Hi Ardo, thanks for the technical comments! We fixed the LaTeX typo and updated the BTC and ETH wealth figures to make clear they’re conditional on a $500k threshold (i.e., what is the centralization of wealth given users greater than a given wealth threshold). If we set that wealth threshold to $0, then the Gini is trivially 0.999.. because most folks have no BTC or ETH.
> Generally a single point estimate is rarely enough to describe a distribution.
Totally agree regarding point estimates, but the post was long enough and we wanted to get the basic ideas out there :). It should be possible to generate a distribution over Nakamoto coefficients given various assumptions to see how robust it is within a given subsystem.
Ultimately, the choice of subsystem decomposition and thresholds (51% or more) depends on one’s notions about which kinds of compromise attacks are likely.
