The point that seems to completely elude 37signals is that not every industry operates the same way as web design and development. They've taken their very narrow range of experience and told us that it applies to all projects in every industry (he cites Denver International Airport and the Big Dig in his post). This is such an outlandish notion that it becomes difficult to take them seriously at all.

As an example, I work as a design engineer in the R&D wing of an automotive supplier. If we were to just start building products on a hunch, without any market research or project planning, we could easily spend hundreds of thousands of dollars or more before we knew if we were headed in the right direction.

Ultimately planning is imperfect and comes at a cost (time and resources), but that needs to be compared to the cost of no planning at all. In the web app business the cost of trying a new idea may be next to nothing, so trial and error becomes the most expedient method. In my industry, we're aware that our plans are imperfect, but if they allow us to cut a few very expensive iterations out of the development process then they're worth it.