Record with consent
An account is written down, dated, and given a reference number before any guidance is offered. Consent is taken in writing and covers what may be published, which is never the identity of the person giving it.
Research & documentation
Documentation first, analysis second, publication only if it survives
Our research is documentation before it is analysis. We record what people report, we check it against records that exist independently of the account, and we publish only the part that would survive someone trying hard to disprove it.
The method
Six steps, in order, every time
The order matters more than any individual step. Analysis before corroboration produces a confident answer to the wrong question.
Corroborate against records
Wherever a document should exist — an order, a receipt, a register entry, a file noting — we look for it, usually through the Right to Information Act. An account that no record can support is held, not published.
Look for the pattern
A single account is a grievance. A recurrence across time, offices, or applicants is a finding. We publish at the level of the pattern, because that is the level at which something can be addressed.
Separate identity from analysis
Identifying detail stays in the confidential register and never enters a draft. Before publication we test whether the published form could be reverse-engineered to a person, and generalise until it cannot.
Read it as a hostile reader would
Every draft is read once by someone whose job is to attack it: what does this not establish, what else could explain it, which sentence claims more than the evidence carries. What survives is published.
Correct in public
If something we published turns out to be wrong, the correction carries the date of the original error and the date of the correction. It is not edited in quietly.
What we look at
Five recurring subjects
These are the areas where accounts keep arriving and where records exist to test them against. We do not research a subject we cannot corroborate.
Administrative delay
Where statutory timelines exist and are routinely missed, and what the missing of them costs the applicant. Delay is the most common failure and the least reported one.
Petty extraction
Small demands at the point of service delivery — the payments that are individually trivial, collectively enormous, and almost never documented.
Access to entitlements
The gap between eligibility and receipt in welfare schemes, and how much of that gap is documentation rather than qualification.
Information refusal
Grounds cited for refusing RTI applications, how often those grounds hold on appeal, and how many applicants never test them.
Rights at the point of contact
What actually happens at police stations, hospitals, and municipal counters, measured against what the law guarantees there.
Limits
What this method cannot do
Every one of these limits narrows what we are entitled to claim. We publish them alongside the method because a method without its limits is a sales pitch.
We are not a statutory investigator. We cannot compel testimony, summon records beyond what the RTI Act allows, or require anyone to answer us.
Our sample is the people who reach us. It is not representative of the population and we do not present it as though it were.
We do not publish allegations against named individuals. A pattern can be described; a person cannot be accused by us.
Where a matter is before a court, we do not publish analysis that could interfere with it.
Learn the method
The Transparency Fellows cycle teaches exactly this
A short, unpaid fellowship in reading public documents properly. Unglamorous, and the only training we run that changes how people argue.