Archives · every published entry
- Entries
- 3 / 3
- Span
- 2026
- Essays
- 2
- Notes
- 1
2026
-
Sequence-NLP + attention-MIL for materials
A research log from a four-hour GitHub rabbit hole into Sidhom's DeepTCR/ATGC/somatic-mutation LLM line — what's going on architecturally, why it's substrate-independent, and a six-week bake-off plan to see whether the same recipe transfers to materials science.
-
Beyond 37%: When Optimal Stopping Stops Being Useful
The classical secretary problem gives 1/e ≈ 37%, but the rule answers a question almost no one actually asks. Six simulations covering recall, cardinal payoffs, full information, two-sided choice, learned priors, and a market-timing backtest where the naive no-policy baseline beats every stopping rule.
-
Habits, Trust, and the Korean Telemedicine Collapse
Korea's June 2023 telemedicine pilot collapsed take-up to 0.16% of NHIS claims despite eligibility rules that explicitly targeted chronic-disease elderly. Using DML and causal forests on linked claims data, I estimate the behavioral residual on always-eligible, always-supplied patients and trace roughly half its variance to habit strength and patient–provider trust.