Even if you are not writing the algorithms themselves, being able to know when to apply a certain approach or select different libraries based on their implemen

Even if you are not writing the algorithms themselves, being able to know when to apply a certain approach or select different libraries based on their implementations (and the tradeoffs they come with) is useful. I feel like I'm endlessly re-re-re-learning various different algorithms and problem classes though whenever I got hunting for a library. Like, algorithm goes into brain, problem is solved, 3 months later it's pushed out by something else. As long as things work, there's no stickyness.