The OOXML debate.

Over the past few weeks I have followed the OOXML debate from a distance. If this was 15 years ago, I would probably have spent some weeks reading it and then made some noise about it -- but these days I don't have the time. According to people in the know it is somewhere between 6000 and 8000 pages (depending on whom you ask). Additionally, there's all the bureaucracy surrounding the process that one would need to understand. To be quite honest, that alone is more complex than I can justify spending time on. Especially since I will probably never have to deal with this standard. There are standards in the works that are far more important to me than OOXML.

What has struck me about the debate is how woefully irrelevant it has been. The whole OOXML debate has been a giant waste of time. It is pointless. Why? Because it has not been about the merits of the specification itself. For some reason the whole debate has been largely about bureaucratic technicalities and unpleasant allegations.

Over the past couple of days I have talked to a lot of people who have involved themselves in the debate. I talked to Håkon Wium Lie, to Martin Bekkelund and to countless other people who have followed the spec, the political manouverings, or both.

I told them about my perspective and how childishly simple the whole issue is from my point of view. It can be summed up in one sentence: "make a hard-copy of the standard and tell me that it seems realistic to expect someone will be able to create a reasonable implementation".

Would you be able to digest the spec? Really? Can you, with complete confidence, tell me that you would be able to read the thing and understand the complete picture to such a degree that you would be able to architect an implementation?

I seriously doubt I could even motivate myself to try. Life is too short.

It doesn't really matter if it is 6000 or 8000 pages to me. It is simply too massive a standard to be of any use. I say this as a programmer. I could care less about the politics. A spec of that size represents an amount of information and complexity that is so massive that it would be foolish, bordering on laughable incompetence, to even think that it can be implemented fully and with a degree of correctness that can even approach the goals we pretend it aims for: complete , seamless and trouble free interoperability between different office solutions.

And this is without taking into account that the standard seems to be an attempt at describing the structure, semantics, and behavior of an existing application suite: there will be missing pieces, there will be quirks.

More importantly: there will be one reference implementation and it will not matter what the spec says because the reference implementation will be what everyone has to be compatible with.

We can pretend that it will not be so, and there are people who really do want to believe, but we have to be realistic.

I would have liked the debate to be about how you design and realize a document format for the future. I would have liked for people not to be in such a hurry. I would have liked for people to focus beyond the current market situation and to make an effort to see 20, 50, perhaps 100 years into the future.

I would really have liked the debate to have been one that discussed the merits of technology rather than being about who can endure and exploit bureaucracy the best and who is better at playing political games. Microsoft won because this is their game. This is what they are good at.

Yes, it would have taken 10 or perhaps 20 years to arrive at a sensible spec. And even that is optimistic. Given a more sensible process we could possibly be able to have parts of it working earlier than that.

And yes, we should have taken the time to do so. I know of standards that are far, far simpler than OOXML, or even ODF, which have been in the works for very long periods of time indeed. I can't say I like this, but it takes a lot of time to develop ideas properly and it is only counterproductive to entertain ideas that the process can be hurried.

But if you want a simple argument: just look at the sheer size of OOXML and tell me you get a warm fuzzy feeling at the thought of implementing it with a useful degree of correctness.