nuocuhz Claude Sonnet 4.6 commited on
Commit
9068cf8
·
1 Parent(s): 5576541

Raise system1_score floor to 3.0 when dominant pattern is gap-consequence enumeration

Browse files
Files changed (1) hide show
  1. rater.py +4 -0
rater.py CHANGED
@@ -192,6 +192,10 @@ Guide:
192
  4.0 = strong System 1 signal
193
  5.0 = dominant System 1 reasoning
194
 
 
 
 
 
195
  ### 5.2 system2_score
196
  Definition:
197
  How strongly the passage reflects slow, explicit, analytical, evidence-based, decomposed, and rebuttable reasoning.
 
192
  4.0 = strong System 1 signal
193
  5.0 = dominant System 1 reasoning
194
 
195
+ Scoring constraint:
196
+ If Issue Prioritization <= 1 AND the dominant pattern is gap-identification with consequence labels rather than causal or methodological reasoning chains explaining how the gap undermines the paper's core claims,
197
+ then system1_score should be AT LEAST 3.0, because flat enumeration of gap-consequence pairs is itself a System 1 heuristic pattern — it reflects the enumeration heuristic rather than analytical decomposition.
198
+
199
  ### 5.2 system2_score
200
  Definition:
201
  How strongly the passage reflects slow, explicit, analytical, evidence-based, decomposed, and rebuttable reasoning.