Plain language. No tricks.

• Your email. Required — it’s how we’ll tell you we’ve launched.
• Optional. The outlet you trust most, and your guess at its political lean. Skip both fields and the form still works.
• Standard server logs. Our hosting provider (Vercel) and our database (Neon) keep operational logs — IP address, user agent, request timestamps — for reliability and anti-abuse. We don’t go reading them to figure out who anyone is.
• In the app: counts, not a history. When you close the app it sends totals for that session: how many cards you saw, how far down you got, and for each story how many times it was opened, expanded, shared, or calibrated. Those numbers are added to a running total for that story on that day, mixed in with everyone else’s. Which stories you read stays on your phone.
• In the app: what broke. When the app crashes it sends the error message and the code path that failed, so we can fix it. We wrote this ourselves rather than adding a crash-reporting company, so nobody else receives it. The report carries no id of any kind — not your random app id, not which story you were reading — which does mean we cannot tell one person crashing fifty times from fifty people crashing once. That was the trade we chose. Emails, web addresses, file paths and long codes are stripped out of the text before it is stored, in case an error message happened to quote one.
• In the app: a random id. Made up when you install, so we can tell how many people come back the next day and the next week. It isn’t your phone’s hardware id, it isn’t linked to your email, and reinstalling makes a new one. It records which days the app was opened — never what was read on them.

• Your name. (Unless you put it in your email.)
• Your phone number, address, or anything else identifying.
• Ad-targeting data, cross-site cookies, third-party trackers. None of that today — and when the first third party does arrive, it’s named at the bottom of this page before it ships.
• Your location.
• Anything about your religion, caste, or political party.
• Which stories you read. The app knows, so it can stop showing you the same card twice. We don’t.

This one is worth being specific about, because Nispaksh is a product about political coverage and “trust us” is not good enough.

What you read is not recorded against you. The table that counts reading has no column for who did it. A story gets a number for the day and everyone’s swipes land in the same number. There is no query we could run — and none a court could order us to run — that returns which stories a particular person read, because that pair is never written down.

What you deliberately place on the spectrum is. When you calibrate an outlet’s coverage, that placement is stored against your app’s random id, because the only way to count one vote per person is to remember who has already voted. So for the handful of stories you actively calibrated, a row exists linking that id to that story. We are telling you because it is true, not because anyone would have noticed.

The distinction is real and worth holding onto: reading leaves no per-person trace, and a deliberate vote does. That id is not your name, your email, or your phone’s hardware id, and the app’s Settings screen erases it and every placement attached to it in one tap.

Your network address is counted, not kept. Every request arrives carrying an IP address — that is how the internet works, and no service can decline to receive one. Because anyone can invent an app id, an address is the only thing stopping one person from posing as a thousand readers and skewing a public tally. So we keep a short-lived count of requests per address. What gets written is a scrambled form of the address, produced with a key held on the server; someone who obtained the database without that key could not read an address back out of it. That is the whole claim — not that the address is unknowable, but that counting works without it being written down. The counts expire within a day, and they are joined to nothing: not your app id, not your placements, not anything you read.

That’s a property of how the thing is built, not a promise about how we behave. Those are different, and only one of them survives us changing our minds.

• Email. To send you one launch email, and at most a couple of follow-ups directly relevant to that.
• Outlet & lean guess. Two reasons. One, we want to know how Indian readers actually perceive their news ecosystem — useful for building the product. Two, when we launch we want to send you a personalised email like “you said The Hindu is Center; the research says Center-Left, here’s why that’s interesting.” Not a gotcha. We mean it kindly.
• App counts. To find out which stories people actually open the comparison for. That number — did the summary make you want to see how six outlets framed it — is the one thing that tells us whether the product is working, and there is no way to get it except by counting.
• Server logs. Standard operational stuff. We don’t use logs for marketing, advertising, or building any picture of you.

In a Postgres database run by Neon, on AWS infrastructure in the Asia-Pacific region. Only people running Nispaksh can access it.

We don’t share it with advertisers, data brokers, marketing partners, or any third party for their own use. We will share it if a court tells us to, and we’ll tell you about it unless we’re legally barred from doing so.

Write to contact@nispaksh.com and we’ll do any of these:

• Show you everything we have about you.
• Fix anything that’s wrong.
• Delete you entirely — we’ll do it within 30 days and write back to confirm.
• Stop emailing you. Every email we ever send will have a one-click unsubscribe link too.

When the app launches it will carry a “delete my data” control in Settings that removes its random id, the record of which days you opened it, and every spectrum placement you made. You won’t need to email anyone or prove who you are, because we have nothing to check you against — the app simply asks us to forget the id it made up, and mints a new one.

These rights come from India’s Digital Personal Data Protection Act (2023) and basic decency. We honour them whether you’re in India or anywhere else.

Your signup stays on file until you ask us to delete it, or until 24 months after Nispaksh stops being an active project — whichever happens first.

Server logs follow our providers’ defaults: Vercel keeps them for around 30 days, Neon for around 7. The per-address request counts described above are dropped within a day of their last use.

App counts are anonymous totals with nobody’s name on them, so we keep those. The random app id, its record of which days you opened the app, and the spectrum placements attached to it are deleted 24 months after your last session — or the moment you use the delete control, whichever comes first.

We don’t set any marketing or analytics cookies. The hosting layer sets a few security cookies (anti-bot challenges, for instance) — we don’t add anything of our own at this stage. The app uses no cookies at all.

This isn’t built for under-18s and we don’t knowingly collect anything from them. If you think a minor signed up, write to us and we’ll delete the entry.

We’ll update this page as the product grows — accounts, saved stories, comments, whatever else. If anything material changes, we’ll email existing subscribers before it takes effect.

One thing we already know is coming, so you can hear it now rather than in a diff. The app will carry ads, because someone has to pay for it — they will be non-personalised, meaning the advertiser gets no profile of you and no interest categories, only the space on the screen. That means a third party is involved for the first time, and it will be named here and declared on the Play Store listing before it ships.

Questions, complaints, deletion requests, or just curiosity: contact@nispaksh.com.