4171

1 files changed, 7 insertions, 7 deletions

diff --git a/Readme.md b/Readme.md
index e1ef1a4e..9df19440 100644
--- a/Readme.md
+++ b/Readme.md
@@ -84,7 +84,7 @@ low-level pointer operations and manual memory management, but much safer,
 paying some run-time overhead to validate pointers. It also provides a number
 of features usually associated with higher-level languages: strong
 type-safety, function overloading, lexical scope, generic functions,
-higher-order functions, and delimited continuations.
+higher-order functions, and [delimited continuations](http://akkartik.name/coroutines-in-mu).
 
 Mu is currently interpreted and too slow for graphics or sound. We're
 actively investigating ways to speed it up. In the meantime, Kartik uses it to
@@ -249,12 +249,12 @@ The names of the variables are important here: a function that takes an
 (immutable) address and returns a different one is different from a function
 that takes a mutable address (and also returns it).
 
-Immutability checks can be annoying sometimes, but the benefit they provide is
-that you always know what a function modifies just by looking at its header.
-They provide all the benefits of [referential transparency](https://en.wikipedia.org/wiki/Referential_transparency)
-that we typically associate with (particularly purely functional) high-level
-languages -- but without giving up the flexibility to imperatively modify
-variables willy-nilly.
+These immutability checks can be annoying, but the benefit they provide is
+that you can always tell what a function modifies just by looking at its
+header. In combination with dependency-injected hardware, they provide all the
+benefits of [referential transparency](https://en.wikipedia.org/wiki/Referential_transparency)
+that we typically associate with purely functional languages -- along with the
+option of imperatively modifying variables willy-nilly.
 
 ---