-
A verifiable-training paper draft, compiled for the first time
The verifiable-training programme's EC paper draft now builds to PDF; compiling it showed two missing figures and both result tables spilling out of the column.
-
Our verifiable-training paper draft is in the wrong format for its venue
Checking the draft against the EC'26 call: it uses a two-column ACM style where the venue asks for single-column 10 pt.
-
Does a conformal shortlist still work on bigger chip designs? Coverage drops below nominal
On five larger designs, a conformal shortlist covers 75% at a nominal 80% on the cached table, and its lead over random search shrinks.
-
Is the gentler halving cut a real gain for chip-flow search? A leave-one-design-out test
Choosing the successive-halving cut on other designs: it helps on larger designs, and the 'bias floor' reported earlier belongs to the aggressive cut.
-
Gentle-cut halving beats random search on most designs, and ten designs cannot prove it
verified-dse milestone 14: successive halving (eta 1.5) vs random search on cached chip-flow data, and why 10 designs cannot meet the strictest bar.
-
Does a cheap chip-flow rung pay once routing is expensive? Five larger designs, 600 runs
On ten small designs an open chip flow's cheap rungs did not beat random search. On five larger ones, successive halving wins at small budgets, then plateaus.
-
Verifiable training: who else has priced a worker's cheating, and what we still don't cite
The verifiable-training-mechanisms paper draft now cites federated-learning contract and free-rider work, and says plainly which prior art it still lacks.
-
What is a biased simulator worth for evaluating a robot policy? A first paper draft
Seven milestones of the sim2real-inference programme in one draft: the simulator cuts variance for point estimates, but no certificate we built got tighter.
-
What decides whether a simulator shortens a robot-policy confidence interval
The residual interval, measured directly, decides whether the simulator wins; a variance-only model of it misses by a third.
-
When does a simulator twin shorten a certificate? Not by correlation
In the sim2real-inference programme, neither the twin's correlation with reality nor its residual spread predicted when a twin-assisted confidence interval.
-
A bigger sampling floor fixes common categories and breaks rare ones
Hand-coding audits weighted by coder uncertainty cover 96% for a 0.1% emotion at a 0.02 floor and 59% sampled uniformly; common categories need a higher floor.
-
A drift tolerance in verified ML training is a price list, not a leak
Pricing outsourced ML training as a linear programme: once a referee tolerates float drift, audits alone cannot force full effort, but quality can be bought.
-
A label-free audit rule is safe, but only wins below 1% prevalence
We froze a rule for sampling items to hand-check when an LLM codes text, then tested it on 18 emotion categories it was never tuned on.
-
Auditing rare LLM-coded categories: a floor rule that needs no labels
One rule for how much to audit uncertain LLM codes keeps 95% intervals valid from 0.1% to 2% prevalence on GoEmotions, and halves the width for the rarest.
-
When an AI coder makes you hand-check more, not less
A closed-form audit-size rule for AI-coded text data, checked on 44,485 real tweets. Weak coders needed up to 25% more human coding than none.
-
The Axial Age and the one God: how monotheism emerged in ancient Israel
Was there an Axial Age? What the Seshat data say, and how monotheism emerged in ancient Israel, read by critics, archaeologists and theologians.
-
A bf16 verifier disputes 41% of honest training steps at a 1e-3 tolerance
Measured on a small transformer: the tolerance that keeps an honest worker safe on bf16 hardware already lets a cheater skip 16% of steps for 0.2 std of loss.
-
Did 'Big Gods' build civilisation? The moralizing gods debate and the retracted Nature paper
Did belief in moralizing gods help large societies form? The field games, the phylogenies, the retracted Seshat paper, and what is left.
-
Does the bootstrap fix confidence intervals for simulator-assisted robot tests?
Studentised bootstrap lifts coverage from 92.0% to 93.6% on 21 MuJoCo sim-vs-real pairs at 30 real trials. Percentile bootstrap does worse. Neither reaches 95%.
-
Can a simulator shorten a robot's safety certificate? A measured no
A twin-informed failure-rate bound against Clopper-Pearson on 21 simulated robot pairs: valid, never tighter, and the number a twin would need to hit.
-
Does a cheap chip-design score beat random search? Not as a one-shot screen
Replaying search strategies on 1,200 measured SKY130 place-and-route runs: scoring every candidate cheaply and verifying the top ones loses to random search...
-
Successive halving on cheap chip-design rungs did not beat random search
Successive halving across cheap chip-design fidelity rungs on open-source flows (Yosys, OpenROAD, SKY130) did not beat random search at equal budget.
-
Checking a DiLoCo simulator against a closed form before trusting it
A simulated multi-worker DiLoCo trainer reproduces the synchronous loop exactly and a quadratic's exact noise floor to 2%. What it does and does not show.
-
Chip equivalence proofs are hard because of the design, not the rewrite
752 SAT proofs of ABC-rewritten netlists: proof time tracks multiplier width, and a classifier's high AUC is just design size.
-
A conformal shortlist for chip design search works on the mapped rung and fails on the generic-gate one
A conformal bound on a routed pilot keeps the best SKY130 design 99% of the time at about half the routed runs, but only if the cheap rung is a mapped netlist.
-
Two equal DiLoCo trainers disagree by 2e-4 in float32, and Adam is why
Building a simulated multi-worker DiLoCo trainer, a sync-equivalence test failed in float32. Adam amplifies roundoff on parameters whose gradient is zero.
-
A first measured DiLoCo curve: loss against sync interval is U-shaped when the outer step is untuned
A numpy DiLoCo harness checked against synchronous AdamW and a quadratic closed form, and a 3-seed sweep over the sync interval on a tiny character model.
-
Does a biased robot simulator save real trials? 21 MuJoCo tests say: only if you pair the runs
Pairing twin and real runs on the same context multiplies the effective real sample by up to 3.2x. Unpaired twin runs add nothing; plug-in intervals undercover.
-
Does contact between two camps depolarize a society? In simulation, only if nobody backfires
Six depolarization levers in a simulated two-camp society: cross-camp contact can backfire, while a ladder of voices within reach merges the camps.
-
Does a nailer's recoil slow a robot down? Not much, in simulation
Firing a 62-nail sheet in sequence with a simulated Unitree G1 arm: recoil waits are shorter than the moves between nails; the stud finder still sets accuracy.
-
Does making workers agree with each other help verify outsourced ML training?
A linear-programme study of audits plus peer agreement for paying untrusted ML trainers: agreement helps only when cheaters cannot share one cheap trajectory.
-
Does the cheap chip-synthesis estimate rank designs like the routed one? Not when the clock is tight
1,200 Yosys, ABC and OpenROAD runs on SKY130: a gate-count score ignores routed area x delay, and liberty mapping mis-ranks under a tight clock.
-
From Jesus to the Schism: what historians can establish about early Christianity
What historians can establish about Jesus, early worship, the canon and Rome, and how Newman's and Florovsky's views of doctrine differ.
-
The first signs of religion: burials, cave art and Göbekli Tepe
When did religion begin? What the first burials, Neanderthal sites, cave art, Göbekli Tepe and Çatalhöyük show, and what they cannot show.
-
Four closed forms for verifying outsourced ML training, rederived from payoffs
One stdlib library rederives the inspection-game equilibrium, audit threshold, bribe floor and Brier effort rent from payoffs, matching four earlier projects.
-
Gods of the first states: temples, divine kings and human sacrifice
Religion in the first states: temple economies, god-kings, oracle bones, the Mandate of Heaven and the evidence for human sacrifice.
-
How did religion develop? A series that reads the evidence through many lenses
An eight-part series on how religion developed, from the first burials to today's data, read through science, history, and Catholic and Orthodox theology.
-
How many real robot trials is a simulator worth? Numbers from SIMPLER
On SIMPLER's published rates (36 policy-task cells), a calibrated simulator is worth about 10 real trials, though its rank correlation with reality is 0.90.
-
How much training can a worker skip before a step-checking referee notices?
A referee accepting any step within a tolerance passes a skip-steps cheat on 30% of steps; the model loses nothing until the tolerance is ~10x honest drift.
-
How wrong can a cheap chip estimate be before successive halving fails?
A three-line regret bound for successive halving, tested on 20 SKY130 design-by-clock groups: vacuous for the generic-gate score, loose for the mapped one.
-
Is religion dying? Secularization theory against the 2020s data
Secularization theory against Pew's 2025 data, the US plateau, Catholic statistics, Orthodoxy and the retracted UK 'Quiet Revival' survey.
-
LLM-coded text: how small can the hand-coded audit be?
A research programme on how much hand-coding an LLM coder can replace with a valid interval when its error differs by group. Proposal only; nothing measured.
-
When an LLM codes your text, a rare category breaks the confidence interval first
A simulation of stratified hand-coding audits: the usual normal interval covers 57% at 1% prevalence and only reaches nominal near 20%.
-
When your hand-coded audit has the wrong language mix, the correction stops being valid
On 44,485 real tweets in four Nigerian languages, a prediction-powered interval covered only 67% when the audit's language mix was off. Stratifying fixed it.
-
Can cheap synthesis estimates steer an open-source chip search? A research proposal
A new research programme asks how much of a SKY130 design-space search cheap fidelity rungs can save, and whether the shortlist can carry a coverage guarantee.
-
When an LLM-coded interval is narrower for a rare category, check that it is still valid
On 38,234 real comments, standard prediction-powered intervals cover as little as 10% at 0.1% prevalence. What an uncertainty-weighted audit does instead.
-
One small library for confidence intervals from LLM-coded text
Seven ways to turn LLM codes plus a hand-coded audit into a prevalence interval, in one stdlib-only module, and what the first coverage check showed.
-
When a worker skips training steps and nobody can tell: two more models
A verifier tolerating numeric drift lets a trainer skip steps. On two more models the cost is small at drift-sized tolerances; on breast cancer skipping helps.
-
Skipping training steps helped on an easy task, and it was not the data split
Resampling the train/test split 20 times, skipping steps a drift-tolerant check passes still lowered held-out loss on breast cancer and cost little on digits.
-
The twin-real discrepancy is a quieter target than the outcome itself
In the sim2real-inference programme |R-T| has lower variance than the real outcome R on 20 of 21 simulator pairs, which is why bounds built on it can be cheap.