This is the historical log. For current claims, start with the research-area index.
2026-09-15
- FRONTIER-METRICS-REVIEW-0001 · Negative Result — Two ways a synthetic world flatters its author, and both headline results withdrawn
2026-09-14
- INTERVENTION-EP-0001 · Method — When the predictor also acts, the observational prediction contract breaks in five places
- HUMAN-MODEL-EP-0005 · Dataset — A dataset portfolio that keeps development, external validation, and internal-load testing separate
- GPU-SCHED-EP-0013 · Finding — The real-cluster claim rests on one job out of 19,100
- GPU-SCHED-EP-0012 · Finding — Replaced it with a statistic whose boundary stays put, and restored the numbers
- GPU-SCHED-EP-0011 · Finding — The detector was not measuring divergence. It divided by the window, so it never moved
- GPU-SCHED-EP-0010 · Finding — Retracting “this is decided by two numbers”
- GPU-SCHED-EP-0009 · Finding — The axis of the flagship phase diagram was confounded by a factor of 18
2026-09-13
- IAA-EP-0008 · Protocol — A 2x2 design separating preparation time from backward CoM in the badminton smash
- HUMAN-MODEL-EP-0004 · Method — A contract that stops human data reaching a model on ambiguous terms
- SIM-WORLD-EP-0001 · Finding — How much of a hidden queueing mechanism can observation alone recover?
- GPU-SCHED-EP-0008 · Finding — Changing only the granularity of the background made the headline result disappear
- GPU-SCHED-EP-0007 · Finding — Measured at the real cluster’s position, and withdrew an explanation carried for four studies
- GPU-SCHED-EP-0006 · Finding — The driver was never pool size. It is how many jobs compete at once
- GPU-SCHED-EP-0005 · Finding — The safe ratio falls as the pool grows
- GPU-SCHED-EP-0004 · Finding — Real GPU clusters never received a job that filled the pool
- GPU-SCHED-EP-0003 · Finding — Whole-cluster jobs, not mean gang size, decide when size-based scheduling breaks
- GPU-SCHED-EP-0002 · Finding — A good average, a job class that never finishes, and three broken stability detectors
- GPU-SCHED-EP-0001 · Finding — Estimation error and restart cost reverse which GPU scheduler wins