This thoughtful post by Christopher Fahey is a great example of how information architects tend to miss the point of their projects: The Holy Grail of Information Architecture. If you read through this post you hear a lot about process and tools and what form the ultimate deliverable from the project should take in order to be most effective. Christopher has concluded that standard deliverables aren’t the Holy Grail of an IA project but he doesn’t quite make the leap to what is: achieving the objectives of the organization.
The point of information architecture should be to help the organization in question to better achieve its goals. The output of an IA engagement should be the achievement of those business goals, not necessarily a nifty flow chart and set of wireframes. Selling widgets, providing membership value, business networking, politics, whatever the purpose of the site: IA should help to achieve those ends more effectively and efficiently. If you can provide that value simply by reviewing and discussing a site in a meeting, great! Don’t get hung up on the deliverables. Do focus on the outcomes to which you are contributing.
C. David Gammel, CAE 
Thanks for the comments. I was really just focusing on the hands-on design aspect of interaction design — which should of course be based on a thorough understanding of business objectives. Even if you have a deep understanding of business objectives and a crystal clear vision of an excellent solution, if you create poor documentation you will have a “weakest link” effect: All of your advance understanding and vision will be for naught.
I agree that you must be effective when creating documentation. But, as you point out in your post, many people leap to documentation before they have really thought through what change they need to help the client implement.