diff --git a/docs/index.rst b/docs/index.rst
index 6126e1f..bdd8aba 100644
--- a/docs/index.rst
+++ b/docs/index.rst
@@ -50,4 +50,5 @@ It copies much of the EDF (EyeLink Data Format) reading code.
 
    usage
    contributing
-   API
\ No newline at end of file
+   API
+   roadmap
\ No newline at end of file
diff --git a/docs/roadmap.rst b/docs/roadmap.rst
new file mode 100644
index 0000000..a780869
--- /dev/null
+++ b/docs/roadmap.rst
@@ -0,0 +1,37 @@
+Roadmap
+=======
+
+This page describes some of the limitations of the current implementation, and
+items in need of discussion.
+
+- **Support for Binocular data**:
+    The current implementation does not support
+    reading binocular data. Adding this support will not be trivial.
+    From a brief look, it appears that our code assumes that each call of
+    ``edf_get_next_data`` will contain sample data (xpos, ypos, pupil size etc.)
+    from a unique time point. But each call to ``edf_get_next_data`` returns data for a
+    single eye, so in the case of binocular data, every 2 calls will return data for the
+    same time point, but for different eyes (left, and right).
+
+- **Support for all sample fields**:
+    The current implementation only supports
+    reading ``xpos``, ``ypos``, and ``pupil size`` sample fields. We should add support for
+    reading other sample fields, such as head position, velocity, etc. This shouldn't be
+    too difficult, but we will want to add a way to specify which fields to read in the
+    API.
+
+- **EDF Message string representation**:
+    The current implementation represents the
+    messages as byte strings (i.e. ``b'stimulus_presentation'``). This was inherited from
+    the original implementation in pyeparse. I'm not sure whether this was to stay
+    compatible with python 2, or if memory was a concern. We could just represent them as
+    regular Python strings, but I will wait to see if anyone provides feedback on this
+    before changing it.
+
+- **Use of Named Arrays**:
+    The current implementation stores data in named Numpy
+    arrays, so that for example you can do ``edf["discrete"]["messages"]["msg"]`` to get
+    an array of the message strings. This was inherited from the original pyeparse
+    implementation. Named arrays are nice but add some complexity (and most users
+    probably aren't familiar with them?). I'm not sure if we should keep them, or just
+    use dictionaries instead. I will wait to see if anyone provides feedback on this.
\ No newline at end of file
diff --git a/eyelinkio/io/edf/read_edf.py b/eyelinkio/io/edf/read_edf.py
index 74efda5..6adb34a 100644
--- a/eyelinkio/io/edf/read_edf.py
+++ b/eyelinkio/io/edf/read_edf.py
@@ -421,6 +421,8 @@ def _handle_recording_info(edf, res):
     # TODO: edfapi eye constants are 1-based, ours are 0-based. Fix this in _defines?
     info["eye"] = defines.eye_constants[e.eye -1]
     res["eye_idx"] = e.eye - 1 # This should be 0: left, 1: right, 2: binocular
+    if res["eye_idx"] == 2:
+        raise NotImplementedError("Reading Binocular data is not yet supported.")
 
     # Figure out sample flags
     sflags = _sample_fields_available(e.sflags)