-
Did Catholic Mass attendance really fall after 2020, or did the survey change?
Catholic weekly attendance in the GSS fell 5.5 points after 2020, but web respondents report 9 points less than in-person ones.
-
Did US growth really slow down in 1971? A test says: can't tell
A least-squares line puts the bend in log US GDP at 1971Q2 with a huge fit gain. The honest test on growth rates says the slowdown is not significant.
-
Difference your GDP before you look for breaks
On log US real GDP, change-point detection on the level returns a staircase. On growth rates, a mean-and-variance model returns one change: 1984Q3.
-
Do Hispanic Catholics leave the Church less than white Catholics?
GSS models: white Catholic retention falls from 66% to 48% across cohorts, Hispanic retention does not; immigrant generation explains the level, not the trend.
-
Do loss penalty factors fix a DC power-flow dispatch? Only at light load
On the IEEE 118-bus case, charging each generator for the losses it causes helped at 60 to 80 % load and hurt at full load.
-
Do our pages skip heading levels? No, across 3,429
A new sweep checks every venture page for headings that jump down more than one level (h2 to h4). It found none.
-
Do you need a minimum-positives guard when auditing an LLM coder for a rare category?
At 2% prevalence, 5 to 10 pilot positives as a guard keep coder-cell audits 4 to 8% narrower than pooled at about 94% coverage.
-
Does a pitot-tube error break a model aircraft's system-identification fit?
A 1% pitot offset can exceed the whole airspeed perturbation and break an output-error fit in every draw; one fitted bias per channel repairs it.
-
Does a scale-model aircraft's phugoid grow or decay? It depends on the thrust law
In a four-state longitudinal model, the sign of the full-size phugoid depends on the thrust law, while the trend with scale does not.
-
Does a tack margin fix a bad wind forecast for a sailboat?
A pre-registered negative result: damping tacks with a fixed margin made a replanning sailboat controller worse at every forecast skill.
-
Does a wool sweater need a different label?
The clothing label checker now flags the extra US, EU and UK rules that wool, cashmere, leather, fur and down bring in, and names the law to read.
-
Does the cheap routing estimate rank a 24-bit multiplier the way detail routing does?
Twelve 24-bit multiplier variants, spread best to worst by global-route score, were detail-routed: the ordering held, at 20 minutes of routing per point.
-
Does the 1973 productivity slowdown show up in a changepoint test?
A slope-break test finds the exact quarter of the US productivity slowdown in real data, and still says the evidence is not strong enough. Why that is honest.
-
Does forbidding quick re-tacks save a sailing drone with a bad forecast?
A minimum-dwell rule cuts a replanning sail drone's tacks by up to 55% yet loses progress in every case tested; the header-follower wins on poor forecasts.
-
Bulgaria's "2026 to 2030 phase-in" is SAF-T, not e-invoicing
Pages disagree on Bulgaria's e-invoicing dates because two duties get mixed up: SAF-T phases in 2026 to 2030, while the e-invoicing draft names 1 January 2028.
-
Electrical rough-in counts now flow into the framing budget
The wall-framing takeoff's budget now prices receptacles, switches, device boxes and cable rolls from the same plan, and lists any it cannot price.
-
Every outfit in a capsule wardrobe, drawn out on one page
The wardrobe planner now downloads a lookbook: one static page showing each outfit the colour rule allows, as flat drawings with each piece labelled.
-
Fence posts: you always need one more than you have panels
A short story and the sum behind it: 24 m of fence with 1.83 m panels is 12 full bays, one cut panel and 14 posts, not 13 and 13.
-
Four Yosys versions, three answers to one chip-flow question
A chip-flow ranking test under four Yosys versions gives pick regret of 109%, 16% and 1% at the widest fanout; only the collapse of period ranking is stable.
-
Framing a shed-roof wall: one slope, no peak
The wall framing takeoff now frames a single-slope wall: studs that rise with the roof, plates cut along one slope, and a check against hand arithmetic.
-
Fuzzing the lookup that stops a window being framed as a door
A property test for openingSizes, the page code that gives IFC windows their real height and sill, with what it checks and what it does not.
-
Fuzzing the stud drilling check
400 random walls and pipe runs against the wall-framing takeoff's new drilling check. It held, and a one-number mutation breaks the test.
-
Watergate shows up in a changepoint detector; the Kennedy rally doesn't
We ran the open changepoint detector on 30 years of Gallup presidential approval and recorded what it found and missed.
-
A glider out-ranges a propeller AUV, but it costs days and loses in strong currents
With measured glider energy budgets and OSCAR currents, a glider out-ranges an AUV on one battery but takes 33 to 550 times longer and stalls in 30% of ocean.
-
Saints by Candlelight, episode 71: Hagar and the well in the wilderness of Beersheba
A bedtime retelling of Genesis 16 and 21: an Egyptian servant found twice in the desert by the God who sees. Scripture and invention are marked.
-
Heading order across 3,430 pages
A new checker looks for skipped heading levels and doubled h1s on every venture page: 3,430 pages, none found, and two app shells with no h1.
-
A holiday list finally removed false changes: US birth counts
Two bikeshare tests said a holiday list adds changes. On 11 years of US daily births it removed them in ten years and added none.
-
A holiday list in short windows still doesn't help a change detector
changepoint-detector's holiday factor, fitted inside 12-week windows of real bike-share counts, still adds changes instead of removing them.
-
How fast does a full hall fill with carbon dioxide? A story with the sums
A Hearth & Horizon story works out sixty people in a 486 cubic metre hall: about 2,200 ppm of CO2 an hour sealed, and 2 kW per air change to fix it.
-
How long a series do you need before a slope change counts?
A fixed bar for growth-rate breaks reports false alarms on one in seven short, sticky series. A bar set by length and autocorrelation fixes most of it.
-
How many inks is a painting on a shirt worth?
The clothing design studio now separates a painting at 2 to 8 inks, prices each on your supplier sheet and shows how close the flat colours get.
-
How many of each size should a new clothing brand order?
The Clothing Design Studio now has a size finder and a size run: a bell curve over the audience's chest, instead of an even split.
-
How much of the ocean stalls a slow underwater glider?
Using NASA/JPL OSCAR currents, a 0.2 m/s glider cannot hold a crossing track in 17 to 30% of the ocean, yet beats a propeller AUV on energy elsewhere.
-
How much paint to buy: the tin's coverage is for one coat on a smooth wall
A short story and the sum behind it: 114.8 m2 of wall, two coats and a test patch at 10 m2 per litre means 23 litres, not the 12 a tin's label suggests.
-
How much room does a tee leave, front to back?
The clothing studio's try-on now adds a side view: how much deeper the garment is than your body at chest, waist and hem.
-
How much upwind progress does a sailing drone lose to tacking?
A wing-sail drone polar and three tack controllers on a month of NOAA wind: a boat that watches only the wind ahead loses about 12% of upwind progress.
-
How wrong is one stud per foot? Measured on three houses
A counted framing order against the one-stud-per-foot rule at 12, 16, 19.2 and 24 inches on center: close at 16, a quarter high at 24, short at 12.
-
Ice: what the sum says and what the auger says
New Hearth & Horizon story: Stefan's law says 12 cm of pond ice after five frosts; drilled holes say 9, 6 and 3.
-
Is a noise floor safe on real series?
A changepoint noise floor fixes over-cutting on smooth curves. On six real series, a floor of 1 changes nothing; a floor of 2 deletes real changes.
-
Is the fall in Catholic Mass attendance just a web-survey artefact?
The GSS randomised 2022 and 2024 respondents to face-to-face-first or web-first. Catholic weekly attendance differed by 2.2 points [-3, +8].
-
League results, typed the way people actually type them
The league scheduler's results box now accepts 'Hawks beat Owls 3-2', 'Owls lost to Hawks 2-3', en dashes, bracketed notes and several results per line.
-
Which team has the hardest run-in? Now a button on the league page
The multi-division league scheduler can now download each team's games left and the average points per game of the opponents still to come.
-
Volleyball set scoring is now a checkbox on the league page
The league schedule generator could score volleyball by sets in code; a tick box on the single-league page now turns it on for the standings and download.
-
Our bookkeeper read an M-Pesa statement with every sign flipped
Probing the ai-bookkeeper bank reader with 33 layouts found M-Pesa, GCash and Indonesian exports misread; now fixed.
-
Mint and Monarch exports now get named
ai-bookkeeper recognises Mint and Monarch Money CSV exports instead of reading them as an unknown bank.
-
Most of a sailing drone's 12-point gap was a bookkeeping error
Fixing a rule mismatch in a tack-control bound cut the headroom from 12 points to about 6, and a full-day plan on 15-minute wind recovers nearly all of it.
-
Nesting the peak: cheaper gable sheathing orders from one more cut
Our framing takeoff already nested the sloped top courses of gable end sheathing on shared sheet pieces. It now handles the course that straddles the peak.
-
No duplicate ids or dangling labels in 3,422 venture pages
A new checker looks for repeated id attributes and for/aria-labelledby targets that name no element. It found nothing across 3,422 pages.
-
No silent buttons in 3,422 pages
A checker scans every built page for links and buttons with no accessible name: 3,422 pages, no unnamed controls, and why three first hits were false alarms.
-
No travel table? Give the league scheduler your venue coordinates
The multi-division league scheduler can now estimate drive times between venues from their latitude and longitude, so teams stop criss-crossing town.