diff --git a/pubeval/REAMDE.org b/pubeval/REAMDE.org
new file mode 100644
index 0000000..773f826
--- /dev/null
+++ b/pubeval/REAMDE.org
@@ -0,0 +1,18 @@
+#+TITLE: pubeval for Python
+
+=pubeval= was originally created by Gerald Tesauro <tesauro@watson.ibm.com> for
+evaluating the performance of backgammon playing algorithms. This Python
+extension is a port of Tesauro's algorithm.
+
+* Usage
+
+The extension is named =pubeval= and can be imported by using =import pubeval=
+after installation.
+
+The only method defined by this extension is =pubeval.eval(race, board)=. The
+arguments to this method matches the arguments required by the original
+=pubeval= function, i.e. the method takes as arguments a boolean specifying
+whether the algorithm is in a non-contact "race" situation and a tuple with 28
+integer entries representing the board in the same format as the original
+=pubeval.c=.
+