Unusual Attitude — privacy

← back to the map

Last updated 21 August 2026.

The short version. This is a hobby project run by one person. If you just look at the map, it records a little. If you create an account, it records considerably more, and it says so before you press the button. What is collected may be kept and used to improve the site and, potentially, to fund it.

I am not going to pretend this is a privacy-first product in order to look good. It is a hobby project that keeps what it is given. What I will promise is to make the best thing I can with it.

Who this is

Unusual Attitude is built and run by Joe Carroll, a private pilot, in his own time. It is not a company, not a commercial aviation product, and not affiliated with the FAA, NOAA or any of the data sources it draws on. It is currently at unusualattitude.pages.dev and will publish at unusualattitude.aero.

None of it is for navigation, and none of it substitutes for an official preflight briefing. That is said on the map itself, repeatedly, and it is not decorative.

Two different situations

What is recorded depends entirely on whether you have an account, and the two are deliberately not merged. An anonymous visitor never agreed to anything, so the things the map promises on screen are kept, and an anonymous visit is not attached to anyone's profile.

 Just visitingWith an account
Pages and layers usedYesYes, linked to you
Your IP addressYes, recorded and keptYes, linked to you
Approximate location from that addressYes, city levelYes, linked to you
Your network operator and its AS numberYesYes, linked to you
Browser, device, language, time zone, referrerYesYes, linked to you
Precise location, recordedOnly if you press “Where am I?” and allow itSame
Precise location, used to centre the mapOnly if you allow it — stays in your browser, never sentSame
Name and email addressNoYes, from your sign-in provider, or the address you asked for access with
Your aircraft, home field, ratings, hoursNoYes, if you tell us
Questions you ask the query boxNot availableYes, with the answer and the data used
Routes you planNot linked to anyoneYes, linked to you

What may be done with it

Plainly: information you provide, or that is made available by your visit, may be captured, retained and used — to run the site, to work out what to build next, and potentially to fund the site so it can keep running. Funding might mean advertising, a paid tier, or arrangements with third parties.

If that is not something you want, you can use the map anonymously and simply not create an account. Everything on the map itself works without one. The account exists for the query box, which costs real money to answer.

Three things that remain specifically true

These are facts about how the code works, not intentions, and they have not changed:

Your address, and what it tells us

Every request carries an IP address — that is how the internet works, and there is no version of this site that does not see one. What changed is that it is now written down rather than only glanced at. Alongside it the site records what Cloudflare already reports about the connection: the network operator and its AS number, the city and region that address maps to, and the browser you used.

This is used to tell a person from a machine. Mail security systems and link scanners open links automatically, and knowing that a click came from a datacentre rather than a home connection is the difference between "somebody read this" and "a robot checked it". It is also, plainly, how visits are understood at all.

Two details of that are worth stating outright rather than leaving implied.

The first is that the record is written on the server, before any JavaScript runs. That is the entire point of it. A link scanner fetches a page and never executes anything, so it is invisible to ordinary analytics — which is precisely how a scanner reading an invitation before its recipient does would go unnoticed. An arrival written here with no matching activity from a browser is a machine, and saying so requires recording the arrival itself.

The second is that alongside the address the site records how the connection was made: the HTTP and TLS versions, the encryption cipher your client chose, and the round-trip time. These describe the software at the other end, not you, and they are kept because they are far harder to disguise than a browser name — a User-Agent is one line of configuration, a TLS handshake is not. They are used for the same single purpose as everything above: telling a person from a robot.

Arrival records are kept for two years, the same as other operational logs.

The city-level position that comes with an address is the centre of a city, not you. It is never treated as your location and never shown as one.

The map asking where you are

The first time you pan or zoom the map, it offers to centre itself on you. Allowing that does not send your position anywhere. The coordinates are handed to the map inside your browser to move the view, and that is the end of them — nothing is transmitted, nothing is written down.

That is deliberately different from the “Where am I?” button, which does record a precise fix, because there the whole point is to put you on the map for others to see. Two different questions, asked separately, and answering one is not answering the other.

The map opens over Atlanta and stays there, or wherever you move it, unless you say yes. Ignore the card entirely and nothing is remembered either way, so you will be asked again next time. If you say yes or no, that answer is kept in your own browser — not on a server — so you are not asked twice.

Signing in

There are two ways in, and they record slightly different things.

Google sign-in. Google tells the site your name, email address, whether that address is verified, your profile picture and a stable identifier. It does not give the site your password, and the site never sees it.

A link sent to your email address, for anyone without a Google account or who would rather not use one. Here the site is told your address and nothing else: no name, no picture, no identifier from anybody. The link works once and expires fifteen minutes after it is sent, and a record of that request is kept so the link can be checked when it is used. Asking for a link is also rate limited per address, which is why asking repeatedly quietly stops sending rather than mailing you five times.

The reply to a sign-in request is the same whether or not that address already has an account. That is on purpose. A page that said "no such account" would be a way for a stranger to find out who has one.

Other providers were considered and are not offered, for practical rather than philosophical reasons: Sign in with Apple requires a paid developer programme, Instagram's consumer login was withdrawn in December 2024 and now only serves business and creator accounts, and X's free API tier was discontinued in February 2026.

A short trial, then a person decides

A new account is not approved and does not wait in silence either. It opens as a trial: 24 hours, and up to 15 questions on the query box, paid for out of the same budget an approved account uses. When the time or the questions run out, whichever comes first, the query box stops answering and the account moves into a queue that is read by hand. That queue exists because the query box calls a paid service, and letting anyone spend that budget unattended is not workable for a hobby project.

A trial ending is not a rejection and nothing is deleted by it. The account, its questions and its history stay exactly as they are, and approving it later simply starts the query box answering again.

Anyone can ask for an account at the access request page. It takes your email address, what you fly and your home airport if you care to say, and a line about why you would like access. All of that is sent with the request, together with the connection details described further up this page, which are recorded in the same way as for any other request to the site. The address is what the sign-in link is sent to, and it is what the account is then keyed on.

Accounts are for Part 91 operations. That is a declaration you make, not something the site can verify, and it is recorded as such.

Contributions, and what they buy

You can put money in. Half of it keeps this being built; half becomes usage credit on your account, which is then spent down at what your questions actually cost to answer. Because it buys something, it is a purchase and not a donation — it is not tax-deductible, there is no charity behind it, and nothing about it is recurring.

Payments are taken by Stripe. Your card number never reaches this site and is never stored here. What this site records is that a payment of a given amount completed, the Stripe identifier for it, and the credit and usage entries that follow — enough to show you a statement and to answer a dispute.

Credit does not expire, and it is not refundable once questions have been answered against it. If you have unspent credit and want it back, ask; it is one person and a database, and you will get a straight answer rather than a policy.

You can also use your own Anthropic API key instead. That key is stored in your browser only, is never sent to this site, and the answers are billed to you by Anthropic directly. The questions and answers are still recorded here, in the same way as any other — where the money came from does not change what the site keeps.

Keeping and deleting

Account records and the activity attached to them are kept while the account exists. Anonymous visit records are kept as operational logs. Payment records are the one exception to deletion on request — the amount, date and Stripe reference have to be kept for tax and dispute purposes even after an account goes, so those rows are detached from you rather than destroyed.

Ask and it goes. Email the address below and your account, your profile and the activity attached to it will be deleted. There is no form and no waiting period; it is one person and a database.

Data the site does not control

The map loads charts, imagery, weather and traffic from the FAA, NOAA, NASA, the Unidata NEXRAD feed, OpenFreeMap and others. Those requests go from your browser to them, and their handling of that is theirs, not mine. The site is served through Cloudflare, which sees requests as they pass. Payments go to Stripe, and answers on your own key go from your browser to Anthropic; both handle that on their own terms, not mine.

Children

This is not aimed at children and accounts are not knowingly created for anyone under 16.

Changes

This page changes when the site does. The date at the top is the last change. Material changes affecting account holders will be mentioned to them.