1) Back to discussion on power analysis. I really enjoyed the discussion with "More of a comment than a question" podcast. What @lakens says makes a lot of sense of course, but the sense that I claim statistical power is an illusion is... https://twitter.com/lakens/status/1294168125617762304?s=20

2) obviously not based on questioning statistical theory. I still think statistics was one of the greatest scientific advances of the 20th century, and not only because my father was a statistician himself (as a kid I shook hands with Jerzy Neyman when he visited Brazil once!).

3) The problem I have is about how "day to day" experimental science is actually carried out as a human activity (here Psychology and Neuroscience). As a pragmatist, power analysis is not reasonable because it's harder to properly estimate power than to do your actual study!

4) Turning this around as @lakens proposes to "knowing what effect size you can actually detect" is a good way to go. But even here it doesn't work *in practice* because scientists want to "learn from data", so they actually care about a family of questions.

5) In other words, they are much less interested in "falsifiable science" (which is another illusion in my view), and much more in understanding "structure in data" that constrains the space of theories or conceptual models about the mind and brain.

6) "Good experimental science" has to be done rigorously/quantitatively. But at the frontier of experimental questions, there's only so much we can do. Experimental scientists should stop pretending to be doing standard "falsifiable science". Rigorous science is what we want.