What the directory reports about itself, and why we show it instead of fixing it
LGD's tables contradict each other in small, countable ways. The API keeps every one of them as a note on the row, so your users see what the government says — not what we guessed.
The Local Government Directory is published as a dozen reports per state, and they were plainly not generated from one consistent model. Loading Maharashtra and Uttarakhand, we found the same handful of contradictions in each:
- A village mapped to a Gram Panchayat whose Zilla Parishad sits in another district — three cases, each a same-named Gram Panchayat picked by mistake.
- Urban bodies whose coverage spans two districts: real for Greater Mumbai, plainly wrong for a Nagar Panchayat in Uttarakhand said to cover a tehsil 250 km away.
- The same ward number and name entered twice in one body, with two codes — 803 rows.
- Census-2011 codes shared by two villages, or swapped between two neighbours.
- A Zilla Parishad spelled differently from the district it governs (Udam Singh Nagar, Udham Singh Nagar).
And the silences: villages in no block (1,506), inhabited villages in no Gram Panchayat (2,562 — almost all of them inside towns), bodies with no wards, coverage given at district level only.
The choice
We could "clean" this. We chose not to, for one reason: the moment we do, the API stops being the government's directory and becomes our opinion of it. A customer reconciling their records against LGD needs LGD, contradictions included.
So every finding is kept as a row in an issue table and surfaces on the affected record in two forms: issues, the machine codes, and notes, the same thing in words — "Gram Panchayat Is Under Another District's Zilla Parishad", "Duplicate Entry In LGD", "No Gram Panchayat In LGD". A dropdown can show the note and still offer the row. Nothing is hidden; nothing is invented.
The one thing we do change is casing. LGD's names arrive as a mix of ETAPALLI(NP)WARD NO-3, saouth side balkhandi and proper Title Case; we normalise them, keeping the abbreviations LGD uses (NP, MB, MC, CT, NMC, OG) and roman numerals in capitals. That carries no information, so it is safe to touch. Everything else is as published.