Data Ownership and Privacy

This is the foundation for everything in self-hosting. When you understand what data ownership means—and why it's the primary reason people self-host—you'll know why you're bothering with Docker containers, reverse proxies, and Tailscale. Every later topic (Backup and Hardware Failure, Container Orchestration, TLS) exists to protect the control you gain here. Practically, you'll be able to assess any SaaS product and ask, 'Who owns the data I put in here?' and answer it honestly. You'll know why a local LLM like Llama 3 on your own machine matters, and why people accept the hassle of self-hosting rather than using ChatGPT. Without this concept, self-hosting is just a collection of commands; with it, every command has a purpose.

Think about your phone's photo library. When you take a photo, where does it live? If you use iCloud or Google Photos, the photo is copied to someone else's computer. That company's engineers can see it (or at least their algorithms can), the government can request it with a warrant, and if their servers are hacked, your photos are stolen—even if your own phone is locked. Compare that to a photo that only ever exists on your phone and your own backup drive at home. Nobody else can look at it, because it isn't anywhere else to be looked at. That's data ownership: the data stays under your control, on hardware you own, and it never travels to a third party unless you explicitly send it.

Now extend that to services. When you use Gmail, your emails live on Google's servers. When you use ChatGPT, your prompts go to OpenAI's data centres. When you use WhatsApp, your messages go through Meta's infrastructure. In every case, you're handing over your data to a company that has its own interests—advertising, training models, responding to legal requests—and its own vulnerabilities. Self-hosting is the decision to run those services yourself, on your own hardware, so that the data you generate stays with you.

The catch, and it's a big one, is that you now have to protect that data yourself. When Gmail's servers crash, Google's engineers fix them; when your home server dies, it's you who has to replace the hard drive. When a service like WhatsApp has a vulnerability, Meta patches it; on your own, you have to update your own software. Data ownership isn't a free lunch—it's a trade. You swap convenience and someone else's responsibility for control and your own responsibility. As the book Self-Hosting with Docker & Linux puts it, 'Total control means total responsibility.'

Imagine you own a small shop with a safe in the back. You keep your cash in that safe. Now, another company offers you a service: 'We'll hold your cash in our vault—it's much more secure, and you don't have to worry about keys or locks.' Sounds good, right? But their vault is shared by hundreds of other shops. If their security fails, your cash is gone. If they decide to charge you for access, you have to pay. If the government asks to see your records, they can open that vault without asking you. You've given up control to gain convenience.

Self-hosting is deciding to keep your own safe, even though it means learning to pick a good lock, maintaining the door, and making sure your keys aren't lost. The safe might be a rugged hard drive in your cupboard; the lock is your firewall and TLS certificate; the key is your SSH password. You—not some distant company—decide who gets in.

Where this analogy breaks down: a home safe can't be replicated for free. If your house burns down, your cash is gone forever. In the digital world, you can make copies of your data almost for free—that's what backups are for. So the analogy understates the risk: data ownership isn't just about keeping one copy safe, it's about keeping multiple copies safe in places you control. Also, a company's vault might actually have better physical security than your cupboard—the point isn't that your security is automatically stronger, it's that you control the decision about how much security you need.

Data ownership and privacy means keeping your data—photos, emails, documents, AI prompts, everything—on hardware you control, rather than on a third party's servers, so that you decide who can access it.

This is the core motivation behind every self-hosting topic in your library. When you set up Docker Compose, you're running containers on your own machine so the data stays yours. When you configure a reverse proxy and TLS, you're making that data safe to access from outside. When you set up Tailscale, you're creating a private network so you don't have to expose your data to the public internet. When you plan backups, you're protecting your ownership from hardware failure—because if you lose your only copy, you've lost ownership. And when you host a local LLM, you're extending this same reasoning to AI inference: your prompts never leave your machine, unlike ChatGPT where they go to OpenAI. This concept is the 'why' behind all those 'how' topics.