Writing software is more like writing books more than anything else. So any process that presupposes that the workload can be divided among roughly equal agents

Writing software is more like writing books more than anything else. So any process that presupposes that the workload can be divided among roughly equal agents is bound to fail. That's my experience after two decades as a software engineer. Every good software team has two or three engineers who knows how to write books, how to structure a story, and can work towards a common vision. The rest of the team more often than not consists of tag-alongs that don't know nor care about good literature. Every good software project lets these two or three engineers get stuff done. Every bad software project suffocates these engineers with processes and/or makes them quit.