I agree with you, freejung - so in my presentation I did not end up using much of their redesign as an example of how to do IA well. I really wonder what the whole story is here.
The main point I was making about IA in my session is that it IS difficult, but it is also worth the resources and time invested. If you do IA well, literally everything after that in the site development process flows rather smoothly.
SEO Blog wrote up a short but accurate summary of the main talking points - check it out if you're of such a mind: [seo.com...]
The one thing that write-up doesn't capture is the area of how many passionate disagreements come up on a team that seriously takes on this job. Those honest disagreements are not a sign that you're doing something wrong - it's the opposite. They're a sign that you are doing it well.
The human brain has an essential function of building categories for its experience, but each brain is somewhat unique. so finding an IA that works for the great majority is a tough job, full of trade-offs and compromises. But everyone should know that ahead of time, and every factor should be open for full team discussion. "Pet" ideas can create disasters.