Your specification documents need to be driven deep into the woods and "eliminated."

Spec docs are just an excuse to pretend that you are doing something important. Nobody really needs them, because all that really matters is that the code ships and people like it enough to use it. Developers who get stuff done don't write spec docs. Sure, they take part in the flame wars on the open standards mailing lists, and they might even write a spec after implementing it, but while all this is going on, the developers that really matter are shipping code at the same time.

The worst offenders are the committees of so called "open standards" people posturing in their academic micro environment. And who is impressed? The users? I don't think so. Doug Crockford came up with the JSON standard all by himself. He even joked in a presentation, calling himself "the one man standards body". I think that's the best standards body I've ever heard of.

Go to this link and read it. Its a walking tour of a mailing list where the early web browser developers are kicking around some ideas for what HTML would become. Go there and read it now, and then come back. You'll notice some ideas on there, some good, and some really bad. But you know what else? Marc Andreesen shipped the code for his browser without waiting for everyone else, and it changed the world forever.

You might complain that HTML doesn't really work very well and needs to be thrown away. But, is the big fat pile of steaming poop called HTML5 any better? HTML5 is the camel created by a committee that set out to create a horse.

Dave Winer recently had this to say in a blog post about the OAuth spec:

If you want to get smart about open standards, you have to watch how these things play out in another open thing -- the market. Because it's the market that just as often shapes a standard as it is a standard that shapes the market.

My sentiments exactly Dave.

I think the best strategy is to take the only the good stuff from open specs and then use whatever you can lay your hands on that will create a minimum viable product the fastest. Don't bother with writing a spec, the code is the best spec you could write. So, let's ship some code this year!