OOXML vs ODF, The Real Test

So various committees are trying to get both OOXML and ODF approved as international standards. The game that's being played is who gets to claim their format is more open? Open is good, right?

A complex open standard is only as good as the reference implementation.

Nobody really wants to parse either of these formats; they want code that will do it for them. Software developers rarely start from scratch to build a new application. Rather, they pull together a set of components that gives them a starting point, and then they add value on top of that base.

Also, specifications this complex are full of ambiguity. There's no way to catch and document the right choice in every decision that an implementer is going to have to make. Typically an implementor will see what the existing code does, so it's important that the existing code be accessible.
HTML is an open standard. If you produce a document in HTML, anyone can read it. There is no chance that you will lose access to your HTML documents. It's not a rich enough format to use as the primary document format for a word processor - and even HTML has run into implementation inconsistencies that lead to documents viewing differently depending on what browser you use.

The complexity of the Office file formats is much greater than that of HTML, so don't expect implementers to work from the spec and produce a good implementation. It's going to come down to what components are available, and what code exists that can be used as a reference.

Unfortunately as far as I know there is only one OOXML (Word) editor and only one ODF (OpenOffice.org) editor.

But the fact that one of them is open-source puts a lot of weight behind ODF. It means that if you want a compatible implementation customized for some purpose, or ported to a new platform, all the information you need is available. With OOXML, you're stuck reverse engineering the reference implementation, instead of having access to source code.

The ultimate goal of the standardization process is to avoid vendor lock-in, and without an open-source implementation of the standard, it's really hard to argue that releasing the specification will really change things in that regard. So like it or not, I expect the whole standardization exercise to end in ODF becoming more prevalent in markets that care about standards.