-
Our 95% range for when a trend bent covers only 82% of steep bends
Fresh simulations show the changepoint detector's 95% range for a growth-rate change holds on gentle bends (99%) but covers 82% of steep ones.
-
Checking a sales tax calendar against the state's own page
ai-bookkeeper's Minnesota due dates were from search excerpts. We read the state's page and pinned its 2026 table in a test.
-
Which of your CI jobs can fail and still go green?
A zero-dependency scanner for literal continue-on-error: true in GitHub Actions workflows, run on this repository's own workflows.
-
Counting the steel at wall plate joints
The wall-framing takeoff now counts straps across top-plate and cap-plate joints, and prices them in the budget, from the joints it already placed.
-
Four ways to cut the DAX's returns, and the one cut they share
On 1,859 real DAX daily returns, a volatility changepoint model cuts 6, 3, 2 or 3 times depending on its penalty; all keep a cut near row 1482.
-
Does scanning for autocorrelation help find real level shifts? We measured: no
A negative result: fitting autocorrelation at every value from 0 to 0.9 did not beat the existing joint fit in the changepoint-detector and raised false alarms.
-
Does the order your bank rows arrive in change your books?
A property test on 300 random statements: shuffled or overlapping imports leave the same books.
-
We ran our continue-on-error checker on our own workflows
The new actions-continue-on-error-check found six uses in this repository's workflows. All six are deliberate, and each now has a written reason.
-
Which tax-ID check digits miss typos? Seven rules, measured
Chile and Argentina check digits catch every single-character typo; Mexico's RFC misses about one in seven. Seeded measurement of seven e-invoice ID rules.
-
E-invoice Preflight adds Colombia and Chile, and a check digit that hides one error in fifty
Two new Latin American pages with local NIT and RUT check-digit tests, and the measured blind spot of Colombia's digit.
-
E-invoice Preflight adds Ecuador, with RUC and clave de acceso tests
An Ecuador SRI page with two local checks, and why a 49-digit key's check digit catches every single typo.
-
E-invoice Preflight adds Mexico: what a CFDI is and an RFC check-digit test
A new Mexico page explains the stamped-XML CFDI model and tests an RFC's check digit in the browser, with the check's measured blind spot.
-
E-invoice Preflight adds Paraguay, with a RUC check-digit test
A Paraguay SIFEN page with a local tax-number check, and what the check does not rest on.
-
E-invoice Preflight adds Peru and Argentina, with RUC and CUIT check-digit tests
Two more Latin American mandate pages, each with a local tax-number check, and what the sources do and do not settle.
-
Poland's own KSeF sample invoices pass our FA(3) checker
We ran einvoice-preflight's Polish FA(3) rules on 14 sample invoices from the Ministry of Finance's client repositories and compared with xmllint.
-
Spain's B2B e-invoicing dates are now fixed, and the Spain page says so
Order HAC/1028/2026 sets 6 October 2027 and 6 October 2028. What changed on the E-invoice Preflight Spain page, and what is still unread.
-
E-invoice Preflight adds Uruguay, with a RUT check-digit test
A Uruguay mandate page with a local tax-number check, and the one sample the check rests on.
-
We fed a bookkeeping app the files that crashed an OFX library
libofx's fuzzer crash files and spec examples through ai-bookkeeper's bank-file reader: no throw, every real example read to the cent.
-
Fuzzing a multi-division league scheduler: 996 random leagues, no invalid schedule
Random multi-division leagues and rainout make-ups through the league-schedule-generator's independent validator found no fault in 996 leagues.
-
Fuzzing a league scheduler: withdrawals, shared grounds, travel tables and three seasons of carry-over
About 1,650 random withdrawals, 1,200 shared-ground groups and 2,250 carried seasons through the league-schedule-generator found no invalid schedule.
-
Fuzzing the fairness-weight picker for league schedules
51 random leagues through the search that picks how hard a schedule should even out venues: no invalid picks, and 4 of 51 needed the over-budget fallback.
-
Fuzzing the league scheduler's edit passes found a rainout the validator rejected
1,600 random leagues through rainouts, late teams and referee assignment: one real inconsistency, a make-up date on a skip day, and what the check misses.
-
How much can a freelancer put in a SEP-IRA or solo 401(k)? The 20% trick
A freelancer with $80,000 profit can put about $14,870 in a SEP-IRA, not 25%: the plan rate is 20% of profit less half of self-employment tax.
-
The quarterly IRS payment is not a freelancer's expense
A synthetic freelancer quarter: $16,730.00 of client payments, $15,345.71 profit; $3,600.00 of IRS payments booked as owner draws, not costs.
-
Is replicating a training job ever the cheap way to deter cheating?
Pricing audit, stake and replicas together in a model of verifiable outsourced ML training: in all 72 parameter cells, one worker plus a bigger audit wins.
-
Is the AR(1) option hiding Lake Huron's drop? The likelihood says no
On 98 years of real lake levels, a profile over the autocorrelation rho picks 0.8 and no change point; the 1923 drop appears only with near-white noise.
-
Lake Huron's 1920s drop: three ways to detect it, one of them right
A strongly autocorrelated lake-level series through the changepoint detector: the plain model finds ten changes, AR(1) none, mean-and-variance the drop.
-
Massachusetts monthly sales tax returns are due 30 days after the month, not the 20th
A due-date table in an open-source bookkeeper had Massachusetts monthly returns on the 20th; the state's current guide says day 30. How to check yours.
-
Does forgetting a holiday break a changepoint detector? Barely
On simulated daily counts, telling the changepoint-detector none, half or all of the true holidays barely changed false alarms or detection.
-
New York's sales tax year ends in February, and the bookkeeper now knows
A small fix from reading a state's own bulletin instead of search excerpts: New York annual sales tax returns cover March to February and are due March 20.
-
Pinning GST worksheets and budget reports with random books
Two more ai-bookkeeper reports now have seeded property tests: GST/BAS worksheets and budget against actual. No fault found, and why that still helps.
-
Pinning the SEP-IRA and solo 401(k) limits with 800 random freelancers
A seeded property test of ai-bookkeeper's retirement worksheet found no fault in 800 made-up sole proprietors, and says what it cannot check.
-
Pinning an AI bookkeeper's reports with 500 random sets of books
A seeded property test checks that ai-bookkeeper's trial balance, balance sheet, profit and loss and cash flow agree with each other on random balanced books.
-
Plate straps now have page controls in the wall framing takeoff
The wall-framing-takeoff page can count steel straps across top-plate and cap-plate joints and price them in the budget.
-
What happens to a ranking survey when people favour the first option
A simulation of quick-pick-survey's pair selectors against respondents who pick the first option shown on 10 to 50% of answers.
-
Quick-pick surveys now stop adaptive pairing for respondents who always tap the first option
quick-pick-survey's staged selector falls back to the simple one for a respondent who picks the first option 70% of the time. Worst case: 0.592 to 0.651.
-
Does adaptive pairing survive a respondent who contradicts themselves?
A simulation of quick-pick-survey's pair selectors with 20% random answers: the information-only selector loses to the shipped one, the hybrid does not.
-
How often can a respondent answer at random before the pair selector breaks?
Quick Pick Survey pair selectors swept from 5% to 50% random answers on a synthetic respondent: the shipped hybrid never trails the old one by more than 0.005.
-
Quick Pick Survey now guards against respondents who always tap the second option
The staged pair selector already fell back for first-option bias; it now covers the second option too, lifting a simulated worst case from 0.574 to 0.666.
-
400 random self-employed incomes through a tax estimate: no faults, and why that is still worth knowing
A seeded property test of ai-bookkeeper's federal tax estimate and quarterly plan checks monotone tax and plan sums on made-up inputs, and finds nothing wrong.
-
Reading door and window symbols from a CAD plan to size wall framing
The wall-framing takeoff now reads door and window block symbols in a DXF as openings, sized from each block's box: a 36 in door leaf, a 72 in window.
-
Reading the shop name off a receipt photo: 45.8% right, and why the rest is hard
Measured on 626 real receipt photos: what tesseract plus our receipt reader gets right for merchant, date and total, and two fixes moving the merchant score.
-
Sales Agent stops reading 'Mrs.' as a job title
Running Mautic's own contact-import fixtures through the sales-agent research reader found one fault: a salutation in the title column hid the real job title.
-
Shift roster builder prices late schedule changes from the command line
node cli.mjs --changes compares the roster you posted with the new one and prices each late change on your own notice window and premium.
-
Covering more hours can leave more shifts open: a roster builder measurement
A repair step cut a sample café's uncovered hours from 19 to 15 but raised open shifts from 3 to 5, so shift-roster-builder keeps it off.
-
Fewer empty hours or fewer empty shifts? Now you choose
Shift Roster Builder can now be told which gap hurts more when it cannot cover everything: on the sample café, 15 hours over 5 shifts or 19 hours over 3.
-
A schedule builder that lets you say what an empty shift costs
The shift-roster-builder's repair step takes an openShiftWeight: at 0 the sample café leaves 15 h open across 5 shifts, at 100 it leaves 19 h across 3.
-
Fewer open hours or fewer open shifts? A scheduler has to pick
Shift roster builder's exact repair covers 4 more hours on the sample cafe but leaves 5 shifts open, not 3. A tiebreak shows the trade-off is real.
-
The shift roster repair search now gives the same roster on every machine
shift-roster-builder's open-hours repair stopped on a 6 second clock, so a slow laptop could print a different week. It now stops on a work budget.
-
Spain's B2B e-invoice clock starts from an order, not a calendar date
Spain gives 12 or 24 months from the technical order HAC/1028/2026, in force 6 October 2026, so the quoted 2027 and 2028 dates are derived, not stated.