Camera Glasses Misconceptions: 'Only What Others Could See', 'Already Out There', and Open Source
'It only records what others could see' is true and beside the point; 'a camera is one more set of eyes' assumes a human observer with human limits; 'it's already out there' conflates self-surveillance with other-surveillance. Google Glass didn't prove the market rejects camera glasses — it proved people reject looking like they're wearing one. The novel element is identification rather than recording, and it lives in the AI backend, which open-sourcing the hardware doesn't touch.
Corrections for the camera-glasses privacy debate, kept out of the concept chunks. ## About the objections to the objections **"It only records what other people could see anyway" is true and beside the point.** A recording contains no information a co-present human could not observe. The harm is not the observation but what recording does to it: permanence, perfect recall, searchability, and instant distribution. Being seen and being recorded are different acts even when the pixels are identical. See Practical Obscurity: The Privacy You Actually Have in Public. **"A camera is just one more set of eyes" assumes a human observer.** Humans forget, cannot be everywhere, cannot identify strangers, and cannot cross-reference. A networked camera with AI has none of those limits, which makes the quantitative jump a qualitative change. **"It's already out there" conflates self-surveillance with other-surveillance.** Phones and payment cards track the person who chose to carry them. Glasses record the people *around* that person, who chose nothing and receive nothing. See Two Axes of the Camera-Glasses Debate: Your Data Versus the Bystander's Consent. **"Some privacy is already gone, so resisting the rest is futile" is the mechanism of normalisation creep**, not an argument against it. Each erosion justified by the previous one has no stopping point. **"Critics propose no solutions" is empirically wrong.** Tamper-proof indicators, facial-recognition bans, on-device-only processing and biometric-consent laws are all concrete proposals — and recording indicators exist *because* of the pushback. ## About the history **Google Glass did not prove the market rejects camera glasses.** It failed on social grounds — conspicuous, expensive, and it marked its wearer. Later glasses that look like ordinary eyewear sell well. What was rejected was *looking like* you were wearing one, and the visibility that killed Glass was also the bystander's notice. See Google Glass and the Glasshole Backlash. **Google Glass didn't fully die in 2015.** Consumer sales ended then; the enterprise edition continued until 2023. **The Kodak panic is a real precedent and a partial one.** The 1888 Kodak provoked "Kodak fiends", venue bans, and alarmed press — and the panic faded by around 1910. But it also produced Warren and Brandeis's 1890 "The Right to Privacy", the foundation of American privacy law. Previous cycles produced law and norms rather than passive acceptance, so citing the cycle as evidence that concern is pointless gets the history backwards. See The Privacy Panic Cycle: From Kodak Fiends to Camera Glasses. **Fading concern is not demonstrated consent.** People may stop reacting because they no longer recognise the device. Habituation and informed acceptance look identical from outside. ## About what is actually new **The novel element is identification, not recording.** A Kodak could photograph your face; it could not return your name, address and relatives in seconds. Objections framed purely around "being filmed" miss the part that has no historical precedent. **I-XRAY invented nothing.** It chained a livestream, a face-search service and a language model — all commercially available. That is why it mattered: the capability is commodity assembly, not a research breakthrough. It was also never released, was not maintained past late 2024, and was published with instructions for removing yourself from the databases it relied on. See I-XRAY: The Demo That Turned Wearable Recording Into Real-Time Identification. **The doxxing capability lives in the AI backend, not the glasses.** Any camera can feed a face-search service. Regulating the eyewear without touching facial recognition databases addresses the visible part of the problem rather than the operative one. ## About open source as a fix **Open source addresses trust and control, not bystander consent.** It lets you verify the firmware and removes the advertising incentive — real gains for the wearer, none for the person standing next to them. **It can make the bystander problem worse.** Open firmware makes removing the indicator light and adding facial recognition trivial. A black market already does this to closed hardware. **"Controlled by nobody" means "nobody accountable."** A platform owner can push a mandatory security fix because it controls the platform. A decentralised ecosystem has no party to enforce norms, patch vulnerabilities, or bear liability.