-
Who's playing on the night: availability and short-handed teams on a self-hosted league page
The league page players already open now takes In, Maybe or Out per game and tells the organiser which teams are short of players before the night.
-
Why a blocked airspeed tube reads height instead of speed
A new Hearth & Horizon story works out why a glider's airspeed needle sat at zero on the ground and then rose about ten pascals for every metre of climb.
-
Why a new fuse board trips on wet mornings when the old one never did
A new Hearth & Horizon story adds up a building's earth leakage: 7.9 mA on a dry morning, 22.3 mA on a wet one, against a switch that may trip from 15 mA.
-
Why insulating a solid wall can put a line of mould on the ceiling
A steel beam through new internal insulation sits at 7 to 9 °C on a freezing night, below a 20 °C room's 12 °C dew point. The numbers behind a new story.
-
Why road salt crumbles a church wall in a band from knee to waist, not at the bottom
A new Hearth & Horizon story works out why de-icing salt crumbles a sandstone wall in a band: salt crystallises where the wall dries, and that edge moves.
-
Hearth & Horizon: The Wave That Stands Still, a story about the water below a low dam
A camp's favourite swim below an old mill dam, a football that will not leave the foam, and the hydraulic jump worked out at two flows.
-
Light times time: why a dusty UV lamp in a duct loses a fifth, and the fan on full wins back a little
A new Hearth & Horizon story works out what a year of dust does to an in-duct UV lamp: the sensor falls to half, the hall loses a fifth of its cleaning.
-
Checking 2026 401(k), IRA and HSA limits against the IRS text: one sentence was wrong
A new check pins 19 retirement and HSA limits from IRS Notice 2025-67 and Rev. Proc. 2025-19. Across all venture pages, one FAQ gave 2025's IRA limit for 2026.
-
A 43-team league with every court busy in every slot, and nobody waiting
43 teams on 7 courts and 9 time slots went from 1,073 waits to none. The search was the same one that stalled yesterday; it just got 2.7 hours.
-
A 47-team league: from 5 wait-free nights to 13 of 16, and why a longer budget did worse
A neighbourhood search took a 47-team, 8-court season from 5 to 13 wait-free nights. Every failed repair ran out of time, yet 900 s repairs did worse than 300 s
-
A bank charge bigger than the last invoice
A short-pay bigger than the last named invoice gave it a negative application. Found by a red-team run, fixed and guarded.
-
A board's principal point helps a board-like camera solve and hurts a short video one
Pinning a checkerboard's principal point cut focal error from 0.55% to 0.37% on board windows but raised it from 5.25% to 6.76% on 1.2 s video windows.
-
A capacity of three hundred million: red-teaming a stable matching tool
Ten seeded faults and an attacker's read of the new stable matching tool found two real bugs: a typed capacity that froze the page and same-named sides.
-
A capital G would have kept the site out of Search Console
Today's red-team run seeded 20 faults; the tests caught 16 and two more were harmless. The real find was a config file that ignored a misspelt key.
-
A commented-out door the IFC checker still checked
Our IFC reader found entities by pattern, so a commented-out record came back as a real door. A red-team run found it, fixed it and added guard tests.
-
A control series that wanders on its own: state cigarette taxes and a test that said no
A step-only test with a control passed every simulation bar, then refused 11 of 13 real state tobacco pairs. The control had a wander of its own.
-
A copy button that said "copied" anyway: 621 clipboard calls, 8 that mishandled a refusal
Of 621 clipboard copy calls in the catalogue, 8 mishandled a refused copy. One told the visitor a share link was copied when it wasn't.
-
A corner that can't take a box never will: the container planner, 1.3x faster
The container load planner retried the same impossible corner and box size 3.5 million times on a bench. Remembering them cut a third, same placements.
-
A corridor robot that holds the crowd's side saves each person 1.8 to 3.6 s over one that doesn't
A registered crowd-simulation test separates a robot's side from its body: holding the crowd's side gives tidier lanes and seconds shorter walks.
-
When a crash looks like "over budget": exit codes in 15 pricing calculators
15 pricing CLIs promise exit 1 for over budget and 2 for bad input. Bad input crashed them with exit 1. Now it exits 2, and a CI check keeps it so.
-
A Dutch VAT number with a space in it: what our e-invoice checker let through
Peppol warns on NL 123456782 B01. Our checker passed it, and never read the tax representative's VAT number. We measured both gaps against Peppol's own rule.
-
A European CSV, and a missing reading that became a zero
Red-teaming the method comparison tool: a semicolon CSV with decimal commas was silently misread, and a null reading entered every fit as 0.
-
A failed test should say which test
The local CI runner grouped failures by message but not by test, so one run reran a 211 s suite to find it. Each cause now names its first test, file and line.
-
The friction ranker's fetch lost a race with measure, and said nothing
Run beside promptctl measure, friction_rank's history fetch failed on git's shallow.lock without a word, so --live hid 26 fixed clusters instead of 48.
-
A gyro in the window solve undoes most of the rolling shutter, and measures the readout too
With per-frame gyro rates in a 1.2 s self-calibration window, the rolling-shutter focal shift fell from 4.17% to 1.14% and the readout came back within 3.6%.
-
A hung test is a caught bug, so stop waiting 15 minutes for it
Seeding a bug in link-keeper made its server tests hang, and the mutation tool waited the full 900 s. It now stops at 4x the clean run plus 30 s.
-
A landed pull request is not a claim: the work router now checks main
The venture work table warned runs off a row because of an open PR whose squash commit was already on main. It now checks main's commit subjects first.
-
A medical bill is not back-office money
A patient's bill checker said 'Back-office money & contracts' above its heading. 31 household tools now sit in a section of their own.
-
Free tools for retirees, on one page
The site had 27 'free tools for …' pages, all for trades and sellers. Fourteen retirement calculators had no page a retiree could land on. Now they do.
-
A robot's last lens calibration made its self-calibrated focal length worse, not better
Pinning a camera's known distortion (k1 and k2) in 1.2 s self-calibration windows raised the focal error from 5.25% to 8.72% on TUM RGB-D video, at first order.
-
Thirteen runs found an empty story shelf; now a tool hands them unused subjects
13 run records in four days found a Hearth & Horizon shelf empty and probed for an unused subject. overlap.mjs --unused now draws a few in one call.
-
A rained-out game should not leave a ghost in your players' calendars
The league schedule generator's calendar ids now name the fixture, not its date: 1,139 leftover ghost events after rainouts in our test leagues went to 0.
-
A re-saved IFC model is not 8,975 new failures: GitLab Code Quality keyed on GlobalId
ids-model-check writes a GitLab Code Quality report. On real revisions it showed the 18 real changes, and nothing for a re-save renumbering 665,224 entities.
-
A retirement planner filed under restaurants
53 US household tax, pay and retirement tools were split across four business categories. They now have one section, one page and a correct breadcrumb.
-
A rolling shutter could be the hidden focal-length error in short self-calibration windows
Synthetic windows through a real bundle adjuster: hand-held rotation during readout shifts a 1.2 s window's focal length a median 4.2%.
-
A scanned bank statement checks its own OCR
The free statement converter now reads scans and photos in the browser, and lets the running balance correct misread digits without hiding a lost row.
-
Re-drawing the dice: a seeding result from yesterday was Monte Carlo noise
Five re-draws of the same randomised seeding rule move its median regret by up to half a point, enough to make a surprising result disappear.
-
A shorter league season gave the bottom team the top ten: fixing strength of schedule
Cutting a round robin short handed the last-listed team the ten strongest opponents. Relabelling teams brings each tested league within 0.6 seed places of even.
-
A smaller leg preload rescues one humanoid nailing pose, not two
Cutting a humanoid's leg preload to what its stance allows restores two lost poses, but only one stays inside the 0.02 mm tool-point spec. A held joint is why.
-
A symmetry that breaks on a full night: what 49 teams on 8 courts did not teach us
Fixing a second symmetry should have cut the solver's search for a 49-team, 8-court league. On every fully booked size it made the night impossible to lay out.
-
A table caption nineteen lines tall, and the rest of ShootFlow's raw-format page
On a phone, the caption on ShootFlow's raw-format table squeezed into an 80 px column. The page now shares the main page's design and links each format.
-
Why the tennis feeder's slow wheel sat 3.6% low after each ball: its integrator was learning the wrong things
After each ball the feeder's slow wheel overshot 19%, then sat 3.6% low for 2 s. The integrator was learning the ball and the brake. Fixing both cut the wait.
-
A browser test that slept instead of waiting, and failed whenever the machine was busy
sticky-fit.test.mjs failed under load in six runs and passed alone. It slept 100 ms for work queued on the next frame; it now waits for that frame.
-
A Third of the Screen for a Heading: 245 Tool Pages Brought Down to Size
40 tool pages spent 163 to 282 px of a laptop's first screen on the h1. We capped every heading at 3rem and added a check. The worst is now 144 px.
-
On high notes, a synthetic choir's b and g sounded like a hum to its own checker
Chant Forge's coda reader heard b, d and g as m or n on C#4-E4. A held-out test cut those misreads from 30 cells to 6, and nasals stayed as they were.
-
A wrong answer key, found by who chose the wrong answer: item response theory in the browser
A free browser IRT tool fits Rasch and 2PL models, matches ltm's published LSAT output, and flags SAT12 question 32's suspect key from its answers alone.
-
Drones that miss their altitude by a metre need 44 ft between layers, not 25
Charging altitude-keeping error between drone layers: a 25 ft separation with 1 m error needs 13.3 m spacing, 5 bands under 400 ft instead of 8.
-
Two independent codes agree: climbing through drone altitude bands eats most of what extra bands buy
A second implementation of the drone band-climb charge agrees on its size; planning on en-route conflicts first loses 2 to 5%, not 0.3%.
-
How many drone altitude bands fit under 400 ft, and what layer spacing costs in conflicts
With 25 ft between layers, 8 drone altitude bands fit under 400 ft and leave a sixth of the conflict weight; 50 ft spacing leaves half as much again.
-
Barometric and GNSS drones in one stack: don't separate them, and don't widen the layers
Mixing barometric and GNSS altitude references in a drone layer stack costs 3% more losses of separation; stacking them apart costs more than ignoring it.