diff --git a/_data/index.json b/_data/index.json
index dbebfbe48c84ec4d1ec4e8ac45e101da422e0a5c..b343915eb3be101154a78e6f54e93de80fe08fc3 100644
--- a/_data/index.json
+++ b/_data/index.json
@@ -77,6 +77,8 @@
 				"name": "2",
 				"topic": "AI Experiments and candidate games",
 				"links": {
+					"metrics": "code/AppMetrics.java",
+					"intro":"slides_pdf/ce810-metrics.pres.pdf"
 				}
 			},
 			{
diff --git a/slides/content/metrics.tex b/slides/content/metrics.tex
index 3515d7cc891d3af1a5bce245185bdbeed8f3c6fe..0350ca7514f7859134e422a77c6801d590b5ba95 100644
--- a/slides/content/metrics.tex
+++ b/slides/content/metrics.tex
@@ -54,7 +54,7 @@
 	\begin{itemize}[<+->]
 		\item We did these yesterday.
 		\item We should load the AIFactory
-		\item Then use the factory in the evaluate
+		\item Then use the factory to build agents in the evaluate
 		\item ai.buildAI("ProRuleRushRangedBlue", settings),
 	\end{itemize}
 	\end{frame}
@@ -83,7 +83,7 @@ public Double evaluate(GameSettings settings) {
 	\end{minted}
 \end{frame}
 
-\begin{frame}[fragile]{Basic}
+\begin{frame}[fragile]{Less Basic}
 	\begin{minted}[breaklines,tabsize=2,fontsize=\footnotesize]{Java}
 public Double evaluate(GameSettings settings) {
 	GameState start = map.buildState(settings);
@@ -101,10 +101,6 @@ public Double evaluate(GameSettings settings) {
 			winCounts[winner]++;
 		}
 	}
-
-	double score = winCounts[1] - winCounts[0];
-	fitnessScores.put(settings, score);
-	return score;
 }
 	\end{minted}
 \end{frame}
@@ -126,5 +122,51 @@ public Double evaluate(GameSettings settings) {
 	\end{figure}
 \end{frame}
 
+\begin{frame}{Results}
+	\begin{table}
+		\begin{tabular}{l|l|l}
+		defRanged&atkRanged&fitness\\\hline
+		1 & 5 & 30\\
+		10 & 6 & 20\\
+		4 & 2 & 20\\
+		10 & 2 & 20
+		\end{tabular}
+	
+		\pause So not a very good fitness function then...
+	\end{table}
+\end{frame}
+
+\section{User Testing}
+
+\begin{frame}{A|B Testing}
+	\begin{itemize}[<+->]
+		\item Put two versions of the game in front of someone
+		\item Ask them what one they enjoyed most
+		\item Ideally should use a large sample size (or stats)
+		\item But this isn't meant to be rigorous
+	\end{itemize}
+\end{frame}
+
+\begin{frame}{Qualitative Data}
+	\begin{itemize}[<+->]
+		\item In inform out conclusions we could ask other things
+		\begin{itemize}
+			\item Basic demographic data,
+			\item Experience with games,
+			\item Open ended questions,
+			\item Game Metrics...
+		\end{itemize}
+	\end{itemize}
+\end{frame}
+
+\section{Advice}
+
+\begin{frame}{AI experiments}
+	\begin{itemize}[<+->]
+		\item Fast iterations - don't spend too long waiting for stuff
+		\item Drop turn limits, repeats, generation counts...
+		\item See what kind of data you're getting and adapt your fitness.
+	\end{itemize}
+\end{frame}
 
 \end{document}