Plainly
WHAT THIS SITE RECORDS
People come here to find out what happens if the police search their car, or whether a landlord can walk in, or what a charge carries. A list of those questions with a name or an address beside it is a thing that can be breached, subpoenaed, or read by whoever holds a database key next year, and no promise written on a page stops that happening. Not collecting it does.
So this is not a policy in the usual sense. It is a list of what the code does, and it is checkable: everything below is in the open source that runs this site.
WHAT IS KEPT
What people search for
The words typed into the search box, which state they were on, and which result they opened. Nothing beside it: no address, no account, no session, no cookie, no fingerprint of a browser.
Why: It decides what gets written next. A question people keep asking that the guide cannot answer is the most useful thing this site can learn.
That a request was made, on the three things that have a ceiling
A counter, and nothing else, on the answer agent, the contact form and the search box. The number is kept against a salted hash whose salt carries the date, so the same visitor is a different number tomorrow and a week of activity cannot be lined up into one person's history. What was asked, written or searched for is never in this counter at all, and neither is an address: the hash is computed here and the database only ever sees the result.
Why: Each of the three can be called without signing in, and each costs something when it is called in a loop: the answer tool costs money per question, and the contact form sends mail from a domain this site owns, which a flood would get marked as a spam source. Counting is all a ceiling needs, and counting does not require knowing who.
Aggregate traffic
Pageviews, which page, and where the visit came from, through Vercel Web Analytics. It was checked before it went in: the script sets no cookie and writes nothing to localStorage, sessionStorage or IndexedDB, and it is served from this domain rather than someone else's. There is nothing in it that can recognise the same person twice.
Why: Without it there is no way to tell whether anybody is reading this, or which parts. It is the difference between building the guide people use and the guide we guessed at.
An account, if you make one
Your email address, which is how the sign in link reaches you, and the slugs of the sections you saved. Nothing else. Not what you searched for, not what you asked the answer agent, not which pages you read: those are written without an account beside them whether you have one or not, and there is no column in any of them that could hold one.
Why: So the file somebody is assembling is still there on the next device, and after the phone is gone. It is the only thing on this site that follows a person, and it follows them because they asked it to.
A message, if you send one
What you write on the contact page, and an email address only if you choose to type one. Leave it blank and the message arrives anonymous. Your address is not recorded either way.
Why: So a gap somebody reports can be answered, when they want an answer.
WHAT IS REFUSED
These are not oversights that might be filled in later. Each one is a thing the site could easily do and does not.
- A reason to make an account. Every section, statute, case and charge is readable without one, and reading creates no record of you at all. An account does one thing, which is keep your saved file across devices, and what it holds is listed above.
- A cookie, unless you sign in. Reading the guide sets none, on any page, ever. Signing in sets a session cookie, because a session is what a cookie is for: it carries the session and nothing else, script cannot read it, and signing out removes it. There is no tracking cookie here at any point, for anybody.
- Your IP address, anywhere. Not in the search log, not beside a contact message, not against a question asked of the answer agent. The ceilings described above count a salted hash of it that rotates every day and is computed before anything is stored, so what the database holds is a number that cannot be turned back into an address or lined up with tomorrow's.
- A visitor identifier for anybody reading. No id, no fingerprint, and nothing that connects two visits by the same person. Somebody signed in has a session, which is the thread they asked for, and it reaches their saved file and stops there: the search log, the contact messages and the answer agent's counter are all written with no account beside them, and the answer agent's counter never stores the question at all.
- Anything YouTube would have learned. The video embeds use the no cookie host and the player is not loaded at all until you press play, so watching nothing means YouTube is told nothing.
ON YOUR DEVICE ONLY
Two things are remembered in your own browser and never sent anywhere: whether you chose the light or the dark theme, and any sections you added to a reading list. Clearing your browser data removes both, and nothing on this site notices.
THE ONE OUTSIDE PARTY
The traffic counter is Vercel Web Analytics, and it is the only third party that runs on these pages. What was verified about it is above. What Vercel does with the counts on their side is theirs to describe rather than ours to paraphrase, so it is worth reading from them directly: Vercel’s analytics privacy documentation.
If something here is wrong
Tell us and it gets fixed, or the page gets corrected. The contact page takes a message without an email address if you would rather not leave one.