Compartmentalise your digital life (one leak shouldn't expose everything)
Most people live their whole online life in one browser signed into everything at once, which is why a single breach, nosy service or data broker can join the dots between your work, shopping, banking and public posts. Compartmentalising keeps those identities in separate boxes (browser profiles, Firefox containers and a different email per service) so a crack in one does not drain the rest; it is course 2 of the Going Ghost track, for lawful personal privacy, not for deceiving anyone, committing fraud, evading a ban you earned, or dodging a lawful investigation.
- Course 1 done first: Build your personal threat model. Compartmentalising has a real convenience cost, so only build the walls your model actually calls for
- A desktop browser. Firefox for the containers part (free, from mozilla.org); profiles work in Firefox, Chrome or Edge. The steps link each one
- Optional, only for the sensitive-only space in step 6: roughly 10GB free disk for a lightweight virtual machine, or a spare account on your operating system. Skip it entirely if your threat model does not call for it
Why compartmentalise (and why not to overdo it)
Understand what linkage is, why separating identities limits a breach, and why more boxes is not automatically better.
Picture your online life as one big room with everything in it: your work email, your bank, your shopping accounts, your real-name social profile and that forum you post on anonymously. Now picture a burglar getting into that one room. They get everything. Compartmentalising is the simple idea of using several smaller rooms instead, so a break-in to one does not reach the others. Security people call this limiting your blast radius.
The thing you are defending against is linkage: the quiet work of joining your separate identities back into one profile of you. Services and data brokers do this using whatever you leave lying around. The main joins are:
- Email address. The strongest join of all. Use the same address on a forum and at a shop, and anyone who sees both now knows they are the same person. A leak from either one carries that address to everyone who buys the leak.
- Phone number. Even harder to change than an email, and demanded by many services. One number across accounts ties them together permanently.
- Cookies and logins. A tracker embedded on thousands of sites (a 'like' button, an ad script) can see that the same browser, already signed into your real-name account, also visited everything else. That is how one signed-in tab quietly labels your whole day.
- Device fingerprint. Even with no login and no cookies, your browser gives off a stew of details (screen size, fonts, graphics card, time zone) that together are often unique enough to recognise you again. More on this in step 4.
Compartmentalising attacks the middle two directly (keep cookies and logins in separate boxes) and sets up step 5 to attack the first (a different email per service). No single move stops all linkage, but stacking a few cuts most of it.
The trade-off: separation costs convenience, and more is not always betterThis is the same lesson as course 1, applied here. Every extra box is:
- More logins to manage. Separate boxes mean separate sign-ins, and no shared autofill between them. A password manager makes this bearable; without one, you will give up in a week.
- More friction, every day. 'Which profile am I in?' becomes a real question. Open a link in the wrong box and you are logged out and confused.
- A chance to leak through the cracks. Paste a work link into your throwaway box, or sign into the wrong account once, and you have linked the two things you were keeping apart. The separation is only as good as your discipline.
So the goal is not the maximum number of boxes. It is the right number for the risks you wrote down in your threat model. For most people that is two or three (a real-name box, a shopping-and-signups box, and maybe a sensitive box), not twenty. We will keep coming back to that.
TipA quick way to decide how far to go: look at your threat model's top risks. If 'a data broker joins my accounts' or 'a service leak exposes my real name' is on the list, this mission earns its keep. If your only worry is spam, you may only need step 5 (email aliases) and can skip the heavier separation.