1. Be an outsider. Don't be unduly awed by authority,...
2. Be fact driven. An obsession with facts can help you avoid common pitfalls and the bending of facts to fit your theories....
3. Respect the unknowable. Work on the assumption that too many people are too certain about the inherently uncertain things.
4. Always question your own convictions. Most of us look for verification of our own beliefs. Instead you should be open to having your ideas criticised. Only through this kind of stress testing can the true validity of your beliefs be determined.
5. Employ meta-cognition. Consciously practise all the steps above to avoid falling into overconfidence when success is experienced.
Actually, this advice is not about doing science. It is from Kathryn Schutz's reflections Five lessons for investors, from Michael Lewis's The Big Short. I read the above summary in The Week.
But, it seems to me it is good advice also for doing better science!