So much scary stuff in here, but my favourite line is, "When I was in graduate school in the 1970s, n=10 was the norm, and people who went to n=20 were suspected of relying on flimsy effects and wasting precious research participants."
My wifi analyses have n=~millions of data points, and I still have no idea if I'm doing it right. The difference is they pay me either way so I have less incentive to go on a 20-year self-delusion binge.
I like the meta-analysis method where they (to oversimplify massively) check for biased results by seeing if they are suspiciously close to the standardized publishing cutoffs. They can even, using methods I don't quite understand, detect the difference between publishing bias (we don't publish negative results) from methodology errors (eg. filtering outlier data points you don't like). It's neat.