Privacy policy

Last updated: 25 October 2026.

This page explains what Psi Census collects about you, what it keeps, what it publishes, and what it cannot take back once published. We have tried to say only what is true of the site as it is built today.

Who runs this site

Psi Census (psicensus.com) is run by Ben Pearson, Alabama, USA. You can write to privacy@psicensus.com about anything on this page, and to admin@psicensus.com about anything else.

The short version

What we collect when you play the study test

When you first open the study test and confirm you are 13 or older, your browser asks our server for a device code. This is a random code with a signature, so we can recognise that the same browser is sending batches. It is not your name, and we do not link it to any other information about you.

Each time you finish a batch of 25 guesses, your browser sends us:

Our server adds the time it received them, gives the batch a random batch number, and stores all of that in our database. We do not store your name, email address, browser type or network address with the batch.

What stays only on your device

These are stored in your own browser (local storage under psicensus.com) and are never sent to us:

The device code is also kept in your browser. The server cannot read your browser’s storage. If you clear your browser data, you lose the device code, your receipts and your streaks, and we cannot restore them.

What is published, and why it cannot be erased

Every submitted batch is published through the site’s public data export (and later in quarterly data releases that anyone can download and check). This is already true of everything submitted so far: all of it is in the public record now. For each batch the record contains: the batch number, the test name, your 25 guesses, the time we received them, whether it came from a player or from our own test robot, and the result worked out from the public random value.

The record does not contain your device code, the batch key, or anything else that identifies your browser.

The record cannot be erased or edited, for three reasons:

  1. Each batch is part of a chain of fingerprints that independent timestamp services signed and that is also anchored in public records. Removing or changing one batch would break the chain, which is exactly what the chain is there to expose.
  2. Taking out batches after the answers are known would let anyone, including us, hide results that did not fit. The study rules forbid it.
  3. Copies are made by researchers, mirrors and archives, and we cannot recall them.

What can be done is unlink the batches from your device. The “Delete my data” button on the Study test page rewrites the private device code stored beside your batches to a fresh random value that nobody holds, so nothing on our side connects those batches to your browser any more, and it clears this browser’s data. It does not remove your batches from the public record. There is no other way to ask for it: we cannot unlink batches by email or any other route, because only the device code in your browser proves which batches are yours.

Please remember: if you share your receipt, a link to the check page, or a batch number, anyone holding it can find that batch in the public record. Your guesses and the time are not a name, but you made them, and the record can be linked to you by whoever you tell.

Participant numbers

Later, when data releases include participant numbers (so researchers can count distinct players), each device will get a random number. It is not derived from your device code or from anything about you. Once a number has been published it is never changed or split. The link between your device and that number is kept privately by us and is not published.

Scrambled network address

To limit abuse, our server briefly keeps a scrambled (salted, one-way) fingerprint of your network address, for about an hour, in its memory and short-term storage, to count how many browsers register from it. The address itself is never stored or logged, and the fingerprint is never written to our database or published.

For completeness, here is how it works. We mix your network address (for IPv6 only its first half, the network part) with a random secret that we replace every 24 hours, and keep only a shortened result. That is used to count how many new device codes a single address asks for in an hour (at most 200). A separate short-term record keeps, for each device code, the times of its recent batches (so that no device sends more than one batch every 30 seconds and 60 an hour). This record contains your device code, not your address. It is kept for about an hour after the last batch it counts and is never published. If you erase your data, this short-term record is not removed at once: your device code may remain in it for up to about one hour, until it expires, and an erase does not reset the batch limit.

Who else sees what

How long we keep things

WhatWhereHow long
Your batches (guesses, time, batch number, result)Our database and the public recordPermanently; cannot be erased (see above)
Device code and batch key stored beside your batchesOur database, private, never publishedUntil you use Delete my data on the Study test page; then it is replaced by a random value nobody holds
Per-device batch times (device code, count, last batch time)Server short-term storageAbout an hour after the last batch
Scrambled network-address fingerprintServer memory and short-term storageAbout an hour; the daily secret is replaced every 24 hours
Backups of the databaseCloudflare’s point-in-time historyUp to 30 days. A device link we remove stays in those backups until they expire
Practice history, streaks, receipts, the 13+ note, the device codeYour browser onlyUntil you clear them or your browser does
Our own server log linesCloudflare’s log view, if it is switched onOur code writes only technical codes (for example “a scheduled job ran”), never your device code, guesses or network address

Cookies and trackers

Our pages load only our own style sheet and script from psicensus.com. There are no ads, no analytics and no third-party trackers on the site, and our code sets no cookies. Your browser’s local storage is used for the items above. This is required for the test to work.

The support page links to Ko-fi, an independent service. Following the link sends nothing from this site beyond the request your browser makes to Ko-fi; Ko-fi’s own policy covers what you give it. We do not list or publish supporters.

Children

You must be 13 or older to take the study test, and we ask you to confirm that before anything is sent. This is a self-declaration; we do not check it. The study test is not meant for children under 13, and we do not knowingly collect data from them. The practice test is not gated, and sends us nothing about your guesses. If you believe a child under 13 has used the study test, write to privacy@psicensus.com; but note that we can only act on batches we can link to a device code (see “Your choices” below).

Sign-in

Signing in is not offered yet. If it is added, you would sign in with Google, and this page will be updated first to say exactly what is stored, and nothing about it applies until then.

Your choices and rights

Changes to this policy

If we change what we collect or publish, we will update this page and its date before the change takes effect, and for a significant change we will say so on the study test page. Older versions of this page are kept in our public repository.