Blog / Research
Methodology, benchmark, and evidence-first research notes, including dataset and evaluation write-ups. 5 posts.
← All posts
Both approaches address multi-agent safety. Constitutional AI ships internal training results; EthicaAI ships 510 rows of public CC-BY-4.0 evidence with Welch t-test and bootstrap CI. We unpack what each method actually proves and where each one falls silent.
Traditional code-evaluation rubrics score against expected output. WhyLab grounds validation in Docker execution against SWE-bench. The 67-problem prefilter showed selective adaptive C2 does not exceed fixed C2 ??a published null result that traditional rubrics would have obscured.
Renaissance Medallion's reported 66% annualized return (1988-2018) is the gold standard. Quant Bot v11 operates exclusively in PAPER mode until 14-day Sharpe ??1.2 and DSR ??0.5 ??a graduation gate we publish in full (HF dataset 8, 375 sections, 9-Layer Kill Switch). Honest scoping over capital deployment.
A methodology-first reference for comparison engines that publish sources and decision rules.
Why research outputs are labeled by maturity and datasets are cited by name and license.