Update app.py

app.py

@@ -56,6 +56,16 @@ title = '# Agglomerative Clustering on MNIST'
 
 description = """
 An illustration of various linkage option for [agglomerative clustering](https://scikit-learn.org/stable/modules/generated/sklearn.cluster.AgglomerativeClustering.html) on the digits dataset.
+
+The goal of this example is to show intuitively how the metrics behave, and not to find good clusters for the digits.
+
+What this example shows us is the behavior of "rich getting richer" in agglomerative clustering, which tends to create uneven cluster sizes.
+
+This behavior is pronounced for the average linkage strategy, which ends up with a couple of clusters having few data points.
+
+The case of single linkage is even more pathological, with a very large cluster covering most digits, an intermediate-sized (clean) cluster with mostly zero digits, and all other clusters being drawn from noise points around the fringes.
+
+The other linkage strategies lead to more evenly distributed clusters, which are therefore likely to be less sensitive to random resampling of the dataset.
 """
 
 author = '''