneficial function of trade associations.

One key aspect of the setting of standards is that they cannot be needlessly exclusionary: this is why there is always the need for multiple boutique bodies, because when a company is unable to get satisfactory inclusion of its technologies or requirements because existing members have "stacked" the process against it (and it should be noted that this is a negative stacking aimed at blocking: there seems to be no such thing as stacking a standards body in favour of a legitimate technology, quite the reverse: a standards body is there to foster agreements) then that company can go elsewhere. The need for a market in standard technologies requires a slew of supporting markets, including a competitive market for member-based standards organizations. (It's turtles all the way down, as the joke says!)

When we get to ISO/IEC JTC1 we run out of competitive standards bodies. At the international level, there is quite a clear difference between the kinds of work that, for example, IEEE takes on and the work that ISO takes on. So if allowing plurality rather than blocking is at the very core for justifying standards (I mean voluntary technical standards used by industry, not regulations or which side of the road to drive on) as market-creators and preventing standards from being feet-in-the-door for cartels, what happens at the apex, at ISO/IEC JTC1 for example, when there are no competitor bodies?

The answer is simple: plurality. ISO/IEC cannot be in the business of allowing cartelization, since the only justification for standards is because they actually prevent cartelization by creating markets.

Trapping a bear

From this light, I hope my support for OOXML getting standardized even though I recommend ODF for public government documents, becomes clearer. The need to support plurality goes to the very heart of the mission of international standards bodies. It is one thing to speak of technical issues, it is another to blanket state "We already have a standard that is good enough for us, therefore you don't need the standard that you think would meet your needs". Because that is just code for "We want to prevent your technology for operating in its market by limiting the market to our favoured technology". That kind of blocking behaviour needs to be exposed and rejected.

The large US multinationals have always been trying to use standards bodies to compete, and they have always shopped around, and none of them like giving up control. The recent defection of some of the leading lights of the Open Document Foundation away from ODF springs out of exactly this issue: the charge that Sun has tried to keep too much control. They all try to play this game, it is not new.

So what can we do? We have to be like bear trappers. The bear is bigger than us, has an off-putting odour, and a taste for honey. But when the bear wanders into a cage, you don't say "Oh, Mr Bear, you are too big" or "Oh, Mr Bear, you stink" or "Oh, Mr Bear, all you want is to raid the honeypot, such a naughty and greedy animal does not deserve to be trapped!" You close the trapdoor and jubilate. The history of these large companies is that they all try to find the route where they can maintain the maximum control, and very often they will get skittlish at the amount of control they have to give up. Even Ecma, which is polloried at the moment for being some kind of a rubber-stamp, would have required giving up too much control for Sun with their Java effort: and you would not want to think that Ecma were necessarily the most accomodating here.

A lot of the anti-OOXML material over the last year has been along the lines "Don't you know how bad MS is" spouted by companies who have been playing exactly the same kinds of games. Think SWT, think DRM, and so on. But standardization can be a real game changer: one of the few game-changers on the horizon. The chance to capture a large mass technology into the review and influence of the international standards organizations comes very rarely and IMHO is not a chance that should be squandered on petty ideological or competitive points. Open Source millionaires and closed source search engine companies, all of them are in the same boat as the rival office suite developers: competitors with vested interests to block the development of multiple markets.

The thing is that competition between these kind of standards is not just good, it is essential. I have just been looking at the new feature list for OpenOffice 3.0, due mid year, and it finally includes tables in Spreadsheets. Now it has been incredible to me that this has not been there before: I don't know how you can make a presentation without tables. But tables in spreadsheets was not something encouraged by ODF before OOXML came on the scene. (It is not a feature suggested for spreadsheet applications in the informative feature table in ISO ODF, in particular.) And the recent changes in OOXML have surely occured in part to catch up with ODF: it is not one sided. The competition is forcing each technology to be improved in places that their original champions did not consider important.

Given the utterly toxic relations between the various players at the moment, which makes any talk of sitting down at the same standards body ludicrous, what we need is frog race. Rival technologies whose stakeholders are attempting to leapfrog each other, but with each jump taking them closer to the goals we have set: open standards, with better QA, harmonized and mappable where possible, supporting plurality, extension and adequate profiles, with decent validation and test suites. The anti-OOXML side tries to claim that the best way to openness it through enforcing a monoculture, but the experience of the last two years, and the substantial improvements in the ODF and OOXML technologies that have occurred and are pending are clear indications that standards need to harness the competitive energies of the stakeholders rather than dissipate them in prolonged committee-room chicanery aimed at maintaining the current "equilibrium".


2008-03-25 07:51:21
Interesting read. I don't see one piece of the situation: standards and specifications bodies rely on members for support and continued existence. The adaptability of the body to the needs of the members and the mission to support a large or small community of users of the products using the standards and specifications can be at odds, and therefore, the survival of the organization can be imperiled.

That was the problem of what Jon Bosak did. By taking SGML with or without permission to W3C, he set the example followed by Sun with ECMA. When one organization doesn't do as asked or demanded, the bigCo finds another that will or as in the case of 3D standards and Intel, they create their own and obligate their own customers (middle tier resellers for example) to join it. Whether or not an individual or company has the right to take the standard being managed by one organization and move it to another at whim is a different question, but one the legal experts would want to examine given that in these organizations, IP has been offered under the rules of membership, therefore there exists a quid-pro-quo relationship that should be protected by law. The example we set with XML is legally dubious even if the particular instance is clean. It isn't something one should try without serious consideration of possible litigation.

That is the standards game at its most cynical, yes, but money on the table is a strong solvent.

A means of adaptation is to create different kinds of memberships with appropriate affordances for the kinds. This balances survival with mission integrity.

Rick Jelliffe
2008-03-25 08:21:27
Len: I take your point, but in the case of XML and SGML I really think everyone was on the same side, ultimately. And here we are discussing the arcana of rigorous markup of competing views of office documents: exactly some of the things that Goldfarb was interested in at the start of SGML (rigorous markup, not restricted to spurious universal formats but allowing different models and custom information-specific markup.) The W3C effort took great pains to keep XML as near to being a subset of SGML as it could, and SC34 reciprocated by upgrading SGML to fit XML where there were problems.

But industry profiles have enormous credibility, and rightly so. I don't really see that the XML/SGML experience is the same kind of boat as your experience with the competing 3D standards.

2008-03-25 13:24:05
Going long...

It's the example it set, Rick. Originally, it was very difficult to get the HTML community interested in standardization at all. Then it was only with the organization they trusted (W3C + SGMLOpen (now OASIS)) because ISO was pilloried. Then when it came to the XML work, it was claimed it was 'impossible' to do that in ISO and besides, ISO was irrelevant because of "Internet time". Yet the private posts revealed the animosities that built up over time. Jon carefully sidestepped those where he could then threw the spec to the floor without explaining that everything taken out would have to be put back elsewhere in even more voluminous packages. It was spin and we all knew it.

I'm not questioning his integrity. I'm saying it set an example soon to be repeated: if an organization's policies don't work to the benefit of some/one driving initiative, it's ok to go to another organization with the same material. Thus the game descended into what it is today with the addition of politics where character assasination in public has become standard operating procedure. "As the twig is bent..."

The 3D experience is a little different. For the XML example to repeat, a consolidated market has to exist. For HTML, there were no other realistic choices given the tangled technologies. For XML, the movement was in place. I doubt that can be repeated.

What has happened in the 3D space is that the standard was created early, the market pancaked, then when it began to grow again, it was too tempting to try to claim it as property. The mistake was SL didn't consolidate the market. It has become like nightclubs on a strip. Once one opens, others follow. Because 3D is a big application space, different markets are tangled up (eg, games and virtual worlds) and none can come out on top.

So now despite IBM's best efforts to spin up an "industry interoperability forum", only a few bit and then they were pushed aside. SL is the big loser there because IBM is more interested in partnering with Forterra, a CIA venture capital firm. It makes sense in terms of their markets but it doesn't do anything for the standards. SL is left hanging and the virtual world market is probably in 'recession' for awhile. Certainly consolidation efforts in industry profiles are dead for a season.

Meanwhile, the ISO 3D standards remain viable, live and still evolving slowly but surely. The question of whether or not a standard is wanted is of interest. The question of whether or not one exists is settled. Now it is a technology race which is where markup found itself post HTML and pre-XML.

We got XML done and we got consensus in the community eventually but it wasn't pretty and it wasn't without damage. IMO, the biggest damage was done to the standard processes themselves. We set examples that have been followed to this day. Some good; some not so good. Can standards recover? They do but they won't ever be without the kind of conflict you describe if they have high value.

OOXML vs ODF has become the example of how all sides can injure themselves, their customers and the infrastructure on which they rely by failing to acknowledge plurality and precedence, but ultimately in my very provincial sense of this, decent manners and reasonable compromises. Not all, of course, but enough that it gets harder with practice, not easier. We got out of it because we had practiced consensus for a long time and we kept the really nasty bits private. I don't see that happening in OOXML vs ODF and I think it has become yet another bad example.

That said, I stand by what I've said before. Michael can weep for it in his keynote speeches, others can dismiss it, but essentially what we did was a burglary in the beginning and that as the process worked out, everyone consented, doesn't change that. Michael claimed I was saying we don't honor SGML enough. I don't say that. I said/say we didn't pay enough attention to the precedents we were setting. Not being able to acknowledge SGML is a symptom of the goals private and otherwise that became public memes.

In the end OOXML will be an ISO standard. ODF may or may not flourish, but the damage by example and collateral is done. I'm not sure either matter now.

Andrew Ziem
2008-03-27 14:15:09
Before 3.0, you could still display tables in using less attractive methods such as pasting a spreadsheet object. (There may be a nice workaround, but I never looked into it.)

BTW, thanks for the link. :)

2008-03-28 20:09:29
I was looking for something else and granted myself the pleasure of re-reading this post one more time. This is one of the clearest expressions I have ver seen of the importance of standards as devices for encouraging good-citizenship among corporate players in an industry.