simulate.py

#!/usr/bin/env python3

# Simulator for https://fivethirtyeight.com/features/who-steals-the-most-in-a-town-full-of-thieves/

import random
import statistics

def simulate(k):
  assets = [100 for i in range(k)]
  for robber in range(k):
    victim = random.randrange(k - 1)
    if victim >= robber:
      victim += 1
    assets[robber] += assets[victim]
    assets[victim] = 0
  return assets

def gini(assets):
  total, lorentz_area = 0, 0
  for a in sorted(assets):
    total +=  a
    lorentz_area += total
  return 1 - 2.0 * lorentz_area / (sum(assets) * len(assets))


def main():
  data = [simulate(1000) for i in range(10000)]

  ginis = [gini(d) for d in data]
  print("gini: {:4.4} {:4.4} {:4.4}".format(statistics.median(ginis),
                                            statistics.mean(ginis),
                                            statistics.variance(ginis)))


  buckets = 75  # bucket together collections of houses for visualization
  for b in range(buckets):
    outcomes = []
    for d in data:
      outcomes.extend(d[len(d) // buckets * b : len(d) // buckets * (b + 1)])
    print("{:>2} {:4.4} {:4.4} {:4.4}".format(b,
                                              statistics.median(outcomes),
                                              statistics.mean(outcomes),
                                              statistics.variance(outcomes)))

if __name__ == '__main__':
  main()