Explore CTR calibration: map SGO scores to click-through rate predictions
Claude
Unverified signature
109fc16
Adds a research doc explaining how SGO's existing evaluation outputs (scores,
actions, bias calibration, counterfactual gradient) can be calibrated to predict
CTR using Platt scaling with a small set of anchor creatives with known CTRs.
Includes a reference script (ctr_calibrate.py) that fits a calibration function
from anchor runs and converts counterfactual score deltas into CTR deltas.
https://claude.ai/code/session_0141cbZmdz7ziFkNsQbq7z5Y
Add CoBRA-inspired bias auditing and calibration for expert panel fidelity
Eric Xu
Unverified signature
76bda55
Integrates research from CoBRA (arXiv:2509.13588, CHI'26 Best Paper) to
measure and calibrate cognitive biases in SGO's LLM evaluator pipeline,
closing the gap between simulated panels and real expert panels.
- Add bias_audit.py: runs framing, authority, and order-effect probes
through the same LLM+persona pipeline, comparing results to human
baselines (Tversky & Kahneman, Milgram)
- Add --bias-calibration flag to evaluate.py: appends bias-aware
instructions to reduce framing/authority/order artifacts
- Add research analysis doc connecting CoBRA methodology to SGO
