diff --git a/.github/workflows/deploy.yml b/.github/workflows/deploy.yml
index 8e6f14b..372875e 100644
--- a/.github/workflows/deploy.yml
+++ b/.github/workflows/deploy.yml
@@ -11,7 +11,7 @@ jobs:
     steps:
     - uses: actions/checkout@v3
     - uses: cachix/install-nix-action@v22
-    - run:  nix build -o _build
+    - run:  nix build -o _build --no-sandbox
     - name: Deploy 🚀
       uses: JamesIves/github-pages-deploy-action@v4.2.2
       with:
diff --git a/src/Measurement_Observation/Binary_Profiling/linux_perf.rst b/src/Measurement_Observation/Binary_Profiling/linux_perf.rst
index e3aad5a..a62e615 100644
--- a/src/Measurement_Observation/Binary_Profiling/linux_perf.rst
+++ b/src/Measurement_Observation/Binary_Profiling/linux_perf.rst
@@ -290,9 +290,6 @@ defined in ``QSort`` that I have elided.
 Inspecting with Perf
 --------------------
 
-Awards
-^^^^^^
-
 To investigate we'll compile and run the program in ``perf``. Perf is a
 statistical profiler that outputs key CPU counters (these counters are called
 `PMC
@@ -972,14 +969,14 @@ Footnotes
        #. :term:`Unlifted` objects cannot be evaluated and therefore don't have
           an entry code.
 
-.. [#] GHC can be built in many different ways which we call ``flavors``. The
-       ``default`` flavor is one such provided by GHC's build tool `Hadrian
-       <https://gitlab.haskell.org/ghc/ghc/-/blob/master/hadrian/README.md?ref_type=heads>`__.
-
 .. [#] I say "typically" because GHC will use ``call`` in some circumstances
        such as creating :term:`CAF`'s, see :ghcSource:`Note [CAF management]
        <rts/sm/Storage.c?ref_type=heads#L425>`
 
+.. [#] GHC can be built in many different ways which we call ``flavors``. The
+       ``default`` flavor is one such provided by GHC's build tool `Hadrian
+       <https://gitlab.haskell.org/ghc/ghc/-/blob/master/hadrian/README.md?ref_type=heads>`__.
+
 .. [#] Larger code size will slow down the program because there is simply more
        code to process. But its contribution to runtime performance is not as
        great as the *locality* between data and the code operating on the data.