I completely agree with everything you've said. I think that an interesting way to gauge the accuracy of the model without doing major empirical research is to create models where talent is the decisive factor, and then compare which model's assumptions better capture what we know about the world. My guess is that for a variable to be a decisive factor, it must correspond to excessive control over others, thus not matching our colloquial definition of talent, but rather of power.
> I think that an interesting way to gauge the accuracy of the model without doing major empirical research is to create models where talent is the decisive factor
Exactly. This paper should be one among many models explored, but it's a good start.
In fact, I'm sure the paper's simulation can be run with a whole variety of parameters for luck's probability, talent, and bad/good luck's effect on capital. We can then see which set of parameters best matches some aspect of reality and possibly gain some new insight there too.
