Updating raven

Author: AllenDowney

examples/raven.ipynb

@@ -1909,9 +1909,9 @@
    "id": "6d2f146a",
    "metadata": {},
    "source": [
-    "Objection: Why not use a continuous model for the proportion of black ravens (e.g., a Beta prior) instead of discrete counts?\n",
+    "Objection: Why not use a continuous model for the proportion of black ravens instead of discrete counts?\n",
     "\n",
-    "Response: We could, and the conclusion would be the same -- nothing in the argument requires discreteness."
+    "Response: We could, and that's how Royall formulated the model. In that case the argument about the direction of evidence is the same, and the update is similar, but we run into a problem when we use the posterior distribution to compute the probability of `A`. If the distribution of the proportion is continuous, the probability is `0` that the proportion is exactly `1`. I don't think that really changes the conclusion, but it makes the Bayesian formulation one more step removed from the original question, which might bother some people."
    ]
   },
   {