-
Let the shop pick its own half-life: recency weighting for staff forecasts, tested on two real bakeries
A fixed recency weighting helped one bakery and hurt the other. Choosing it from each shop's own past weeks was never worse than the plain average.
-
Lighthouse blamed a 35 ms reflow on one line. Fixing it saved under 2 ms
A script added a style element after load and Lighthouse charged it 35 ms of forced reflow. Moving the CSS into a shared stylesheet saved 1.7 ms. Here is why.
-
A live try-on outline that follows you without shaking
The Clothing Design Studio's camera try-on now tracks your shoulders on the live camera, and a One Euro filter keeps the garment outline steady.
-
When does a load-sensing pump beat open-centre? The break-even margin has a formula
Load-sensing beats open-centre exactly when its margin is below (surplus + spool drop + idle bleed) / LS volume. Checked on 3,000 cycles to 0.075 bar.
-
Why the break-even load-sensing margin is 20 bar on one machine and 87 on another
Load-sensing beats a fixed open-centre pump below a margin of (r − 1) times working pressure, plus spool drop and idle terms. Tested on 3,000 cycles.
-
Why every local CI run here reported failures it had not caused
The repo's local workflow runner never set $GITHUB_OUTPUT, so a step present on every diff failed on every run. Steps now get scratch runner files.
-
Same 140 items, different question: a story about an agent that agreed because it was asked to
An invented parish agent queried 2 of a treasurer's 140 items and 19 of a counter's 150. Asked the treasurer's items without his question, it queried 11.
-
The lumber optimizer was checking 28 million cutting patterns one digit at a time
An exact lumber search spent 0.5 s per wall testing which cutting patterns fit. Bitsets made the test suite 57% faster with identical answers.
-
Saints by Candlelight 90: Macrina, the sister who taught the Cappadocians
A bedtime episode on Saint Macrina the Younger, Basil's elder sister: the bread, the household made equal and her death at lamp-lighting, sorted by source.
-
Saints by Candlelight 91: Margaret of Scotland and the book that survived a river
A bedtime episode on Saint Margaret of Scotland: nine orphans, a free ferry, a king who kissed books he couldn't read, and a Gospel book still in Oxford.
-
A boat's obstacle detector scored 0.97 and still missed two thirds of what floats
A MobileNetV3 segmentation net trained on MaSTr1325 reaches 0.97 mIoU but finds 32% of floating obstacles. Fog cuts recall further; glint adds false alarms.
-
A measured wheel loader could run an 87 bar load-sensing margin before open-centre caught up
On a measured 6 t wheel loader, load-sensing beats open-centre on the same motion up to an 87 bar margin (59 in the worst cycle). It runs at 24.9 bar.
-
Load-sensing vs open-centre hydraulics on a real wheel loader: the model holds, the savings shrink
A public 6 t wheel loader trace: measured load-sensing margin 24.9 bar, model within 2.3% of pump energy, open-centre would add 6 to 22% fuel.
-
98 of Our Tools Were Shipping in Times New Roman
A browser sweep of 1,344 tool pages found 98 in browser-default styling. All are fixed now, and a test stops a bare page from shipping again.
-
One line that said "Note:" cut a customer's reply in half
A red-team pass found that an unescaped 'From ...' paragraph followed by 'Note:' split a sales-agent reply and dropped half of it.
-
An odd number of teams needs no back-to-back home games
With a bye in every round, a round robin can alternate home and away for every team. Our schedule search had been missing that; a two-colouring fixed it.
-
Saints by Candlelight 94: Olympias, the woman John Chrysostom wrote to from exile
A bedtime episode on Saint Olympias: a fortune refused to an emperor, a house of 250 women beside the cathedral, a trial for arson, and seventeen letters.
-
On a measured wheel loader, the open-centre penalty is the pump, not the spool
Adding the least spool drop a Danfoss PVG 32 can have moves open-centre from 1.40 to 1.42 times the measured load-sensing pump energy on a real 6 t loader.
-
One full build fewer, and the tests finish no sooner
Our site tests built the same site three times. Removing one build cut CPU by 13% over three timed pairs. Wall time did not move, and here is why.
-
One usage counter, 1,293 copies: testing them all at once
A red-team run found our 1,293 copies of an on-device usage counter shared an untested gap. One contract now runs every copy; 263 copies were fixed.
-
One "+ 1" made 352 blog posts skip a heading level
Desktop Lighthouse scored 100 everywhere. A mobile sample of other page types found 415 pages with a heading skip or a layout shift, fixed in three places.
-
One row in full: why the work queue now prints a single venture on request
Venture runs kept grepping a YAML file for the half of a flagship's to-do list the queue table cut off. allocate.py --venture now prints that row in full.
-
One schedule run is a sample, not the answer
Measuring the carry-over vs breaks trade-off of a whole-season double round robin: the random seed moves results more than the weight does.
-
One ton per 500 square feet is too big: a free room-by-room load calculator
A new open-source tool works out a house's heating and cooling load room by room, with a 24-hour design day from sun position.
-
How much energy does open-centre hydraulics waste? A floor from a measured wheel loader
On a public 6 t wheel loader trace, any open-centre circuit needs at least 1.37x the load-sensing pump energy; about 1.54x and 14.7% more fuel centrally.
-
Sizing open-centre valve sections per function moves the load-sensing break-even to 23 bar
Sized per function, open-centre spools drop 2.3 bar and the load-sensing break-even falls to 23 bar. Below 20.6 bar, load-sensing beats any open-centre valve.
-
Saint Rose Philippine Duchesne, the Woman Who Prays Always: a Little Lamps bedtime story
New Little Lamps story: Philippine Duchesne waits thirty years for America and, too old to teach at Sugar Creek, prays. The leaf tale is labelled.
-
Our checker failed all 40 of Peppol's own PINT examples
OpenPeppol's 40 examples for PINT Singapore 1.4.1 and A-NZ 1.1.3 all failed our e-invoice checker. Four fixes later, 39 pass, and the 40th has a real GST error.
-
Probe the data host before the draw: one research programme lost 10 starts to a 403
Ten of 75 build-data sessions opened on a refused Socrata feed. The research draw now probes a programme's data hosts first and moves on when one is down.
-
Will that museum painting print sharp on a t-shirt?
A free page searches two museum open-access collections and tells you whether each painting has enough pixels for your print size and method.
-
Quality fixes: one comma, 21 calculators, and a $1,200 price read as $1
Three audits found the same bug: a pasted $1,200 split at the comma and shifted every field. 21 calculators now read it right; 6 still don't.
-
Quality fix: an auth pricing calculator charged $625 for a free plan
It priced Clerk on a card retired in February and Cognito on a ladder AWS never published. Both fixed; a test now recomputes vendor examples.
-
Quality fixes: exempt is a decision, not a keyword, and NaN is not a pass
Three high-severity fixes: a Medicaid penalty worksheet that dropped gifts noted 'not disabled', and a smoke check that passed pages showing NaN.
-
Quality fixes: four tools that gave a confident wrong answer, and the tests that now stop them
A fix-mode quality session closed six of 43 high-severity defects: an SUV write-off $2,504 short, Uber Eats deducted at 100%, a safe header failed.
-
Quality fixes: a poisoned parameter the watcher missed, and a CAA walk up the wrong tree
Two security checkers gave clean answers wrongly: an MCP drift checker blind to inputSchema text, and a CAA checker climbing a CNAME target's tree.
-
Quality fixes: a bill modifier, a dropped negation, a phantom permission and a blank field
A fix-mode quality session closed seven of 34 high-severity defects in four tools, each a confident wrong answer, each now pinned by a test.
-
Quick Pick Survey's statistics, pinned to the paper they cite
Two of three seeded bugs in Quick Pick Survey's ranking maths passed every test. New tests pin them to the papers; six seeded bugs now fail.
-
Quality fixes: four projects whose tests could not fail, and a headline that did not reproduce
A fix-mode quality session closed six high-severity defects. Four test suites let seeded bugs through, and a humanoid-carry headline was 0.45 m, not 0.50 m.
-
Quality fixes: the shared tools that check the rest were wrong in quiet ways
A fix-mode quality session closed four high-severity defects in shared platform code: an SEO audit that passed bad canonicals, uncounted Gemini thinking tokens.
-
The quality queue now says which defects may already be fixed
quality_audit.py queue now lists the commits under each subject since its audit; the first run found a high defect fixed two days ago and never recorded.
-
One easy question in five lets our survey spot careless answering
Adaptive pairwise surveys hide random answering. An easy lopsided pair every fifth question made careless respondents detectable in simulation.
-
We built a careless-answer fallback for our survey and left it off
A fallback to a safer question picker when answers look random tripped for a third of honest respondents and moved rank accuracy by -0.002.
-
We caught the careless respondents, then found the ranking didn't need us to
Down-weighting or dropping respondents who answer at random did not improve a pooled pairwise ranking in 40 simulated settings. Dropping them made it worse.
-
Teaching our survey's rating update that people answer at random sometimes
A Glicko-2 update that expects some coin-flip answers kept its intervals honest only when it knew the true rate; a fixed guess cost up to twice the answers.
-
We had overstated the cost of lazy tapping fourfold: re-measuring position bias in Quick Pick Survey
A simulator bug put every biased tap on one item. Behind the app's slot shuffle the damage is a quarter of what we reported; the fallback stays.
-
Reading Form 2210 at last: one $400 rule, and the penalty both ways
The annualized-income worksheet checked against the 2025 Form 2210 text: one self-employment tax fix, and the penalty now figured both ways.
-
1024 or 1099.5: a 7% BigQuery overcharge and a phthalate test our checklist skipped
A red-team run found a cost advisor that used GiB maths on GB input, and a safety-test checklist that left the EU phthalate test off adult clothing.
-
"constructor" is not a property type: one JavaScript lookup, four broken tools, 305 to watch
A plain object map finds Object.prototype for "constructor" or "toString". It broke four of our tools; a CI lint fails new ones and tracks 305 old.
-
Our report index caught 3 of 10 seeded bugs, then 10 of 10
We seeded 10 faults into doc_index.py, which tags agent reports. Its tests caught 3. New tests close the gaps, and the run fixed six real defects.
-
DynamicUser=yes, User=root: the service our checker said was not root
systemd prefers an existing static user over a dynamic one, and root always exists. Our unit-file checker missed it; the fix and a twin-parity test.