TRANSPARENCY FIRST
How the Constituency Rating Works
Every number on the Seedhi Baat leaderboard is derived from a transparent, auditable formula. This page explains exactly how we calculate constituency scores, protect anonymity, and decide what data MPs see.
The Growth Score Formula
The Growth Score ranks constituencies on the leaderboard. It is a composite of four signals, each weighted by its importance as a measure of genuine civic engagement — not just raw sign-up volume, which is easy to game.
SIGNUP COUNT (40%)
Total active members in the constituency
The absolute number of people who have joined Seedhi Baat from a given constituency. This is the primary signal — it measures how much of the constituency has chosen to participate. Weighted heaviest because it directly reflects civic mobilisation.
WEEKLY GROWTH RATE (30%)
New signups as a % of existing base, week over week
A constituency that grew 20% last week signals momentum — someone is spreading the word, a local issue is galvanising people, or a ward leader has picked it up. Growth rate prevents a large but stagnant constituency from permanently outranking a smaller, faster-growing one.
REFERRAL RATE (20%)
% of members who referred at least one other person
Organic referrals are the strongest quality signal. If people are inviting their neighbours, it means the product is delivering real value and the civic feedback feels meaningful. Referral rate is calculated as (members with at least one referral) ÷ (total members).
RESOLUTION RATE (10%)
% of filed complaints marked resolved
Complaints marked resolved by the filer indicate that the system is working — feedback led to action. Weighted lowest because resolution depends partly on factors outside Seedhi Baat's control (government responsiveness, issue type, MPLADS fund cycle). It is a quality indicator, not a primary rank driver.
How Anonymity Is Preserved
Anonymity is not a feature toggle — it is the architecture. Here is exactly how it works.
- No phone numbers in public display. If you choose to sign up with a phone number (for notifications), it is stored in a separate, encrypted table. The public leaderboard, constituency pages, and complaint feeds never contain phone numbers. Your phone number never appears in any export or MP report.
- Anonymous IDs.Each complaint is assigned a random, non-reversible anonymous ID. There is no lookup table that maps anonymous IDs back to users, even internally. A complaint cannot be linked back to a specific person — not by Seedhi Baat, not by an MP, not by a court order on Seedhi Baat's systems (because the linkage does not exist).
- Aggregate reporting only. MPs in Phase 2 receive only category-level aggregate data: "152 complaints about Roads; 87 about Water Supply; 23 about Air Pollution." No individual complaints, no names, no anonymous IDs, no metadata about when a complaint was filed.
- Public mode is explicit opt-in. If a user chooses to be public (to organise their neighbourhood, for example), they make an active, informed choice. The default state of every new complaint is anonymous.
How Feedback Aggregation Works
Raw complaints become constituency intelligence through three mechanisms.
CATEGORY BUCKETING
Ten standard civic categories
Every complaint is tagged to one of ten categories: Roads & Potholes, Water Supply, Electricity & Power Cuts, Garbage & Sanitation, Street Lighting, Drainage & Flooding, Air Pollution, Health Facilities, Education, and Other. This standardisation makes constituency-level totals meaningful and comparable across geographies — a complaint about a blocked drain in Lucknow and one in Hyderabad are both counted under Drainage & Flooding.
CO-SIGN WEIGHTING
Co-signs amplify but don't duplicate
When a citizen co-signs a complaint, it increases that complaint's visibility on the constituency page and in the MP report. Co-signs are counted separately from original complaints — a complaint with 50 co-signs is not counted as 51 complaints. The Growth Score uses co-sign volume as a quality signal for the referral rate component.
SPAM DETECTION
Rate limiting and de-duplication
A single session cannot file more than 5 complaints in 10 minutes. Complaints with identical text filed in the same constituency within 48 hours are flagged for review and excluded from MP reports (though they remain visible on the public feed with a low-confidence marker). This prevents coordinated complaint flooding that would distort the data.
Phase 1 vs Phase 2 Data
What is available — and to whom — changes when a constituency crosses the 40-member threshold.
| Data point | Public (Phase 1) | MP report (Phase 2) |
|---|---|---|
| Total complaints by category | Yes | Yes |
| Top 5 categories ranked | Yes | Yes |
| Co-sign counts per complaint | Yes | Yes |
| Month-over-month trend | Yes | Yes |
| Individual complaint text | Yes (anonymous) | No |
| Individual user names | No | No |
| Phone numbers | No | No |
| Resolution rate by category | No | Yes |
| Ward-level breakdown (sub-constituency) | No | Yes (Phase 2+) |
| Referral network map | No | No (never) |
Why This Methodology
The design of Seedhi Baat's methodology is grounded in three bodies of evidence.
Civic tech research consistently finds that participation barriers — language, digital literacy, fear of retribution — are the primary reason citizen feedback systems fail to scale in India. The anonymity-first, vernacular-first design directly addresses these barriers. A 2019 study of e-governance adoption in Rajasthan found that anonymous reporting increased complaint volume by 3.4x compared to identity-required systems.
RTI effectiveness research demonstrates that civic engagement produces the best outcomes when it generates a verifiable, time-stamped record. The co-sign mechanism on Seedhi Baat creates exactly this — a public, immutable record that a given number of people in a constituency cared about a specific issue on a specific date. This is the kind of evidence that changes political calculations.
Democratic accountability theory argues that elected representatives respond to aggregated constituency sentiment more reliably than to individual petitions. A single complaint about a broken road is noise; 500 co-signed complaints about roads from a Lok Sabha constituency of 1.5 million voters is a signal that cannot be ignored without electoral consequences. The 40-member threshold before MP notification is calibrated to ensure the signal is statistically meaningful — not a single bad actor, but a genuine community concern.
We publish this methodology publicly so that citizens, MPs, journalists, and researchers can audit, challenge, and improve it. Accountability flows both ways.