Science is a complicated business nowadays. The data sets are large, the measurement technologies are complex, the statistical methods are specialisms in their own right, as also in many cases are the code bases.
It's pretty easy to feel overwhelmed by all this. And I also find that a lot of effort has to be put into guarding against errors and misunderstandings in the work you're doing. This vexes me at some level, so I spend time thinking about how to improve the situation. This is tricky because at some level, the complexity is somewhat innate. But I do think one can adopt a strategy that helps to some degree.
Refine and simplify everything.
What I mean by this is actively (even obsessively) try to refine and simplify every aspect of the work you do. Make your next paper more succinct. Try to come up with a small number of very clear, easy-to-express conclusions. Make your code more compact and tidier. Refine your statistical algorithms so they're not so arcane. Make your data processing procedures as simple and clear as possible.
I think this could be hugely beneficial. Papers would be easier to read and digest. Conclusions would be easier to communicate and develop. Code would be less buggy and easier to work with. Statistical algorithms and data processing pipelines would be faster.
And I think clear, concise scientific ideas are vital. Think about the great ideas in science. There's a purity and beauty to them. So perhaps we should be actively trying to refine and improve the clarity of the ideas we work on, just in case they turn out to be really important.