I have a general life heuristic that I will only ever try to be in the ballpark of optimal. By analogy, I want to get the Big O right, and not worry about the constants.
A picture I have in my head related to this is the scene at the end of Schindler’s List where he’s lamenting that he could have saved more people by selling his car, or his jacket, or whatever. But he’d saved hundreds. I decided I didn’t want to live thinking that way.
Can you find a way to bound your error?
I have a general life heuristic that I will only ever try to be in the ballpark of optimal. By analogy, I want to get the Big O right, and not worry about the constants.
A picture I have in my head related to this is the scene at the end of Schindler’s List where he’s lamenting that he could have saved more people by selling his car, or his jacket, or whatever. But he’d saved hundreds. I decided I didn’t want to live thinking that way.
Maybe see also Slack (or Studies on Slack).
Or also consider that if you optimize too hard you might Goodhart.