diff --git a/README.md b/README.md
index b76d4c0..94982f9 100644
--- a/README.md
+++ b/README.md
@@ -80,3 +80,4 @@ Notes on the [Python benchmark](benchmark_python/):
 
 ## Concluding remarks
   - how is all of this relevant for the users of a computing cluster?
+  - Never trust benchmarks! See for example [Producing Wrong Data Without Doing Anything Obviously Wrong!](https://users.cs.northwestern.edu/~robby/courses/322-2013-spring/mytkowicz-wrong-data.pdf)