diff --git a/nucypher/policy/policies.py b/nucypher/policy/policies.py
index 4b14c2f95..008f9138c 100644
--- a/nucypher/policy/policies.py
+++ b/nucypher/policy/policies.py
@@ -376,28 +376,14 @@ class Policy(ABC):
 
         # d = ensureDeferred(self.get_publication_threadpool())
 
-        for node in self.bob.matching_nodes_among(self.alice.known_nodes):
-            d = deferToThreadPool(reactor, threadpool,
-                              network_middleware.put_treasure_map_on_node,
-                              node=node,
-                              map_id=treasure_map_id,
-                              map_payload=bytes(self.treasure_map)
-                              )
-            def llama(result):
-                return result
-            d.addCallback(llama)
-            publication_deferreds.append(d)
-            # d.callback(47)
-        self.publishing_mutex = PolicyPayloadMutex(publication_deferreds, percent_to_complete_before_release=10)
-        def _finish(all_responses):  # TODO: This degree of management is too much to expect of the called function.
+        target_nodes = self.bob.matching_nodes_among(self.alice.known_nodes)
+        self.publishing_mutex = NodeEngagementMutex(f=self.put_treasure_map_on_node,
+                                                    nodes=target_nodes,
+                                                    network_middleware=network_middleware,
+                                                    percent_to_complete_before_release=10)
 
-            self.alice._policy_queue.put(all_responses)  # TODO: Do something more interesting with the responses?
-            threadpool.stop()
+        self.publishing_mutex.start()
 
-        # TODO: errback that stops the threadpool.
-        self.publishing_mutex.addCallback(_finish)
-
-        # return self.publishing_mutex  # I dunno.. return this?  Why not just use the composed version?
 
     def inject_alice_publication_threadpool_into(self, f, *args, **kwargs):
         d = ensureDeferred(self.alice.get_publication_threadpool(policy_id=self.id.hex()[0:10]))