Accountability
Scoreboard
Every council that has published records or received questions, with what can be measured about it. Accuracy is the share of a fixed, published set of test questions the answer service got right against that council's records. The server scores each run itself from the answers it gave, and every run can be inspected case by case.
| Local government | State | Published records | Questions, 30 days | Answered | Open gaps | Test-set accuracy |
|---|---|---|---|---|---|---|
| Obio/Akpor | Rivers | 31 | 179 | 87% | 12 | 100% (60 cases, 11 Oct 2026, scored by the server, case by case) |
| Ifelodun | Kwara | 0 | 4 | 0% | 1 | not yet run |
| Offa | Kwara | 0 | 2 | 0% | 1 | not yet run |
| Obi | Benue | 0 | 1 | 0% | 1 | not yet run |
| Kano Municipal | Kano | 0 | 1 | 0% | 1 | not yet run |
| Ilorin West | Kwara | 0 | 1 | 0% | 1 | not yet run |
"Answered" means answered from published records; a refusal is counted as not answered even when refusing was correct. Open gaps are distinct questions the LGA has not yet added a record for. Demo LGAs carry illustrative records and are marked. Records compiled from public sources are labelled on the council's pages until the council confirms them. "Staff-reported" runs predate server scoring and carry only the totals staff submitted. The test set is eval/questions.json; a run is only as broad as that set. How answers are checked