Lots of good ones on here. I'm not particularly good at this either, but one thing that has helped occasionally is keeping multiple tasks running concurrently.

Lots of good ones on here. I'm not particularly good at this either, but one thing that has helped occasionally is keeping multiple tasks running concurrently. With software, among other things, it's very easy to settle into a grove where I find myself getting too comfortable in solving particular problem. So to temper this effect, I switch tasks and work on something else every once in a while, kind of like skipping on a lily pads. A machine learning equivalent is over-fitting. To not overfit is to not feed the brain too much of the same data or otherwise you end up needing to retrain / remove old memory from slots, depending on which metaphor you tend.