This is true of startups too. You launch to learn what you should have built.
Paul Graham notes that startups, like computer science, involve launching to discover what should have been built. He references Fred Brooks' observation that debugging specifications is central to programming.