The productivity gains I'm seeing using Cursor are so striking that I completely lost all faith that any of the negative LLM productivity/quality studies have any validity. I am not an outlier, if any of them were correct I would not be seeing these improvements. This is not some murky questionable finding, it is immediate and obvious with use.