The Education of Gary Edwards

By Rick Jelliffe
September 8, 2008 | Comments: 1

One of the more interesting characters in the recent standards battles has been Gary Edwards: he was a member of the original ODF TC in 2002 which oversaw the creation of ODF 1.0 in 2005, but gradually became more concerned about large vendor dominance of the ODF TC frustrating what he saw as critical improvements in the area of interoperability. This compromised the ability of ODF to act as a universal format.

Edwards increasingly came to believe that the battleground had shifted, with the SharePoint threat increasingly needing to be the focus of open standards and FOSS attention, not just the standalone desktop applications: I think Edwards tends to see Office Open XML as a stalking horse for Microsoft to get its foot back in the door for back-end systems.

Edwards and some colleagues split with some acrimony from the ODF effort in 2007, and subsequently see W3C's Compound Document Formats (CDF) as holding the best promise for interoperability. Edwards' public comments are an interesting reflection of an person evolving their opinion in the light of experience, events and changing opportunities.

I have put together some interesting quotes from him which, I hope, fairly bring out some of the themes I see. As always, read the source to get more info.

I don't want to comment on CDF here, because I still agree with the comments I made on a blog item about 18 months ago ODF versus OOXML: Don't forget about HTML in particular
For simple word processing style documents, if you need interoperability (and you want to get it by restricting the kinds of structures in the document so that the documents can be read by many different applications and be easily repurposed), then HTML is the format to consider first: validated, standards compliant XHTML in particular. Think of it in terms of a continuum, with HTML at one end (simple WP documents), PDF at the other end (full page fidility but read-only): HTML, ODF, OOXML, PDF. And certainly not to forget the ultimate premise(!) of markup: to rigorously label the important information in your documents accroding to its rhetorical and semantic structures, which sometimes simply requires custom schemas and microformats, extending or augmenting or even replacing the standard formats.

Also, I am still totally cynical about snake oil and naive assurances that things that are known to be difficult are easy.:

I guess, behind it all, there is this idea that there will be one true document format. The future will be beautiful because we will have full-potential HTML. Or the future will be beautiful because we have full-potential ODF. Or the future be beautiful when we adopt the same common microformats regardless of the framework. I tend to the view that the future will never be beautiful in that kind of monochromatic ('Stalinist' is entirely too dramatic) way, but that we need to to encourage a rich library of standard technologies, widely deployed, free, unencumbered, explicit, together with the awareness of when each is appropriate and with an adequate set of profiles and profile validators (using ISO Schematron!). Plurality. (HTML browsers are not weaker because there is GIF, JPEG and PNG, let alone TIFF, even though there is almost complete overlap.)

Edwards has one of the clearest comments about the positioning of ODF as a technology that I have seen.

But the purpose of these efforts wasn't exactly to the point of trying to establish OpenDoc XML as some kind of global destiny for all information. (That's phase II :)

Rather, the purpose seemed to be that of crafting OpenDoc XML as a useful "universal transformation layer" sitting between working legacy information systems and, rapidly emerging enterprise level publication and content management systems.

What should the role of vendors be?

After the release of ODF 1.0 (and after IBM joined the ODF TV two days before the release of ODF 1.0), Edwards and associates formed a group confusingly called the Open Document Foundation:

But this is exactly the reason why we founded the OpenDocument Foundation - to balance out the corrupting and undue influence of powerful vendors more interested in goring the Redmond ox and seizing marketshare via government mandate legislation than providing mankind with what we thought was most important; a universal file format that was open, unencumbered, universally interoperable, web ready (compatible and fully compliant with W3C standards) and totally portable - as in application, platform and vendor independent.

When Edwards split from the ODF effort, he was scathing about the domination of large vendors of all kinds, initially voicing concerns about Sun's

For the near five years that i have been a member of the OASIS ODF TC, Sun has opposed any and all efforts to improve interoperability with Microsoft applications, documents, and bound workgroup-workflow business processes.
This goes all the way back to the very first TC meeting on December 14th, 2002, when the enterprise publication, content and archive management systems contingent of the OASIS TC wanted the "proposed" ODF charter amended to include as one of the primary objectives, "compatibility with existing file formats and interoperability with existing applications".
And yes, that proposed charter change specifically included compatibility and interoperability with Microsoft applications, documents and processes!!
Sun opposed that change and has consistently opposed all interoperability enhancements since.

Then subsequently on IBM

IBM has to discredit us because anything that damages ODF or helps MS-OOXML hurts the IBM business plan of lobbying for legislative mandates requiring the rip out and replace of MSOffice. The thing is, we don't care much for the IBM business plan. Our focus is entirely on the needs of a marketplace trying to break the iron grip of relentless and determined monopolist.

Edwards pointed out

The last official ODF TC vote i reviewed had Sun with six votes, IBM with three, and one independent.

I verified this voting concentration myself in a recent blog which was partially deleted unfortunately: going back more than a year, I couldn't find any ODF TCs where the voting membership was not dominated by a handful of vendors, basically Sun and IBM. This is in sharp contrast with the early period of the ODF TC up to the introduction of ODF 1.0, and has rather coincided with a period of stagnation or slowness by the TC: I suspect this is also because of a desire to make sure that ODF 1.2 is a much better standard than ODF 1.0.

The point I drew from this recent domination was that more participation by other parties was needed, and I have been pleased to see a little more variety in recent rollcalls (Update: E.g. see the much better balance in the latest meeting's rollcall.) But nevertheless it goes back to an old point of mine: that "openness" needs to be able to avoid both crypto-monopolization and cartel-ish technology-exclusion behaviours in committee. I call this Verifiable Vendor-Neutrality. I think my readers will be pretty aware that standards bodies and competition watchdogs are acutely aware that standards bodies must avoid both cartelization and single vendor dominance; however, standards that are made from scratch intended for universal use clearly have different aspirational criteria than standards made to document existing technology: in the former case, the technology is under development while in the latter the technology is relatively fixed: exposing it in a way that disentagles it from other proprietory technologies is the name of the game.

Too pro, too anti?

Edwards' early comments on ODF pushed the line that ODF was ready to take over the world. For example, in this comment in 2005 you would think that compatibility was total:

The OpenDoc TC was very fortunate to have a wealth of expertise in reverse engineering the legacy maze of incompatible MS binary file formats. Experts from Corel Office, StarOffice, Boeing, Stellent, ArborText, and SpeedLegal among others had long made their living reverse engineering MS file formats.... To be honest, i don't know of any Microsoft worker bees old enough to be packing the level of expertise that was routine on the OpenDoc TC. If Jean Paoli has a MicroSerf in mind who can stand toe to toe with Phil Boutros, i would pay for a front row seat.

Since the first 18 months of the OpenDoc TC's life was spent on legacy issues, most of which concerned MS file formats, it doesn't surprise me at all that they cloned the OpenDoc specification. But then to try to take credit for the enormity of work the OpenDoc TC put into the transformation process? Who are they kidding? The least they could do is send a thank you note and give credit to the real experts who actually did the work: Phil Boutros (Stellent), Paul Langille (Corel), Tom Magliery (Corel), Simon Davis (Australian National Archives), Jasson Harrop (SpeedLegal), Daniel Vogelheim (Sun StarOffice), Michael Brauer (Sun StarOffice), Doug Alberg (Boeing), Paul Grosso (ArborText), Patrick Durusau (Society of Biblical Literature), and David Faure (KOffice-KDE).

The frustration had set in by mid 2007

There is no possible way anyone can claim that today's OASIS ODF TC would welcome Microsoft and make accomodating changes to the specification! No way! And the proof of this hostility can be seen in the actual disussions and rejections of Microsoft specific interoperability proposals.

And by the end of 2007, Edwards had to admit admit what others of us had been pointing out

The simple truth is that ODf was not designed to be compatible - interoperable with existing Microsoft documents, applications and processes. Nor was it designed for grand convergence. And as we found out in our five years participation at the OASIS ODf TC, there is an across the boards resistance to eXtending ODf to be compatible with Microsoft documents, applications and processes.

Now, I don't want to go overboard in the extreme: the idea that ODF and OOXML are completely incompatible would be just as risible position as that they are completely substitutable. They have different features and approaches, but these will not effect all documents, and even where they do, we can hope for graceful degradation in many cases. The conclusion that Edwards came to was that there were five key areas where ODF needed to be upgraded, and that by sticking to a subset of ODF with these five extensions, you could get pretty good interoperability. Here is a quite lengthy selection from comments to a Jesper Lund Stochholm blog:

IBM's plan of "ripping out and replacing" MSOffice isn't going to work, as amply demonstrated by the year long Massachusetts pilot study. The subsequent year long plug-in pilot similarly failed because of the incompatibility between the way OpenOffice implements the five incompatible document structures; lists, fields, sections, tables and page dynamics (breaks), and the way MSOffice implements those same structures.

We see the same problems in the Belgium and Denmark pilots. Without a subset of ODF geared to compatibility with MSOffice, ODF is pretty much useless to the over 550 million desktops that represent 94% of the market. No posturing by Rob Weir or anyone else for that matter can overcome the fact that ODF was not designed for the conversion of MSOffice documents. Nor was it designed to be interoperable with the existing MSOffice installed base and the bound business processes they rely on.

From July 12th of 2006 through February of 2007, there were five ODF iX "interoperability enhancement" approaches introduced to OASIS. None of them passed or even made it through discussion.
If workgroups are able to swallow the disruptive cost of "rip out and replace", and make that transition to ODF, they still face a single vendor future. Which is exactly what they were trying to avoid in the first place. Because ODF compliance is optional and interop a mess, the only solution left for users desiring to implement ODF is to standardize on a single ODF application. A situation that seems to work fine for big ODF vendors, who are all pushing their own particular ODF applications.
This is hard for many to accept, but ODF has serious interoperability problems. It simply wasn't designed for the conversion of existing MSOffice documents, or interoperability with existing MSOffice desktops. Go ahead. Try it and see for yourself.

The conversions problems however can be effectively handled by an ODF subset designed exactly for the conversion of MSOffice documents. Couple that with a much needed interoperability framework, and ODF can be made to work. We're talking five generic elements here! Not the thousands of pages of application specific elements and attributes that would no doubt result if Microsoft were to somehow join the ODF TC.


It is difficult to talk about ODF and standardisation without mentioning OOXML!

Edwards, intimately involved in ODF since its inception, ended with a position which can be contrasted with another long-term ODF participant with independent credentials (i.e. not just a suit), ODF editor Patrick Durusau, who remains passionately committed to the ODF project. Durusau has a history of standing up for freedom in standards (I think I have mentioned his participation in the successful effort against the OASIS Rights Language attempt by MS, IBM etc circa 2004, culminating in the boycott call against OASIS. Yes, even OASIS has controversy!

Edwards comment, previously quoted, is something that I would not be surprised to hear from Patrick's mouth:

...providing mankind with what we thought was most important; a universal file format that was open, unencumbered, universally interoperable, web ready (compatible and fully compliant with W3C standards) and totally portable - as in application, platform and vendor independent.

But my understanding of Dr Durusau's position is that for ODF to succeed, the standard needs to be uncompromisingly excellent and complete, and that IS29500 (OOXML) can provide an useful source of information that will only help strengthen the ODF effort by providing more information.

As the editor of OpenDocument, I want to promote OpenDocument, extol its features, urge the widest
use of it as possible, none of which is accomplished by the anti-OpenXML position in ISO.

I gather that during Durusau's review of DIS29500 he become more aware of the hard technical differences, though he has been constantly scathing about the draft and its inadequacies, and the need to push through better processes at JTC1. His recent view in Not with a Bang is typically salutory against those who drum up unnecessary partisanship

The campaign against Open XML was at its start, in the middle and at the end an anti-Microsoft
campaign. The merits or demerits of Open XML were simply a convenient launching point for
criticisms of Microsoft.

The ODF TC was founded by Sun's Jon Bosak, who also founded the XML effort at W3C. He gave Sun's official attitude to OOXML as this

We wish to make it completely clear that
we support DIS 29500 becoming an ISO Standard and are in complete
agreement with its stated purposes of enabling interoperability
among different implementations and providing interoperable access
to the legacy of Microsoft Office documents.

Sun voted No on Approval because it is our expert finding, based
on the analysis so far accomplished in V1, that DIS 29500 as
presently written is technically incapable of achieving those
goals, not because we disagree with the goals or are opposed to an
ISO Standard that would enable them.

I didn't see any sign of Edwards changing his position that Office Open XML should not have been made a standard (and I think his position is a quite reasonable one, just not mine), though he does want to see a conspiracy in Sun's public position. I think it is simpler: Sun people like Jon Bosak have a genuine understanding that open processes have to be open to everyone, not just the people you like; and they are aware that the anti-trust law also reinforces this, competitors who try to use standards processes to lock out a competitor (even if on the most excellent and innocent of motives) are on shakey ground. And I expect that Jon Bosak and his team are certainly under no illusion that in order for ODF to succeed, it must succeed as ODF the implemented technology not ODF the standard: ODF cannot 'win' by goodwill but by extensive coverage by multiple implementations.

I used to think that some of the stridency of the anti-OOXML movement came from the some negative aspects of the writing personas of the Open Document Foundation folk (Edwards, Marbux, Hiser), however I am having to revisit this suspicion: their departure from the scene has not resulted in any increase in calmness (indeed, I think Edwards experienced first-hand the level of intolerance against deviations from the party line from some quarters), and re-reading many of Edwards' posts his comments (though initially rather starry-eyed in the ODF honeymoon, and perhaps idealistically held to longer than sweet delusions should be) have many good insights and raise many serious issues. Not the least of these issues is the folly of glibly seeing that world as black hats and white hats.

Sometimes, when I quote Durusau's comment that we need to get to a position where we accept that reasonable people can reasonably disagree, I get pounced on with claims that I never give any credence to anyone with a different view than mine. While I hope this is not true, nevertheless I am aware that I tend to respond to cranks more than civilized people. In that spirit, Gary Edwards is clearly someone who I disagree with on most of his conclusions, but I think I share most of his premises.

I think we differ in that I see engagement with corporate stakeholders as absolutely vital (though I think Edwards did increasingly come to see the need for on- and off-ramps for transitioning technologies and for mixed environments, rather than probably doomed big-bang substitutions) though this needs to be tempered procedurally to prevent vendors (rather than users or nations) from having effective control of proceedings, which is the utter opposite of real openness. From an old blog of mine:

It is not bad that large corporations try to influence standards developments: a standard is an agreement and they are key stakeholders. Standards help promote and channel innovation. But what is good for General Motors in not always good for America. ... A standard that is too far away from the strategic or profit interests of corporations large or small will not be supported by them, unless pressure can be applied from outside or unless good people inside the company weedle it in.

You might also be interested in:

1 Comment

Any very large emotionally laden event will reverberate past the initial set of stimuli or people who set it in motion. That is endorphin addiction pure and simple. The human brain rewires itself based on co-incident occurrence (see Hebb throty) and given some statistically large group sharing the occurrences, it becomes self-organizing and self-sustaining. Sometimes I refer to this as beating the Mean Drum although it is just as true of the kind chord.

Rationality is a weak predictor of human behavior. It isn't necessary for the us vs them, good vs evil or spy vs spy models to play out consciously. Because of the way the brain wires itself up, the players not only don't hold the same viewpoints, they physically can't.

Enlightened individuals, in the sense of self-awareness within awareness of an environment, learn to rewire their own brains. Others don't.

What we ascribe to social movements or even technological movements are sometimes nothing more than a statistical event of similar wiring patterns reinforcing the behaviors that propagate their own inner models to others. A meme is not a thing. It is a wiring diagram being shared.

News Topics

Recommended for You

Got a Question?