diff --git a/paper/TODO.org b/paper/TODO.org
new file mode 100644
index 0000000000000000000000000000000000000000..c3dd37010678fc4992157391a7d2d996baf3d93e
--- /dev/null
+++ b/paper/TODO.org
@@ -0,0 +1,45 @@
+* TODO Requested Revisions
+  1) [X] Add names to the rules
+     1) [X] names everywhere
+     2) [X] make it not ugly
+  2) [-] Revise Intro
+     1) [X] List of Contributions
+     2) [X] Better overview of eta and the counter examples.
+     3) [ ] Have co-authors review
+  3) [-] Related Work
+     1) [ ] Degen-Thiemann
+     2) [X] Henglein
+     3) [ ] AGT
+* TODO Clarity
+  - 4.1 call-by-push-value is dense
+  - multiple reviewers: there isn't a "we did this" part of the intro
+  - numeric citations as nouns: L1025 and "elsewhere".
+  - 108: *dynamic* gradual guarantee
+  - "AGT is based on operational semantics more than it's based on
+    axiomatic semantics, but it's primarily based on typing—*static*
+    semantics.  I got the point eventually, but it was a little
+    jarring."
+  - CBPV mnemonic: liFt, thUnk
+  - You correctly saw the need for a section to give background on
+    CBPV, but it's also important to clearly summarize the two papers
+    you're generalizing.
+* TODO Proper Attribution
+  - Blame soundness as a concept is not due to Wadler and Findler, but
+    to [Tobin-Hochstadt and Felleisen 2006] , although you may be
+    thinking of a more specific result here from Wadler and
+    Findler. The argument in this paragraph (the continuation cannot
+    affect the result and thus does not deserve blame) is very similar
+    to the criteria given in "blame correctness" by [Dimoulas et al
+    2011].
+  - Our work on constructing models of the dynamic type should cite
+    some of the obvious stuff on type tags, type Dynamic, etc.
+* TODO Response Issues
+  1) L514 is unclear, and talk about completeness a little.
+  2) multiple blame labels
+  3) Lines 116-117, typed equivalences in static portions of code
+  4) Gradual Languages where ? doesn't exist: this is actually a
+     strength of this paper
+  5) restrict to non-effectful programs: 
+  6) "particular equalities one might want to reason with while
+     programming"
+  7) Eta is only used in one direction