It looks like ~10 is the optimal value of the threshold (because of the way the population wanders away and then comes back). This would not be a random amount, but in fact equal to the amount on the genesis block. Perhaps we should try changing the amount of PC on the genesis block to see if this pattern holds.
These observations are presently inconclusive.
Not sure what the proposal fraction should represent, or what an ideal value might be
Need to write something which quantifies "branching" and "dead branches" in order to understand the visualization better
Probably need to run the simulation longer when there are this many validators. This could have an impact on the evolution results since the population grows over the simulation, so either the run length needs to depend on the number of validators or I need to fix the population size. The latter would be much faster computationally, so might be the better option.