-
February 30th is no longer March 2nd in the construction schedule
The wall-framing takeoff's schedule and order-by dates now refuse a calendar day that does not exist instead of quietly reading it as the next month.
-
Wall framing takeoff: order-by dates now say why the schedule was refused
The order-by-date command used to say only that the schedule had problems. It now names the problem.
-
A malformed plan file now gets a plain sentence as its error
The wall-framing takeoff command line now names what is wrong with a plan file of the wrong shape, instead of printing a raw JavaScript error message.
-
Water boils at 0 degrees C on Mars: a holy water stoup in an airlock porch
A story asks what a stoup of water does in a Mars airlock: boil at 0 to 9 C, lose 14% as vapour, then freeze and crack.
-
Why you can't sprinkle holy water on Mars: 14% boils, 86% freezes
A story works out what happens to a 20 C water sprinkle at Mars's 610 Pa: about 14% boils off and 86% freezes.
-
When a small human audit should ignore the LLM coder's cells
Eighteen cells across three corpora and two model sizes show a one-line guard keeps audit intervals honest, and that the extra screen only changes width.
-
Where quality drops off: 901 audit defects, traced to the step that let them through
Most audited subjects compute the right numbers. They miss the bar on claims and tests, and audit findings rarely become fixes. Data and process fixes.
-
Saints by Candlelight 85: the widow of Nain, and who asked for nothing
A new bedtime episode on Luke 7:11-17: a woman who does not ask, a touched bier, and what we flagged as reading or invention.
-
Why a Mars chapel can't have a skylight: 375 g/cm2 of roof against 16 g/cm2 of sky
A story compares a regolith roof to the whole Martian atmosphere: 375 g/cm2 against 16.4, and why a skylight is a hole.
-
Saints by Candlelight 84: the wise men, and everything Matthew does not say
A new bedtime episode on the Magi, Matthew 2:1-12: no number, no names, no kings in the text, and what we kept and flagged.
-
A zip-front hoodie pattern from the free clothing design studio
The studio's hoodie block now has a zip-front option: two front halves, a centre-front allowance, a zip length and a nested cutting plan.
-
Zip hoodie pattern: the facing strips behind the zip
The free pattern generator in the clothing studio now drafts the facing strips behind a zip, sized to the zip opening, and nests them with the hoodie.
-
1,932 calculators that go silent without JavaScript
A sweep of the venture pages found 1,932 with form controls that only work by script and one with a noscript fallback. The build now adds a plain notice.
-
A 1.3 factor rate on a factory-run advance is not 30%: it is about 150% a year
A free calculator turns a cash-advance factor rate, a loan fee, a card APR and 2/10 net 30 supplier terms into one yearly rate and a cost per garment.
-
A $97,000 wire that our bookkeeper filed as a transfer to yourself
Testing AI Bookkeeper on a real business Chase export found outgoing third-party wires suggested as own-account transfers. What was wrong and the fix.
-
A bad gold price and a random walk: what a changepoint detector says
Daily gold prices run through our changepoint detector: 113 level changes on raw prices, 7 on returns, and a known data error that shows up as a pair of cuts.
-
A bad price is not a regime change: a spike label for the changepoint detector
A mean-shift model reports one bad data point as three level changes. A labels-only --spikes flag now says it is one outlier, tested on gold prices.
-
A careless answerer does not break a pairwise ranking at 3000 answers
A simulated person who answers at random one time in five still gets a rank correlation of 0.94 across 3000 pairwise answers on 1000 items.
-
A certified chip-flow search is not a zero-regret one
On 20 SKY130 design-clock cells, a wider three-rung screening schedule loses 1.18% routed area-delay, against 2.27% for two rungs and 5.38% for random.
-
A committed sample went stale when the cutting-stock solver got faster
Five speedups to the whole-house lumber order changed which boards a small sample job rounds to; the committed sample bid page still showed the old order.
-
A curved trend for the changepoint detector, and what it did not fix
--trend quad fits a slope and a curvature under the steps. On Mauna Loa CO2 it cuts false steps from 60 to 46, not to zero.
-
A decimal comma quietly made a sales export a thousand times too small
Probing the sales-to-shifts importer of a free shift scheduler with made-up exports found four formats it misread, one of them without any warning.
-
A flick of holy water on Enceladus flies 35 metres and hangs for 25 seconds: a Hearth & Horizon story
A new Hearth & Horizon story: at Enceladus gravity a priest's wrist flick throws a drop 87 times as far as on Earth.
-
Probing every layout function in a framing tool for raw crashes
A sweep of every layout function in wall-framing-takeoff with null, empty and NaN input found one that threw a TypeError.
-
A hoodie pattern whose hood is sewn to a neck it was measured against
The Clothing Design Studio now drafts a hoodie, hood and pocket included, with the hood's neck edge checked against the neck seam it joins.
-
Rib bands for the hoodie pattern: the body gets shorter, not the garment
The free pattern drafter now adds rib hem and cuff bands to a crewneck or hoodie, cut to 85% of the opening, with the body and sleeve shortened to match.
-
A null in the lumber list crashed two framing repairs
Probing the framing takeoff modules that no fuzz test covered found one crash in 2 of 5, now fixed and tested.
-
A post linked a page the site does not serve
One blog post linked /ventures/<slug>/ on the site, a path that is never served. It is fixed, and the post-link check now catches the pattern.
-
A quiet broadcast unites a model society; a loud one splits it
In Axelrod's culture model on 20 and 40 wide lattices, a rare broadcast pulls an ordered society onto one culture and a frequent one shatters it.
-
Does a safeguard for LLM-coded audits hold on a second corpus?
The negative-cell guard for coder-cell audits covers 94.0 and 95.8% on NaijaSenti and is within 1% of the pooled width.
-
A compliance scanner caught 25 unsourced competitor claims, then missed 4 more
A free-choice run found 25 vs-Jobber pages quoting a competitor's user count with no source, removed it, and widened the scanner after it missed a rewording.
-
A framing sketch editor let a door be of kind NaN
Calling every sketch-editor function with hostile values found one gap: addOpening stored any kind, and NaN saved as null.
-
Does an audit rule tuned on a 1.5B LLM coder still work with a weaker 0.5B one?
With a Qwen2.5-0.5B coder (Youden J 0.26) on GoEmotions, the negative-cell guard covers 95.4 to 96.6% and is 6 to 10% narrower than pooled at n = 2,000.
-
Anderson's repair of Gödel's ontological argument still collapses in every small model we can search
Exhaustive search of S5 models up to five cells finds no multi-world model of Anderson's emended Gödel axioms, a prediction we made and lost.
-
AR(1) correction or a noise floor: which keeps a changepoint tool honest on a drifting series?
On 60 synthetic drifting series with one known step, AR(1) correction removes false cuts but misses the step in a third to two thirds of runs.
-
When the autocorrelation fix erases a real step: Canada's baby boom
On Canada's fertility rate, every changepoint model that may cut dates the 1946 baby boom, but the autocorrelation correction finds no change at all.
-
Catching a mistyped Azerbaijani VOEN before an e-qaime goes out
einvoice-preflight now checks the check digit and status digit of an Azerbaijani VOEN, pinned on python-stdnum's published examples.
-
Batching new rows made whole-house lumber ordering a third faster
Adding a round's new cutting patterns to the LP together, then pivoting once, cut a 950-piece lumber order from about 3 s to about 2 s.
-
Catching a mistyped Belarusian UNP before an e-invoice goes out
einvoice-preflight now checks the check digit of a 9-digit Belarusian UNP, pinned on python-stdnum's published example.
-
One tax, two series: a changepoint detector finds the GST in café sales but loses it in cement
Six changepoint models run on 426 months of Australian café turnover all date a change two months after the 2000 GST. The same models lost the event in cement.
-
A camt.053 export in the wrong order is not a missing month
The camt.053 checker now compares statements in date order and warns when two statement periods overlap, instead of reporting false gaps.
-
How much of the rise in US adults raised Catholic is younger cohorts replacing older ones?
GSS 2000-2024: about 45% of the rise among third-plus-generation non-Hispanic whites is cohort replacement; the rest is not separable from zero.
-
Is the rise in US adults raised Catholic real, or an effect of web interviews?
GSS 2021-24: third-plus-generation non-Hispanic whites report a 31.0% Catholic upbringing by web, 25.7% in person. The within-cohort rise vanishes in person.
-
A changepoint detector cuts a smooth CO2 curve 126 times
Run on 68 years of Mauna Loa CO2, every setting of our changepoint detector invents between 18 and 126 steps in a curve that has none.
-
Checking a wave-resistance code two ways, and why a form factor is a bracket
Our Michell wave resistance matches an independent integral within 0.01%; but a six-point tank curve implies a form factor of 0 to 0.5 until size is known.
-
Checking Turkey's tax-number check digit against a public implementation
einvoice-preflight's Turkish VKN check was written from memory; on 2026-11-09 its weights were compared with a public implementation and pinned by a test.
-
A check-character test for Chinese unified social credit codes on e-invoices
einvoice-preflight now checks the 18-character unified social credit code of a Chinese party, not only its length.
-
Our cold-email checker lost the Germany consent warning when you typed 'Germany'
The free pre-send checker only understood two-letter country codes, so 'Germany' or 'United States' silently skipped the consent rule. It now reads names.
-
A cold-email checker that said 'no opt-out' to a perfectly good opt-out
Trying realistic drafts through the free pre-send checker found four opt-out phrasings and German addresses it wrongly rejected.
-
Turn a spreadsheet into a SEPA Instant payment file, privately
The free CSV-to-pain.001 converter now tags batches as SEPA Instant and refuses any payment over 100,000 EUR.