diff --git a/publishing/article/discussion.tex b/publishing/article/discussion.tex
index 7023bcb6..d504f598 100644
--- a/publishing/article/discussion.tex
+++ b/publishing/article/discussion.tex
@@ -1,6 +1,6 @@
 \section{Discussion}
 
-In this article we present an automated workflow for full, end-to-end article reexecution.
+In this article we present an automated workflow for full, end-to-end article reexecution and quantification of results divergence at the level of the reproduced article manuscript.
 We generate the full research communication output (including inline statistics, figures, and brain maps) from solely the raw data and automatically executable code.
 This effort substantiates the feasibility of article reexecution as a process, based on a real-life peer-reviewed study example.
 Additionally, our effort provides a sample implementation of technologies required to provide this capability and increase the adoption of reexecutable research outputs.
@@ -13,7 +13,7 @@ \subsection{Reexecutability}
 In contrast to arbitrary reporting standards, the property of reexecutability implicitly guarantees that instructions for all pertinent instructions are documented fully.
 %Further, reexecutability is a far more scalable process, allowing the reliable estimation of the intrinsic variability of a research workflow.
 
-We demonstrate the feasibility of full research output reexecution by integrating cutting-edge technological capabilities, and publish all resources for open access, re-use, adaptation, and inspection.
+We demonstrate the feasibility of full research output reexecution by integrating cutting-edge technological capabilities, and publish all resources for open access, inspection, re-use, and adaptation.
 The article reexecution system which we produced isolates data and original resources, and does not make assumptions about the internal structure of a reexecutable article, and is of course, not domain-specific.
 Our system addresses artcle execution via a Bash entry point, and dependency resolution via an ebuild-style specification present in the target article.
 Given these two attributes, our system can conceivably be adapted to additional target articles.