Thoughts from my boss, in response to this article:
"Make as many mistakes as fast as you can, as fast as possible without hurting yourself or any of your stuff. The number of mistakes required to achieve success is a fixed number, to the first order. To the second order, the number of mistakes needed is proportional to the time between mistakes. Thus, there is a mistake rate at which no progress is made. For example, if we made one unsuccessful attempt at a task per year, and eight attempts were required to achieve success, it's very likely that success would never occur. This is especially true if the forgetting of lessons is considered. If a lesson is remembered and effective for only five years, then we'd always be three attempts away from success. Also, make mistakes at the small scale as fast as possible, where the political, economic, and capital cost are also small.
Successful engineering occurs when the executor fails to build the object incorrectly."