Repository navigation
Where Humans Actually Yield
A measurement study on operator fatigue. The subject under grading is the instrument — the published curves, and the regulations that are supposed to track them. Never any person, and never any worker's fitness for their job.
Status: RESULTS — 2026-09-08. Six tests comparing what regulators require against what the literature measures, each verdict computed rather than asserted: four AGREE, one CONFLICT, one NOT_COMPARABLE. Every figure below is parsed from decimal text to an exact scaled integer and digest-pinned. No individual is assessed, scored or profiled anywhere in this work.
You are eleven hours into a shift. Are you still safe to do this?
Almost nobody can answer that from the inside, and the reason is measured rather than moral. It is also the reason the honest version of this page had to be written before the useful one.
Human performance under load is one of the best-measured things in the safety literature. These are not estimates from a model; they are counts from very large populations.
| axis | measured yield | source |
|---|---|---|
| cumulative excess wakefulness | attention lapses near-linear beyond 15 h 50 m awake (s.e. 44 min) | Van Dongen 2003, SLEEP 26:117 — n=48, 830 subject-days in-laboratory |
| extended shift ≥ 24 h | near-miss OR 5.90 (5.40–6.30); motor-vehicle crash OR 2.30 (1.60–3.30) | Barger 2005, NEJM 352:125 — 2,737 interns, 17,003 monthly reports |
| circadian phase | driving 02:00–05:00, adjusted OR 5.6 (1.4–22.7) | Connor 2002, BMJ 324:1125 — 571 cases / 588 controls |
| hour within duty | risk "exponentially increasing beyond the 9th hour at work"; 12th hour RR 2.00 | Hanecke 1998 — 1.2 million German accidents |
| prior sleep debt | ≤ 5 h in the prior 24 h, adjusted OR 2.7 (1.4–5.4) | Connor 2002 |
| consecutive nights | vs the 1st night: +6% / +17% / +36% on the 2nd / 3rd / 4th | Folkard & Tucker 2003 — 7 studies |
| rest breaks | risk doubled in the last 30 min before a break vs the first 30 after | Folkard & Tucker |
And one randomised intervention, which is worth more than all the observational work combined:
Landrigan 2004, NEJM 351:1838 — randomised crossover, the same interns, the same intensive care units, 2,203 patient-days. Removing 24-hour shifts moved serious medical errors from 136.0 to 100.1 per 1,000 patient-days, and serious diagnostic errors from 18.6 to 3.3 — a factor of 5.6.
Within the one study that measures three axes with a single outcome, design and adjustment set (Connor 2002), the point estimates order as sleepiness 8.2 > circadian 5.6 > sleep debt 2.7. All three pairwise confidence intervals overlap. Reporting "sleepiness beats circadian phase" from these data would be reading a rank out of noise. All three clear the null; which is largest is NOT KNOWN, and this page does not pretend otherwise.
This is the part we did not expect, and it is the good news.
The duty-hour floor lands exactly on the measured knee. AGREE. FAA Table B minimum and EASA Table 2 minimum are both 540 minutes — 9 hours exactly. Hanecke measured accident risk turning exponential beyond the 9th hour at work across 1.2 million German accidents. Two regulators, on two continents, independently floor at the hour the curve bends.
The floor moves earlier for night starts, and so does the knee. AGREE. Both tables shorten permitted duty as start time moves into the night. Hanecke measured the knee itself moving from the 9th to the 8th hour with later starting times.
The consecutive-night cap lands on the step, not before it. AGREE. FAA 14 CFR 117.27 permits at most three consecutive duty periods infringing the window of circadian low. Folkard & Tucker put the +36% risk step on the 4th consecutive night. The regulator forbids exactly the shift that carries the step.
The defined trough is the measured trough. AGREE. FAA 14 CFR 117.3 defines the window of circadian low as 02:00–05:59. Connor's measured crash-risk trough is 02:00–05:00. The same clock band.
Which matters more, the clock or the workload? The two regulators invert each other. CONFLICT.
- FAA sheds 300 minutes for circadian phase against 150 minutes for task density — circadian dominates 2 : 1.
- EASA sheds 120 minutes for circadian against 240 minutes for density — density dominates 2 : 1.
Circadian phase has a measured effect size. Task density has none — no study in this corpus measures flight segments as an exposure at all. So this is a genuine disagreement between two published priors, and the evidence cannot adjudicate it. Stated as CONFLICT, not resolved.
The axis with the best curve is the one nobody writes down. NOT_COMPARABLE.
Van Dongen's threshold is about time since waking. Every regulator ceiling is about duty length. These are not the same quantity, and converting between them requires knowing how long an operator was awake before the shift began — which no source in this corpus measures, and no operational record collects.
That is not a criticism of anyone. It is a statement about what is currently written down.
The strongest single predictor in the entire corpus is a person's own rated sleepiness — OR 8.2 (3.4–19.7), higher than circadian phase. It is also the one that fails when it is needed most.
Van Dongen restricted 48 people to 4 or 6 hours in bed for 14 days. Subjective sleepiness rose at first and then flattened — self-ratings "did not significantly differentiate the 6 h and 4 h conditions" — while objective attention lapses kept climbing, dose-dependently, for all fourteen days. In the authors' own words, subjects were "largely unaware of these increasing cognitive deficits." At day 14, with performance at its worst, they "reported feeling only slightly sleepy."
If you have been short of sleep for a fortnight, you are the last person who can tell. That is a measured property of human self-assessment, not a failing of character, and it is the single most useful thing on this page for anyone working long rosters.
We call this: the published fatigue curves and the published duty limits agree on four of six tests, and where they disagree it is with each other rather than with the evidence. That is a better state of affairs than we expected to find, and it is worth saying plainly. Regulators did the work.
We refuse to call any of it a fitness-for-duty determination. Nothing here assesses a person. Every figure is a population rate. A yield point is a property of a state — hours awake, hours on task, phase of the clock — and the same person rested is a different state. No operator should be evaluated, ranked or excluded on the strength of anything on this page.
We refuse to convert duty length into time-since-waking. That conversion requires a number nobody collects, and inventing it would be the kind of fabrication this program exists to refuse.
Where a bench should point. The missing measurement is pre-duty wakefulness. It is the axis with the best-characterised curve and the worst operational coverage, and until it is recorded, the strongest available science cannot be applied to the operator in front of you.
No individual data of any kind. No genotype, no diagnosis, no medical information, no personnel record. The corpus is published literature, published regulations, and public accident records. Where a source was gated it was declared and routed around, never approximated — EUR-Lex consolidated 965/2012 returned 0 bytes and is replaced by the UK CAA rendering with the caveat recorded; one BAC-equivalence result is carried only through its independent replication and marked as not file-backed.
Nothing here is medical advice, and nothing here should change anyone's treatment or employment.
git clone https://github.com/gaiaftcl-sudo/uum8dSolarResearch.git
cd uum8dSolarResearch
xcrun swiftc -O -swift-version 5 reproduce/fatigue-yield-vs-regulator.swift -o /tmp/fy
/tmp/fyEight self-test arms run first, in both directions, and the program emits nothing if any fails —
each comparator is shown able to return the opposite of what it returns here, so a verdict of AGREE
is a measurement rather than a tautology. The regulatory instruments are pinned under
corpus/fatigue-yield/ and are re-fetchable from the eCFR: 14 CFR 117, 14 CFR 121,
49 CFR 395.
MARKER FATIGUE_YIELD_VS_REGULATOR__FOUR_AGREE_ONE_CONFLICT_ONE_NOT_COMPARABLE
sha256 d5fab6529c34126003aec8d8ea2f80cb1558ccf9feb3b4896eaef739758206a1
Zero floating point on any decision path. Times are integers by nature; odds ratios are scaled to integers and compared by cross-multiplication. A duty limit that moves with the rounding is not a limit.
The measurements, the corpus digests and the verdict logic are published so that anyone may check them. Re-deriving these figures from the cited public sources requires no permission and no agreement with us. Affine.Earth asserts no claim over the underlying literature or the regulatory instruments, which belong to their authors and to the public.
Rights — source-available, all rights reserved. This wiki and its repository are published for public inspection and to let anyone re-derive the figures. They carry no LICENSE; under default copyright, all rights are reserved. No right is given or intended to use, run, or deploy it for any purpose other than re-deriving the published figures, nor to modify or build on it — any other use requires a written licensing agreement with the authors. · Affine.Earth · zero float · zero shear
Each step is the reason the next one exists. Nothing here is medical advice, and no page calls any medicine safe or unsafe.
1 · Why an exact safety screen at all
- Cures Without the Gatekeeper — the medicine front door: six real written medicines, one screen anyone can re-run
- The library admission law — what may enter, and the 71 arms that prove it refuses. The primary artefact.
2 · The three libraries, which grow rather than close
- The Library of Compound Cures — exact off-target maps for the medicines the registry publishes
- The Library of Proteins — 80,080 generated sequences, novel chemical matter, graded honestly
- The Library of Material Systems — what a system is, what was measured, where the law lives. C-007 absolute: no recipes
3 · The maps — every place a molecule could act, counted
- The off-target atlas — every nucleic-acid medicine the registry publishes a sequence for: WHERE it can pair
- The order of the bases — WHETHER THAT BURDEN IS UNUSUAL: 472 strands ranked against sixteen rearrangements of their own bases
- Where else could this guide cut? — the whole human genome, counted
- Designed, or forced by its own bases? — every clinical CRISPR guide, with its own composition as the control
- What a public genome deposit will tell you — and four ways it will mislead a health tool first
- Study 45 — which of nine billion answers a laboratory can act on — a safety review of AlphaGenome Atlas, measured live on 1,200 real variants at two genes. The headline score separates every one. The detailed tracks do not: splice-site usage hands back 950 of every 1,000 values shared with another variant at HBB and 998 at CFTR, and the shared values pile up in the quiet band where a bench clears a variant
4 · One medicine at a time
- Zilganersen — the first treatment for Alexander disease, screened on the real approved sequence
- A drug an AI designed — rentosertib for pulmonary fibrosis, and exactly what our instruments reach
- CAR-T, halted — the verdict a regulator could re-derive
- N-of-1 antisense — the only safety net at a population of one
- VERVE-102 — the off-target lattice a stranger can re-derive
- PM359 — prime editing, certified before anyone is dosed
- Del-Zota — the one safety question that can be made exact
5 · What keeps a disease alive, and what moves it
- Study 26 — master regulator bonds — 17 tumour types, 7,673 tumours; eleven compound pairs where no single agent among 20,308 cleared any
- Study 20 — Rife frequency — light and frequency, measured rather than dismissed
- Study 37 — five molecules — 37,910 "validated discoveries", 5 distinct molecules; why per-item validation cannot see a corpus-level defect
- Are the generated cures new? — 80,080 peptides against the human proteome
- Study 16 — disease type · Study 17 — chemistry InChIKey · Study 14 — protein lattice
- No language model in this stack — what the answers here are made of: measured 2026-09-12, no cell runs a model process, opens a model port or holds an unmasked model unit, and a gate refuses their return
- Run any study in your browser — all ninety programs open on your own device, forty-nine run there, and the run tells you whether it printed the sealed bytes
- The ontology — grades, terminals, controls, and what each page may say
- Zero Float · Zero Shear — the method in one page
- Ask someone you trust to check this — what to hand a sceptic
- Readers’ guide · Program index — all 42 studies · White paper · Roadmap
- The full-grade replacement — 49 retired instruments, 4 verticals
- The exactness seam — the business case
- Build a study — Falcon walkthrough — how to add one yourself
The same move every time: take a domain where a floating-point model is the accepted instrument, compute the same quantity in exact integers, and seal the cases where the two render opposite verdicts. The subject under grading is always the instrument, never the phenomenon.
- Study 48 — the atom already has an address — silicon dimers 3.840 Å apart, the smallest commanded scale on the board: a length carried in single precision mis-addresses its first atom at step 8,783; an address cannot
- Study 49 — the phase code never needs π — a phase-only modulator takes 256 codes per pixel; the code is a ratio of integers
- Study 50 — CMS raw data from the LHC, read exactly — CMS's 2011 collision bytes streamed from CERN Open Data into the Affine IDE and read in exact integers, every collision a hologram you can turn: 138 of 3,564 bunch slots carry 93,110 of 120,742 collisions, and in 3,854 the event record reads its slot exactly 3 lower than the pixel boards · public release
- Study 55 — IceCube: the light in the ice, hit by hit — IceCube's calibrated hits read byte for byte: 4 published files, 9,749 events, 2,289,821 hits, a census seal per file
- Study 47 — translation shear: the meaning that survives a language — LAW FROZEN · LIVE CLAIM, measured 2026-09-11 and again fleet-wide 2026-09-12: translation as an exact coordinate transform, charts derived in memory at every start from the raw rows of a pinned public weight file and never written down; one lattice digest on 9/9 cells, zero drift, every refusal named. The generative comparison arm is ABSENT — there is no generative translator in the stack
- Study 34 — the observer-invariant verdict — why a safety verdict needs an exact law, not a bigger computer
- Study 35 — the safety brain that forgets — deaf in 8.4 seconds, forgets across machines, disagrees with itself
- Study 36 — the language game of Fermat's Last Theorem — guess and shear, or project
- Study 40 — the number the simulation throws away — their ICO result computed as a fraction; in float the effect returns 0 at every width, and an effect returned as zero cannot be searched for
- Study 41 — fifty years of solving the wrong problem — the ordering was never about time, it was about arithmetic; 177× the work and 2,400× the wrong guesses to return the answer the machine already had
- Study 42 — The Exact Contract — 2.7M flood settlements in Int128 cents; the step exists and the rigidity does not
- Study 29 — continuous-model shear
- The lattice holds · Impact study — continuum dead · Death of continuous shear
- Fourier Phantom — Anima FNO vs 11+12+13 · Stellar dynamo kill shot
- QCD: freedom is dilation · UUM-8D vs IUT — WIN
- Peer-review bundle · Conjecture alignment
- We need fusion — the verdict every machine can check
- Affine Fusion Control — the local exact-integer court · public release
- Fusion researcher's guide
- Study 33 — the fusion control verdict court
- Every season, fifty tonnes — the biosphere-safety case
- The forcing nobody measures · Impact study — the SpaceX trajectory
- Study 31 — the biosphere joint ledger — LIVE on the court, 9/9 cells
- Study 28 — the wet-bulb threshold court — Act 1 sealed
- Study 32 — the taxi-out floor court
- Where humans actually yield — the fatigue curves, and where the rules already agree
- Study 30 — sovereign edge pod · Manufacture contracts
- The detector that flags the whole market — a manipulation geometry in exact integers, and the regulator's own indicator scored against a legitimate quoter
- Study 43 — almost every order is cancelled, and that is normal — nine sessions, three operators, two continents: 935 to 998 of every 1,000 orders that ended, ended without trading. A check that flags almost everything is a denominator, not a detector — and the stock you pick moves it further than the exchange does
- Study 44 — nine billion answers, four billion ways to say them — AlphaGenome Atlas ships 9 billion predictions in single-precision floats, which hold 4.28 billion distinct values: 52 of every 100 variants MUST share a score with another. Agreement and exhaustion look identical on the wire
- Study 38 — the loss-reserve triangle — a reserve is an exact rational; 481 of 482 verdicts identical in both arithmetics; the sixteen-billion figure comes from an unchecked premise
- Study 39 — the actuarial domain — life, pensions, multi-state and aggregation; the margin is 8 significant digits at its tightest
- Run any study in your browser — the ▶ badge beside a program name opens it in the Studio, already built and carrying its inputs, and runs it on your machine with nothing sent back
- Explore the live courts
- MCP user guide — all 51 tools · Deterministic no-float courts for LLMs
- Court Client — generic wasm IDE for every court · Court-client checkpoint
- Coding Court — the verdict IS the artifact
-
Zed — the coding agent, for developers — set Zed 1.20.2 up on
https://affine.earth/v1, no language model anywhere; what a turn does, the wire, the autonomous closure -
Zed — Minecraft comes to life — the two-person interaction, sealed: it asks, cites, clones a sibling with a value you supply, verifies by replay; the court flips
REFUSED_UNKNOWN_BUDGET → WIN - Zed — the agent that teaches the whole domain — architecture, protocols, server management and git, each answered from lines it read and instruments it ran; five closures PROVEN, and the cattle question answered with a counter the fleet did not have
- Math Court on Glama · Math Court user guide · Example app — entire court
- Quantum algorithms inventory · Shor witness certifier
- MCP clients (public)
- Glama connector
- Look in the UI (no visitor data)
A study appears here under the state its evidence has earned, and above under the question it answers. The two are different filings of the same work, on purpose.
✅ LAW FROZEN · DATA SEALED
- Study 06 — explosion vs earthquake · Study 07 — Sgr A* raw visibilities
- Study 11 — Ehrhart volume · Study 12 — parallel repetition · Study 13 — Connes rigidity
- Study 14 — protein lattice · Study 16 — disease type · Study 17 — chemistry InChIKey
- Study 18 — material STD · Study 19 — Go First dice
- Study 26 — master regulator bonds — 17 tumour types, every finding published
🔴 LIVE CLAIM — standing, not sealed
- Study 02 — launch ionospheric holes · Study 02 — regulatory alarm
- Study 09 — global convective bond · Study 20 — Rife frequency · Study 21 — stellar dynamo
- Study 22 — 2-local Hamiltonian · Study 23 — spin glass · Study 24 — N-representability · Study 25 — exact permanent
🌊 CHARTER · OPEN — the findings, published either way
- Study 03 — flare SIDs — archive went dead · predictions and validations
- Study 04 — tsunami vs surge — partial seal · Study 05 — Forbush decreases
- Study 08 — Gaia BH1 — no corpus until DR4 · Study 10 — Fermi / dark matter — does not disprove DM
- Study 15 — Skala DFT shear · Study 27 — exact nuclear scattering
- Overview · First 27 days · Success criteria
- The science, and what history says · Blind spots — five stories magnitude models miss
- Historical corpus · Data archives — every source, exactly how to reach it
- Model shear · Benchmark results · Prediction registry
- Substrate architecture — how a shadow becomes a geometry
- Operations runbook · Satellite & aviation advisory