As I see it the less cognitive overhead I have when reading code the better. The more I have to think about the code I am reading the more likely I am to make a

As I see it the less cognitive overhead I have when reading code the better. The more I have to think about the code I am reading the more likely I am to make a mistake because my brain is not a machine. With f" I simply read the code left and right and understand the output as I go. When I see the output I'm interested in I stop and see the context instantly. With .format() I read left and right, stop once I get to what I'm interested in, go back to count the {}s, remember the count, and finally count the arguments. Multiple places to make a mistake. And doing this when refactoring code is how you get bugs.