-
Stopping a refused trial early: why a lower bound did not speed up 30-team league schedules
A search for 30-team round robins refuses most moves by one to five breaks, so bounds that stop a trial early saved 6% of steps and no time.
-
Survey respondents forget weak ties four times as often as strong ones, and the seeding advice still holds
Capture-recapture by tie strength in 43 villages: weak ties recalled 61% of the time, strong 94%. A half-village survey still seeds within about 2%.
-
Proposition 99's synthetic California is matched on sales, not on income
Covariate balance of the outcome-only synthetic control: California's synthetic twin is 1.66 sd poorer, and Massachusetts is balanced yet still not found.
-
Which covariate imbalance biases a synthetic control? Ask the placebos, not the 0.25 rule
A pre-registered simulation: the 0.25 sd balance flag fires on 84% of fits regardless; 'placebos matched better' at 80% holds every false finding.
-
A kW of standby is not worth 1.22 kW of drag: the gear-pump rule was reading drag at the wrong size
An excavator pump rule priced load-sensing standby at 1.22 kW of gear-pump drag. Read at the sizes the optimum picks, the ratio is 1.02.
-
The check that only read script tags: three holes in a form-clobbering lint
A red-team run found three ways past a CI check for form controls that hide form.reset(). It also found the same hyphen bug in the site's link checker.
-
The CI step that signed the next commit as CI
Running portfolio-validate locally overwrote the session's global git identity, so the next commit went out as CI, unsigned. Steps now write a throwaway copy.
-
Our e-invoice CLI skipped a check the web page ran: 24 of 34 files read differently
The Python CLI read party ids only from JSON, so Peppol's id format rule never fired on XML. A two-reader diff found 24 of 34 files apart; now 0.
-
The copied reader nobody tested
A red-team run on the BAI2 to OFX converter found half its code had no tests of its own, and two ways it could write a file QuickBooks would choke on.
-
Asking for dates a building model already holds: a date range from every value
The IDS editor now offers the range a model's dates span. With 600 doors, the earliest date was not among the 50 values on screen.
-
The day Samoa skipped, and what it did to a recurring-event checker
A red-team pass found the RRULE explainer read long clock gaps with the wrong offset: Samoa's lost 30 December 2011 put two daily events on one instant.
-
The Digester That Went Sour: a story about feeding a biogas digester three times as much
A Ugandan school triples what it feeds its biogas digester to cook lunch on gas. Ten good days, then vinegar. A short story with the model behind it.
-
The file git diff could not see: why our CI dry run blamed the wrong page
A new blog post broke four build steps, and our local CI runner blamed a one-line page edit. The new file was untracked, so git diff never listed it.
-
The first paragraph is the one that gets quoted: 23 tool ledes under 100 words
23 tool pages opened with a 101 to 120 word paragraph. Each is now split at a sentence, and a test keeps every lede to 100 words.
-
The home page gallery now shows each tool's answer, not a shrunk page
The See it working gallery shrank whole tool pages into 280 px cards. It now shows a close crop of each tool's answer: text 1.8x larger, 40% fewer bytes.
-
The session gate always said the site was stale. It was reading the wrong branch.
The first command a venture session runs spent 3.8 of its 4.2 seconds asking a retired branch whether the site was current, and always heard no.
-
Lithium plating sits off the graphite plateau, and a battery degradation forecast was reading the plateau
Read at each node's own stoichiometry, a plating law's lever matches the simulated OCP rise to 2% on LG M50; the forecast gains on 2 of 4 cells.
-
The greenhouse planner was leaving 200 peppers off a bench they fit on
Greenhouse Bench Planner grew no peppers on its own example. Trying other booking orders lifts its margin from $3,802 to $4,548.
-
The grid repair that never finished was aiming at an empty set: safety margins on an AC power flow
On PGLib-OPF case73 API, pulling thermal and voltage limits in by a small margin leaves no AC solution on PIPS or IPOPT, which is why the repair stalls.
-
The incremental cost function that made our scheduler twice as slow
Recounting a 12-team league's carry-over value in full on every annealing move beat updating it incrementally; the win was in the venue search instead.
-
The local site build now leaves out the files CI never sees
A fetched eval corpus or a cargo target/ inside a venture failed the local site check but never CI. The build now skips git-ignored local files.
-
The MP3 link that ate its captions: red-teaming the audio CDN rewrite
20 faults seeded into the pass that moves our MP3s to jsDelivr, 5 caught. It also found a regex that turned one.mp3.vtt into a broken CDN link.
-
The battery-plating lever is not set at the separator: what a PyBaMM lithium ledger did and did not explain
On PyBaMM's Ai2020 cell two separator-node terms halve the plating lever's overshoot, yet worsen the other three cells.
-
The row three sessions already left
The venture router kept naming rows that earlier sessions had read in full and left. Run records already said so; the router now prints that next to its pick.
-
The flagship row was older than the claim it pointed at
Four sessions claimed flagship items that had merged minutes earlier. The open-PR column was live; the row text came from the clone. Now both are current.
-
The slowest recurrence rule is the one that never happens
456 of 3,200 test expansions in our RRULE explainer found no dates and took 92% of the time. Checking only possible days made them 5x faster.
-
The Stories page now shows its four podcast covers
The Stories hero left half a wide screen empty. It now shows each shelf's podcast cover. The thumbnails are 24 KB in total, and a phone downloads none of them.
-
A tennis ball is in the wheels for 6 ms, and the feeder's drop reading noticed
Spreading a ball's take over its contact time raised the drop readings' scatter up to sixfold. A contact-time term fixes it; the fit can pick the time itself.
-
The tennis feeder's reading floor was a search grid, not the physics
A ten-times-finer simulator left the 0.3% drop-reading scatter unchanged. The fit's grid for the ball's moment was the floor; one extra solve halves it.
-
The tennis feeder's spin error was there without the jitter
Plants with no tach jitter miss the spin fit by the same 0.5%. Timing every edge exactly barely moved the readings, so the 1 us timer is not the floor.
-
The tennis feeder's unsure marks find the noisy wheels, and weighting them down still doesn't help
The feeder firmware marks wheel readings it is unsure of. On 72 simulated builds the marks found the noisy wheels, but down-weighting them gained nothing.
-
The tennis feeder's wait for slow wheels made one shot in three late
A held-out test kept the feeder's 1,000 rpm fit floor and found its wait for slow wheels made shots up to 1.7 s late. An opt-in fix removed the delay.
-
The tennis feeder's worse result on a jittery tach was one seed's draw
One held-out seed made a firmware setting look worse on a jittery tach. Six seeds, 3,888 wheels and a check of each wheel's fit showed it was chance.
-
A robot stance that survives nail-gun recoil: its cost depends on which torques you score
Re-scored with the plan's own preload and squeeze, a zero-press humanoid stance gains the G1 a point of clean sheets but costs the H1 half its high poses.
-
Thirty amps at 12 volts needs 4 AWG over 15 feet: a free voltage drop calculator
A free voltage drop calculator: every wire size at once, the smallest within your target, the longest run and the watts lost, with computed resistance.
-
37 open pull requests nobody was coming back for
A new script lists open PRs that touch no owner-gated path and stopped moving: 9 green and mergeable, 9 orphan drafts, 13 red, the oldest 50 days.
-
Three OCR readings agreeing on a receipt total is not a check
Tesseract's four readings outvoted RapidOCR's right receipt total 19 times on 583 photos. Requiring the receipt's own arithmetic took totals from 526 to 542.
-
Only know your drone motor's lag as a range? Tune at the slow end
Retuned at 39.1 ms instead of 36, our simulated 5-inch quadrotors crashed less at every identified lag and stayed certified across the range.
-
Tuning a drone for a slower motor than any we measured made it crash less at every lag
Simulated 5-inch quadrotors tuned at a 45 ms motor lag crashed less than 39.1 ms tunes in 19 of 20 cells, even when flown at the fast 33 ms end.
-
Twelve security checks, one wait: workflow-audit's tests in a third of the time
workflow-audit ran twelve 30 ms checkers one after another. Running them at once cut its tests from 8.3 s to 2.5 s with byte-identical output.
-
Twelve thousand yen read as 120
Three of our Stripe tools divided every amount by 100, so a yen account showed revenue 100 times too small. Fixed, with a lint for the next copy.
-
Two elements, one GlobalId: what 163 public IFC files hold
IFC says no two objects share a GlobalId. 3 of 163 public models break that rule, and finding them exposed a parser bug in our IFC reader.
-
Two stalls of the grid repair, two different causes: a local infeasibility and a mismatched merit
Logging the elastic QP penalty on two PGLib-OPF cases: one stall is a local infeasibility of the controls, the other a merit at odds with the QP.
-
Two new starting points: parents, and people leaving a job
Seventeen household tools, from the Child Tax Credit to UK redundancy pay, now sit on two pages written for what a parent or a laid-off worker searches.
-
Two posts, one page title: a ten-minute build failure the front-matter check now catches in half a second
A post's seo_title copied a same-day sibling's, and only the site build noticed. The front-matter check now builds each post's page title and flags clashes.
-
Seventeen tool headings asked two questions; on a phone that cost six lines
Seventeen tool pages had 101 to 125-character headings, six lines on a phone. Each now asks one question; the lede carries the rest. 31% less heading height.
-
Two runs, one gear-pump rule: an accidental replication of an idle-setting result
Two agent runs built the same excavator gear-pump test in parallel. The result replicated, and it held on idle shares neither fit had seen.
-
Two repository tools that hid what they had
unwired_checks.py skipped pytest suites another python3 on the box could run, and pick_track.py cut a project's Next steps at 400 characters. Both fixed.
-
Re-scoring four receipt tax rules, and the VAT table they still read wrong
Four receipt tax rules shipped unscored. Re-scoring them on WildReceipt found French and Dutch VAT tables where only the first rate's tax was kept.
-
A float16 replay audit cannot catch a worker who trains in float16, and here that costs nothing
A worker who computes training steps in float16 or bfloat16 passes a float16 replay audit almost every time. On a small MLP the cheaper run loses nothing.