What did he test though? That over a given decade, unemployment might be higher in one nation-state than a collection of others?
What's the hypothesis? Where's the valuable data gotten from this?
My idea of equivalent action: Me stating "the president 10 years from now will be the first 3rd party president ever."
Why? Who knows! And if I was right, and make a blog post that says "See, I was right!" And then point to a bunch of things that happened after my prediction that led to this result, what did we learn? Anything at all?
The point is that these are bets between people who disagree with each other. They negotiate the data and terms and hopefully learn from the result and change their minds. This is one way to help tune bayesian priors about what ideas and people to trust and investigate further.
This is an economist betting that higher regulation causes higher unemployment, and winning. That's a useful step better than simply stating the opinion. It's much worse than a randomize controlled trial, and it would be delightful if the European Union could be convinced to carry those out for science. But employees and employers tend to object to being experimented on.
What's the hypothesis? Where's the valuable data gotten from this?
My idea of equivalent action: Me stating "the president 10 years from now will be the first 3rd party president ever."
Why? Who knows! And if I was right, and make a blog post that says "See, I was right!" And then point to a bunch of things that happened after my prediction that led to this result, what did we learn? Anything at all?