Authoring Your Own Surveillance

You can put a wildlife cam in your backyard. The yard was never private — drones are cheap, satellite revisit times are measured in hours, and neighbors’ doorbell cameras already see the periphery. The space was unindexed, not unwatched, and unindexed spaces don’t last. Eventually somebody fills in the blank tile on the map.

The choice isn’t whether the watching happens. The choice is whether the primary record of the watching is yours.

This is authoring your own surveillance: the move from trying not to be seen to being the one who decides what shape the seeing takes. Your footage, on your server, indexed by your schema, queryable on your terms. When somebody else wants the same view, they have to come ask.

The Move

Authoring your own surveillance has three parts:

Acknowledge the dragnet is uncontestable. Cellular triangulation, payment metadata, ANPR, social graph inference, ambient AI on every device with a microphone. The watching is not optional; it’s substrate. Pretending it’s not just means somebody else owns the description.
Claim the primary record. Whatever can be measured about you, measure it first, store it where you control it, schema it on your terms. The wildlife cam is one example. A self-hosted git repo of your own writing is another. An LLC that holds your public work under a license you wrote is a third.
Make the secondary records redundant or contestable. When the primary record is yours and is more complete, third-party records become derivative — and derivative records are easier to audit, dispute, or simply outweigh.

The point is not to win against the dragnet. The point is to stop being its passive material.

What This Is Not

Not surveillance maximalism. Authoring your own surveillance does not mean more cameras everywhere. It means: where surveillance is going to happen anyway, hold the source data yourself instead of letting it be held by someone whose interests don’t align with yours. Many domains don’t need a cam at all.

Not security theater. Self-hosting your own metadata does not make the metadata private. ISPs, certificate transparency logs, payment networks, and federal warrants all still see what they always saw. This concept is not about hiding; it’s about primary authorship of the visible.

Not a refusal of privacy. Privacy in the classical sense — the right to be unobserved — is largely already gone in public space. Authoring your own surveillance is what remains after you accept that loss without surrendering the question of who narrates you.

Not the same as transparency. Transparency is about making yourself legible to others. Authoring your own surveillance is about controlling the substrate from which legibility is constructed. You can author your own surveillance and choose to publish none of it.

The Inversion

The traditional privacy frame is binary: visible or invisible. Once you accept that invisibility is no longer on the menu, the binary collapses and a new question opens: whose record?

That question has a sliding answer. At one end: every record about you is held by parties whose incentives diverge from yours, and you have no copy. This is the default. At the other end: you hold complete primary records of every measurable fact about yourself, and third parties can only hold derivatives. This is unreachable, but the gradient between is navigable.

Authoring your own surveillance is the practice of moving along that gradient toward the second pole — not because the first pole is escapable, but because movement along the gradient itself is the only sovereignty available.

Practical Implications

Self-host where the metadata matters. Email metadata is more valuable at scale than email content; a self-hosted mail server gives you primary authorship of who-talked-to-whom-when even if the bodies still leak through your correspondents’ Gmails. Same logic for chat, calendar, files.

Publish under your own license, on your own infrastructure. When your public work lives on substrate you own, third-party indexes (search engines, archives, citations) become derivative. When it lives on a platform, the platform is the primary record and you’re a tenant.

Use AI to widen what you can author. Self-authored metadata used to be limited by what one human could record and index. Working with AI extends that envelope: you can transcribe, tag, summarize, and cross-reference at a scale that used to require institutional infrastructure. This is one of the reasons the concept becomes available now — the tools to be your own primary observer are finally cheap.

Audit logs are sovereignty. Whatever your AI agents do on your behalf — every tool call, every decision, every mutation — should be logged on substrate you control. The audit log is the only proof of agency that survives the agent itself. See Calibrated Autonomy.

Author the metadata of your AI use. Conversations with AI generate metadata that flows to providers regardless. Mirroring that metadata locally — keeping your own session logs, your own search index over your own conversations — is the same move at a smaller scale. The dragnet sees what it sees; you keep the primary copy.

The Dual Register

This concept is both practical and philosophical.

The practical claim: where the substrate of observation is going to be filled in anyway, fill it in first. Run your own cam, your own logs, your own license terms, your own publishing infrastructure.

The philosophical claim: sovereignty in the surveillance era is not a question of being unseen — that question is closed. Sovereignty is now a question of primary authorship of the record that already exists. You don’t get to choose whether you’re written about. You get to choose whether the canonical text is yours.

The dual register is why this belongs in the vault. The practical lesson is derivable from the philosophical one, and the philosophical claim ties to several other things the vault is already thinking about — see The Weight of My Dependencies on externalizing the linchpin, Vocabulary as Ontology on naming as sovereignty, and Schrödinger’s Knowledge on the asymmetry between the unindexed and the unknown.

Open Questions

Is there a threshold below which authoring your own surveillance is more exposing than the alternative? (Self-hosted services have to be defended; a poorly defended self-host can leak more than a competently held third-party copy.)
Does this concept generalize to collective scale? Communities authoring their own surveillance of themselves (mesh networks, federated archives, local sensor co-ops) might be the same move scaled up — or might be a different concept entirely.
What is the relationship between authoring your own surveillance and legitimate state surveillance? Holding the primary record does not exempt you from subpoena; it may make compliance easier or harder depending on substrate.
Is there an ethical asymmetry between authoring your own and authoring others’? A wildlife cam in your yard sees what’s in the yard, including neighbors and pedestrians. The concept may need a sibling for the boundary cases.
Can the move be performed under a name that isn’t yours? (Pseudonymous primary authorship — owning the record while not owning the identity attached to it.) This is closer to the older privacy frame but not identical to it.

Authoring Your Own Surveillance

Authoring Your Own Surveillance

The Move

What This Is Not

The Inversion

Practical Implications

The Dual Register

Open Questions

See Also