I think there is a lot of counters you might want both.

For example, supposed you have a list of ids accessing some system, you may want to count the raw number, but then also have some transformations thereof that are still ordered by count.

Example

  l = getListOfAccessIds()
  counts = Counter(l)
  total = np.sum(counts.values())
  fractions = {k:v/total for k,v in  counts.most_common()}

  #use the fact new dictionary is still ordered by most common
  plt.semilogy(fractions.values())
  plt.plot(np.cumsum(fractions.values()))

  #still works like dict
  print(fractions[id_of_interest])