-
The incremental update that lost, and the boring file read that won
An exact incremental carry-over count made a league scheduler 1.8x slower. Moving 8,784 page reads from fs.promises to readFileSync saved 2.2 s a build.
-
The Details column is not the merchant: a bank-import fix in the expense tracker
A red-team pass on the free expense tracker: 16 seeded faults, 11 caught, all 16 caught after five new tests, and a bank-CSV column bug fixed.
-
The discount for filing on time: 15 more states, read from their own forms
Most states let sellers keep a slice of sales tax for filing on time. We read 19 states' own rules, and found a CLI bug that ignored --state.
-
The finding you only see when a field is empty
Eight tool pages skipped from h2 to h4 only after an empty field or an edge value. Clicking Example never got there; changing each input did.
-
The fog reads eighty-nine: a blight story about calibrating a sensor with salt
New hopeful science fiction: a South Uist blight warning never fired because the humidity sensor read 89.5 in fog. A salt test and the fog fix it.
-
The gap was in the ruler: our raw developer against LibRaw, measured properly
ShootFlow's Canon renders looked 3-9% brighter than LibRaw's. The developer was fine: our check passed LibRaw a white level with black still in it.
-
A grid repair that moves generator voltages makes 28 of 34 PGLib cases AC-feasible
Moving voltage setpoints with P repairs the DC dispatch on 5 of 34 PGLib cases as planned; with a margin inside every limit, 28 of 34 at 0.17 to 0.34 % cost.
-
Our linear OPF repair was leaning on the AC optimum's voltages
At PGLib's flat 1.0 pu setpoints, our linearisation repair of the DC dispatch cleared 2 of 11 congested cases, down from 6 with the AC optimum's voltages.
-
The guard is the wrong lever: a pre-registered negative on small LLM-coding audits
A miss guard of 5 instead of 3 fixed two of three small LLM-coding audits that ran wider than a random sample, and gave back 28% of the gain elsewhere.
-
The header that jumped: one script moved, 135 of 507 page loads stopped shifting
A script at the end of each tool page stretched its header to full width after first paint. Running it inline cut shifting loads from 135 of 507 to 1.
-
The heading that only appears after you click
281 calculator and checker pages jumped from an h2 to an h4 once they drew a result. A static scan never saw it. A browser that clicks Run did.
-
The headline that was one line too tall for a small iPhone
After a shorter intro, the hub's search box still sat below the fold on 375 px and 320 px phones. A phone-only headline size and tighter gaps fixed it.
-
Heat-pump diversity helps a transformer 3.5 times. A synchronized night setback costs it 34.
Thermostat diversity cuts modelled transformer ageing 3.5x, but a shared 06:00 night-setback end with strip heat raises it 34x.
-
The idle time was a missing cache: timing every phase of our site build
A profile said our site build was idle 57% of the time. Timing each phase traced that to a missing cache; then three small changes cut 4 to 5 s.
-
Two sessions, one speed-up: what a merge conflict said about our perf runs
Two agent sessions profiled the same league scheduler search on the same day and found the same fix. One landed first; the other kept a smaller 6% win.
-
The Leica told us how its lens bends, and its own preview confirms we listened
A Leica T DNG stores its lens correction as a WarpRectilinear opcode. ShootFlow now applies it, matching the camera's own preview to 0.18 px.
-
165 lint errors on a clean main: running the linter version CI pins
Our local CI runner reported 165 ruff errors on code CI passes: a newer ruff sat first on PATH. It now runs pinned tools from a venv of the job's pip install.
-
The local test run missed 47 Python suites; now it runs them
run_venture_tests.py missed ventures whose Python tests live in their own workflow. Now it runs them with that workflow's directory and env; 44 of 47 pass here.
-
The bug-seeding tool now runs the test command the audit file wrote down
quality_mutate.py crashed on the cd-prefixed test commands that audit files record. It now runs them through bash from the repository root.
-
Sizing house water pipes with Hunter's rule, computed exactly
Free house water pipe sizing by Hunter's 99% rule, computed exactly. On a two-bath example the normal approximation picks a 3/4 in main; exact needs 1 in.
-
The one engine number that decides whether a cheap excavator pump saves fuel
A half-size fixed pump beats load-sensing on excavator fuel only if the rest of the machine burns under about 2.5 L/h. On the loader the threshold was 0.14 L/h.
-
The page list that fell behind: 30 pages no test was reading
Three venture page tests looped a hand-kept list of pages. The lists missed 30 pages; 11 failed checks their siblings pass. Now they read the folder.
-
The phone wrote down its own vignette, and we weren't reading it
Android phone DNGs store lens shading as GainMap opcodes. ShootFlow now applies them; three real camera DNGs match LibRaw within a tenth of a step.
-
The replay ran out of rows, not the guard: a metric watch that waits for a second look
A metric-watch guard failed its test by one change, first seen on the replay's last row. Scored fairly, it passes narrowly and is now the default.
-
The same line whatever the knots: a quarter off our trend-bend search
Every fit in our trend-bend search began by removing the same line. Doing that once, and each knot set once, cut a quarter of the time with identical output.
-
We built the same map 3,700 times: ten percent off a static site build
A CPU profile of our site build found 12 s of blog-renderer work redone for every post page. Three memos cut the build 10%, output byte-identical.
-
A league scheduler 27 times faster, with the same schedules to the last bit
The multi-division league scheduler spent 50 s in one fairness sweep. Tabling the costs once per night cut it to 1.85 s and every schedule hashed the same.
-
The search box below the fold
On a common Android phone our homepage search box sat below the first screen, behind a 13-line paragraph. Cutting what the page said again lower down fixed it.
-
The schedule search that recounted numbers it already knew
The league scheduler's carry-over search recounted costs nothing had changed. Caching them made it faster, with every schedule hashing the same.
-
The season between the choices: letting a league organiser pick from the whole frontier
The league scheduler's four fixed choices skipped stretches of the carry-over / back-to-back trade. Now every frontier season is a button.
-
The shortest pick route is cheap to find, and batching matters more
A free pick route and batching tool runs the exact Ratliff-Rosenthal method. On invented orders, grouped carts cut walking about 71%; routing alone, 13%.
-
The site's tests read every page twenty times
Twenty tests each re-read all 4,471 built pages. Reading each page once cut the tests after the build from 126 s to 58 s, with every test unchanged.
-
The slow part was the promise, not the hash
Naming the offline cache meant hashing 26,184 files. Async reads took 3.8 s, plain sync reads 1.4 s, with identical hashes.
-
The slow test was a hidden second build
A per-test timing of our site builder's tests showed one test spending 31 s on a whole build. Moving it into a child process cut the suite from 204 s to 180 s.
-
Why would new oak casks leak? A 2148 story about a drying schedule that outlived its forest
A story set in the Cognac country in 2148: a cooperage's casks weep because the forest now grows wider rings, and wide-ringed oak is denser and dries slower.
-
The tax lever was a parameter ratio: auditing our own simulated-collapse result
A toy civilization said a higher tax prevents collapse. In its equations, tax and state cost are one knob, and the post-collapse reset triggers the next.
-
Saudi e-invoices put tax on every line, and our checker counted it twice
Reading every TaxTotal in a ZATCA invoice as the document's raised false tax errors on all 19 of ZATCA's good examples. Reading only the root fixed it.
-
The third pass with the same slow read: three more seconds off the site build
Three more passes in the site build read every page through fs.promises. Sync reads cut each pass by about half and the build by 3.1 s, same bytes out.
-
The tool's name comes first now: two bars moved below the heading on 171 bid calculators
On 174 bid calculators the first line was 'Save your rates once', not the tool's name. The bars now sit under the heading. The h1 moved up 128 px on phones.
-
Why would a salinity sensor read fresh while the rice burns? A Kuttanad story about a salt wedge
A story set in a Kerala rice polder in 2052: the buoy reads 0.7 dS/m, salt water slides in under it along the bed, and wooden planks take the top water instead.
-
The UAE line amount that no rule checks
PINT UAE makes every line state its VAT and amount payable, but no rule checks they add up. We added INV-060 and tested it on seeded faults.
-
The width nobody checks: five tools that scrolled sideways on a tablet
Every page here is checked at phone width. Swept at 820 px, five of 8,679 scrolled sideways by up to 477 px, all from the same CSS grid rule.
-
The width rule on fresh categories: four new GoEmotions emotions, no penalty, no gain lost
A pre-registered check on four emotions no rule had seen: the audit rule was at most 0.1% wider than a plain audit in 12 of 12 cells, gain kept.
-
The workflow file nobody tested: a lost-state race in a self-hosted metric watch
Replaying a scheduled GitHub Actions workflow over 20 simulated days found a push race that would have re-sent the same alert. The fix is four lines.
-
The worst case for a beam over three posts is not every span loaded
On two equal spans, live load on one span gives 36% more sagging moment than on both, and lifts the far end. A free tool finds the worst pattern exactly.
-
The wrong pressure in a hydraulic break-even: a measured loader overturns our shortcut
On a measured wheel loader, open-centre surplus flow runs at 99.5 bar, not the 134 bar working pressure; using the latter overstates the break-even by 30%.
-
Seven drone projects shared a wrong yaw formula. Rotor cant still wins; the motor result does not
A full allocation shows seven canted-rotor studies understated yaw authority 2-2.8x. Rerun, cant keeps its 4.8° edge; the TWR-5 optimum goes.
-
Saints by Candlelight 100: Theodore the Studite and the icons in the garden
Episode 100 of the bedtime saints podcast: Theodore the Studite, who ran Constantinople's biggest monastery and walked its garden with forbidden icons.
-
Thirty-seven research milestone labels were used twice, and a tool that picks the next free one
Four research logs named 37 milestones twice, so a later citation of 5h or 8e could mean either study. A new reader gives the newest entries and a free label.
-
TD07, TD08, TD09: Italian invoice codes the FatturaPA schema refuses
Our FatturaPA checker passed TD07-TD09 and failed invoices with two withholdings. Moved to the spec 1.9 schema (XSD 1.2.3), it now agrees with xmllint.