I've been writing software for 30+ years and I think the reason methodologies don't work is that they are simple. Software is an attempt to implement something that will be used by people, interact with data, help make decisions, and much more and the real world is extremely messy. People want software to work one way on weekdays and another way on weekends, and differently if it is day or night, across multiple timezones, and in multiple political locations. They want it to know their workflow. They want it know things people know, like the weather or things in the news or specialized knowledge domains. So I think the main problem with methodologies is they do not address the fact that people use software, and all current computers and software are very poor at fitting into human life, desires, and expectations. AI is starting to change that a little, but there is much more to do before computers fulfill their promise. Methodologies may help along the way, but new inventions are probably a greater requirement. Methodologies are how developers interact with software and each other and management, and they won't solve the problem of how software interacts with people and the world. It's the elephant in the room.