diff --git a/notes/008-geolog-to-datalog-subset.md b/notes/008-geolog-to-datalog-subset.md
index 2965580..2b994af 100644
--- a/notes/008-geolog-to-datalog-subset.md
+++ b/notes/008-geolog-to-datalog-subset.md
@@ -440,6 +440,56 @@ Here is a useful one-sentence definition:
 
 ---
 
+## How Useful This Subset Would Be
+
+This subset would be genuinely useful, but only for the right jobs.
+
+It would be useful for:
+
+- understanding a Geolog theory in simpler rule form,
+- debugging the rule part of a theory,
+- exporting some Geolog programs to Datalog tools,
+- comparing Geolog evaluation with Datalog evaluation,
+- and teaching the parts of Geolog that are close to rule-based database logic.
+
+It would be especially useful for theories that are mostly:
+
+- relation-based,
+- rule-based,
+- and not dependent on creating fresh objects during chase.
+
+Good examples include:
+
+- graph reachability,
+- transitive closure,
+- propagation rules,
+- dependency rules,
+- and many database-style logic programs.
+
+But this subset would not cover the full point of Geolog.
+
+It becomes less useful when a theory depends on:
+
+- existential conclusions,
+- equality-heavy reasoning,
+- or the full meaning of native functions.
+
+So the best way to think about it is this:
+
+- it is a very useful **bridge**,
+- a useful **debugging view**,
+- and a useful **restricted compilation target**,
+- but it is not a replacement for full Geolog.
+
+If built well, the ideal behavior would be:
+
+- if a Geolog program fits inside the subset, export it cleanly,
+- if it does not fit, explain clearly why not.
+
+That alone would make the subset valuable.
+
+---
+
 ## Final Summary
 
 If you only want a practical translator, the right plan is:
@@ -451,4 +501,3 @@ If you only want a practical translator, the right plan is:
 - reject existential and equality-heavy features.
 
 That gives a subset that is useful, understandable, and realistic to translate automatically.
-