Two Axes of the Camera-Glasses Debate: Your Data Versus the Bystander's Consent

Arguments about camera glasses persistently talk past each other, and the reason is that two distinct concerns share the word "privacy". Nearly every popular argument in the debate lands cleanly on one axis and does nothing on the other. ## Axis one: control over your own data Who holds the data your devices generate about *you*, what they do with it, and whether you can verify any of it. This is where most people's intuitions live, because it is the axis their own experience is on. Arguments that operate here: - **"It's already out there — credit cards, phone GPS."** Substantially true on this axis. - **"Open-source it and remove the corporation."** A real fix here. - **"I don't want companies and governments having more of my data."** ## Axis two: the bystander's non-consent Whether the person standing next to a wearer — who bought nothing, carries nothing and agreed to nothing — can know they are being recorded and identified. This is where the backlash lives, and it is the genuinely new problem. The distinction that makes the axes come apart is **self-surveillance versus other-surveillance**. A phone and a credit card track the person who chose to carry and use them, in exchange for something. Camera glasses capture *everyone around* the person who made that choice. Existing systems mostly instrument participants; glasses instrument the people next to a participant. That is a categorical difference, not a difference of degree. ## Scoring the common arguments **"Only captures what others could see anyway."** Deflates the hysterical version, leaves the substantive core intact: being seen and being recorded-identified-stored-aggregated-shareable are different acts even when the pixels are identical. See Practical Obscurity: The Privacy You Actually Have in Public. **"It's already out there."** Right on axis one, and two problems on axis two. First, "some privacy is already gone" would make every incremental erosion unresistable, each step justified by the last — which is the mechanism of normalisation creep rather than an argument against it. Second, it is *different* data: a card knows a transaction; glasses plus AI know your face, your movements, your conversation and your identity continuously — the first-person visual layer that ties every other dataset to an identified physical person. **"Critics only object and propose nothing."** Empirically false. Tamper-proof recording indicators, bans on facial recognition, on-device-only processing, and biometric-consent statutes are all live proposals. More importantly, the objection *is* the mechanism: recording indicators and tamper-disable responses exist because of pushback. This is a negotiation over terms, not progress versus obstruction — the same shape as seatbelts and traffic law. ## The open-source scorecard Open-sourcing the glasses is a serious proposal and deserves a serious answer rather than dismissal. **What it fixes.** Auditable firmware means you can verify that the indicator light cannot be silently disabled, that nothing is uploaded covertly, and that no facial recognition is running locally. It dissolves the manufacturer-specific trust deficit and removes the surveillance-advertising incentive entirely. On axis one, it is close to a complete answer. **What it does not fix.** The bystander is still recorded and identified without consent. Who controls the device is orthogonal to that: open source protects the **wearer's** data, not the **bystander's**. **How it could make things worse.** Open firmware makes stripping the indicator light a one-line change and adding facial recognition a feature. There is already a black market for physically disabling the LED on closed hardware; open hardware makes that trivial and legitimate-looking. And "controlled by nobody" means "nobody accountable" — a platform owner can push a mandatory fix precisely because it controls the platform, while a decentralised ecosystem has no party able to enforce norms, ship security patches, or bear liability. **And the identification lives elsewhere anyway.** The doxxing capability is in the AI backend, not the glasses. Perfectly ethical open-source glasses can stream to a face-search service. Open-sourcing the hardware does not touch the facial recognition databases — that is a separate fight, and the more important one. ## The synthesis No single lever works. "Ban it" ignores that the capability is commodity technology; "open-source it" fixes the axis that was less broken. The combination that addresses both axes is: **transparency** (auditable devices), **hard limits on the identification layer** (regulating facial recognition and biometric databases, which is where the real harm is concentrated), **a consent signal that is difficult to defeat**, and **social norms** about where wearing them is acceptable. The throughline: arguments that feel obviously right to the person making them usually feel that way because they are correct on axis one. The backlash is on axis two, which is newer, harder, and not addressed by anything on the first list.