+++ title = "Timeline: stax-experiment register" date = 2026-05-15T13:55:57Z draft = false description = "Daily counts of confirmed/refuted/inconclusive verdicts and Type-1/Type-2 catches over the most recent 14 days." +++
Why
The experiment register records every load-bearing claim made on this workstation through a pre-registered hypothesis + falsifier discipline. This page surfaces the rate at which claims are confirmed/refuted and how often the framework catches its own operator in a Type-1 (overclaim) or Type-2 (missed-risk) on their own work.
| date | reg | ver | confirmed | refuted | inconclusive | T1 | T2 | bench | | -— | -—:| -—:|-—:|-—:|-—:|-—:|-—:|-—:| | 2026-05-14 | 45 | 40 | 32 | 4 | 4 | 3 | 7 | 0 | | 2026-05-15 | 62 | 52 | 44 | 4 | 4 | 3 | 9 | 32 |
Reading the table
- reg / ver: registers minted vs verdicts recorded on the day.
- confirmed / refuted / inconclusive: verdict outcomes.
- T1 / T2: verdicts that caught an overclaim (T1) or missed-risk
(T2). Zero across a long stretch when the operator is making real claims is a flatline warning; either the work is unusually clean or the audit lens is asleep.
- bench: structured benchmark events.
Source
- Register:
~/.local/state/stax/experiments.jsonl - Tool:
stax-experiment(https://github.com/SMC17/stax-experiment) - Generator:
stax-timeline-viz 14