diff --git a/anthropic_shadow.myco b/anthropic_shadow.myco
index 733f944..0ff8ab7 100644
--- a/anthropic_shadow.myco
+++ b/anthropic_shadow.myco
@@ -2,6 +2,6 @@ As a corollary to [[quantum immortality]], events which would kill all humans, s
 
 * Python has bad dependency management because all ML code is written in it. If it were good, we would have AGI.
 * RL doesn't work stably or reliably because it would be too powerful - imitation learning is less likely to do "weird things".
-* LLMs are what we got because they are slow to develop ([[scaling laws]]) and can do some useful tasks but are bad at [[agentic]] action.
+* LLMs are what we got because they are slow to develop ([[scaling laws]]) and can do some useful tasks but are bad at [[agentic]] action. Nobody training at large scale will adopt various things which plausibly substantially improve performance because they would become too capable.
 * Google can never ship any AI products competently because their infrastructure and researchers are too powerful.
 * Kabbalah (approximately, extremely deep research into Jewish lore) exists to divert many of the world's highest-[[power level]] autists away from engineering fields.
\ No newline at end of file