Distribution of power-law exponents for various fitting methods. Power-laws are fitted to a coalescence and fragmentation process simulated with size-biased kernels as in (2, 3), M = 10 4 , F = 0.3 , and 20,000 samples. We write MLE and KS-MLE as in the main text, “PL” for power law, “Range” for finite range, “Exp” for an exponential cut-off, “R” for the poweRlaw package, “Python” for the powerlaw package, and “Num” for the numerical maximum likelihood estimation in the bbmle package. The...