> The key evidence to me that full stack is mostly BS is that it only crops up in web roles. Scientific computing / Data Science / Machine learning is an area

> The key evidence to me that full stack is mostly BS is that it only crops up in web roles. Scientific computing / Data Science / Machine learning is an area that benefits from people knowing the full stack. Siloed environments, where the scientists don’t talk to the engineers and the engineers don’t talk to ops are especially poisonous to this type of work. Research and science requires a lot of trial and error and the compartmentalised waterfall style of development doesn’t work well.