Add script to experiment average number of draws needed to get Bob.

Intuition says it will be roughly half, and looks to be confirmed here. However,
analysis patterned from Section 5.2 (Indicator Random Variables) of CORMEN3
suggests that it should be ln N. I wonder where my analysis is wrong.

experimental-statistics/exchange-gift/bobfinder.py

@@ -0,0 +1,22 @@
+#!/usr/bin/env python3
+
+import random
+
+class BobFinder(object):
+
+    def __init__(self, shuffler=random.shuffle):
+        self.shuffler = shuffler
+
+    def find_bob(self, participant_count):
+        pool = ["Participants %s" % i for i in range(participant_count - 1)]
+        pool.append("Bob")
+        self.shuffler(pool)
+        return pool.index("Bob") + 1
+
+def experiment(participant_count=40, runs=1000):
+    bob_finder = BobFinder()
+    bob_posn_sum = sum([bob_finder.find_bob(participant_count) for i in range(runs)])
+    return bob_posn_sum / runs
+
+if __name__ == "__main__":
+    print(experiment())