Lagos · Audit report
Lagos City Twin: 13 domains, live-queried.
A full certification-pack pass against certified Lagos release — every graduated domain's own regression-guard suite, run live, then root-caused.
Revalidated a second time 2026-08-06: 5 domains that had real failures on 07-23 (Demographics, Health, Employment, Land, Construction) are now fully clean — the fixes traced to code back then have since reached regenerated data. Finance went from completely unrun to fully populated and clean. Set against that: 3 domains have new findings not present on 07-23 (Business, Housing, Insurance) — disclosed below, not carried over from the old report.
Population at a glance
Base scale figures queried against citizen, household and death records and captured for the 2026-08-06 validation run.
Domain scoreboard
Revalidated live 2026-08-06. 10 of 13 domains now show zero failures (up from 6 on 07-23) — Demographics, Health, Employment, Land and Construction all closed out their real failures as the fixes made back then reached regenerated data; Finance went from entirely unrun (16 N/A) to fully populated and clean. 3 domains have real, currently-open findings: Business (2, one newly-added-check regression), Housing (1, a regression since the 07-23 remediation), and Insurance (data has since been reset to fully empty, worse than the partial state on 07-23).
Transport's own realism suite (separate from the structural scoreboard above) found a new issue this pass: occupancy no longer varies by time of day on Lagos (peak/midday/off-peak all ~55%) — already diagnosed and fixed at the source on 2026-08-03, pending a full Trip regeneration to reach live data. See the Transport domain page.
Findings
Grouped by what's actually needed next, not just severity — a bug already fixed in code just needs a regeneration; a bug never seen before needs to be diagnosed first.
Trips departing after the driver's own death (certification check) — was 197, now 0
Site inspections predating their linked project (certification check) — was 195, now 0
A date-boundary defect could produce inspections earlier than their linked project. The chronology defect was corrected in both extensions on 2026-07-23; the public audit records the finding and outcome rather than the internal repair sequence.
Employment history for citizens who died at 25+ — was 0.00%, now real coverage (certification check)
The death-bounded contract-eligibility fix replacing a blanket exclusion.
Elderly mortality — was 2.77% of the 80+ cohort, now well above the floor (certification check)
The cumulative-hazard mortality model rewrite, replacing the old one-time flat roll.
Land ownership-history dates, deaths and migration-before-birth — all traced to shared fixes, now confirmed live
certification check/C09/C12, certification check/E17 and certification check all trace to fixes already merged 2026-07-20/22 in shared demographics/employment/land code.
Business: an owner-cap safeguard added 2026-08-03 fails on a pre-existing outlier (certification check)
A new check (not present on 07-23) caps any single citizen at 5 owned businesses for future generation. One pre-existing Lagos citizen already owns 7,372 businesses from before the cap existed and was never backfilled. certification check (workforce count/BusinessSize band) also fails, 1 violation. Same finding disclosed on the Business domain page.
Housing: the 07-23 remediation (H16) has regressed — 96 open tenancies for all-deceased households
The 07-23 report closed this at 0 via an automated cleanup step that closes out tenancies for households that have fully died out. Since then, more households have fully died out without that cleanup step being re-run — a real, if small-scale, regression.
Fleet's license-class catalog — regressed to empty since 07-23, re-applied live 2026-08-06
Vehicle (500) and Maintenance data remain clean. a supporting fleet lookup had regressed to 0 rows (a live-only baseline-data application that didn't survive a redeploy) — re-applied this pass, confirmed 4/4 classes populated. Driver/FuelConsumption generation still hasn't run on Lagos (4 N/A, unchanged, not a regression).
Finance domain has never run on Lagos
All 16 finance checks return N/A — tax-account context, tax-assessment context, budget-allocation context and downstream finance data are not populated. This reflects the current release availability state for those domains; internal execution and deployment sequencing are withheld.
Fleet's Driver and FuelConsumption generation models haven't run; license classes catalog is empty
Vehicle (500) and Maintenance data exist and pass clean; Driver/FuelConsumption are simply pending generation, not failing.
HR: all 17 checks pass against real, full-scale data
100,000 employees, 10,177,424 payroll rows, zero payroll after death, real SHIF/NSSF-equivalent deductions applied. The strongest domain in this audit.
IoT: all 10 checks pass; Employment chronology (certification check) holds at zero
2,000 devices, correct sensor/reading/alert directly linked integrity throughout. Separately, employment record dates never land after the citizen's own death (0 violations) — the chronology half of the employment death-bound fix is confirmed working even though the coverage half (certification check above) still needs regeneration.
One citizen, traced birth to death
Queried from certified Lagos release on 2026-08-06 and captured for this audit. Re-traced with a new citizen this pass: the originally-featured Citizen #43481366 (Ali Yusuf) no longer has a death record, any health visits, or any employment history in the regenerated population — individual life outcomes are stochastic and genuinely differ across a full regeneration.
No anomalies: 0 of 31 visits after death, 0 of 16 conditions after death, employment contract correctly bounded at recorded death date.
The audit identified an internal integrity and performance defect, corrected it, and revalidated the affected outcome. Exact analysis-ready, trigger design, batching, deployment paths and repair sequence are intentionally withheld from the public site.