Sensing in public places requires noticeable and persistent notice

What it is

Sensing in public space — cameras, microphones, people counting, Wi-Fi probes — requires notice to the people covered that is noticeable (visible on entry, checkable while present) and persistent (coextensive with the sensing, not fading over time). A one-time announcement, a slogan glimpsed at the entrance, small lettering on the device, or a link to terms: none of these constitutes effective notice — most of the people being sensed never learn that sensing is happening.

"Persistent" mirrors the temporal structure of sensing: collection is a continuous behaviour, and the notice's visibility must not fall below the sensing's presence. An adequate sign answers three things: there is sensing here; what kind it is (image / sound / counting); and who operates it (an accountable and approachable party).

The contrast with personal devices is the default entry point: a phone has system-level indicators and a permissions page to consult; public sensing has no corresponding "settings entry", and the physical sign is the only surface where it shows. The notice standard for public sensing must therefore be higher than for personal devices, not aligned with them.

Why it happens

• Attention competition: a notice in public space competes with hundreds of visual elements; weak notices (small type, high placement, jargon, single exposure) are attentionally nonexistent. Survey research on CCTV signage recall repeatedly finds very low numbers — a wide gulf separates "a sign was posted" from "people know".
• Notice as commitment: posting "recording here" is not just an information transfer but the operator's self-binding — what is publicly declared (type, purpose, retention) becomes the anchor for later accountability. Conversely, concealed sensing, once discovered, destroys far more trust than the sensing itself deserved: the "covertly recorded" narrative draws its sting from the concealment, not the footage.
• Invisible coverage extents: a camera's field of view, a microphone's pickup radius, a probe's range are not visible from outside — the covered cannot even tell whether they are in range. Notice therefore also carries the function of externalizing coverage, a layer personal-device notice never needs.

Studying it

• Signage recall surveys: intercept people entering and leaving, measuring three layers of recall — noticed the notice, can name the sensing type, can name the operating body. Low recall is the stable result in CCTV-signage effectiveness studies, correlating with sign position, size and pictorial form.
• Surveillance studies: qualitative work on notice practices and the relation to trust and behavioural adaptation, explaining why formally compliant notices still fail.
• Compliance audits: checklist audits against transparency requirements (notice position, language, pictograms, registry access) as the institutional-side metric.

Methodological caution: treating "a sign exists" as the pass criterion is formalism — the terminal test of effectiveness is recall. Studies must distinguish "never noticed" from "noticed but content not retained"; the first calls for changes to position and size, the second to content comprehensibility.

Where it stops holding

• Effective notice is not consent: notice solves "do they know", not "can they refuse" — the absence of choice is structural in public monitoring, and notice must not be packaged as a consent credential.
• Notice intensity should track intrusiveness: anonymous counting and face-recognizing imagery do not carry equal notice duties. A uniform maximal standard overburdens low-intrusiveness sensing; a uniform minimal standard hands highly intrusive collection a "we count visitors" fig leaf. Grading is necessary — and misgrading is itself misleading.
• Dense multi-source spaces cannot be signed source by source: with ten systems in one hall, per-source signs are unreadable and infeasible; a space-level consolidated notice (list of sensing types in this area plus a single enquiry point) is needed, with the per-source registry retained for query.
• Sustained maintenance is routinely forgotten: fading, occlusion and post-renovation invalidity break "persistent"; a notice without a maintenance plan fails automatically over time.

Applying it

• Put signs in the entry path's field of view (eye level, not high on a wall), pictogram plus the main local language, with the three elements: type icon, one-line purpose, operator and complaint contact.
• Place signs at the boundary of the covered area, not beside the device — devices hide in ceilings and corners; only boundary signs externalize coverage. Mark large zones with ground markings or entrance signs.
• Update signs and registry whenever sensors are added or purposes change, making notice maintenance part of the facility-change process rather than a one-time sticker.
• How to check: periodic intercept surveys of the three recall layers (noticed / named the type / named the operator), accepting on recall rather than on "signage installed"; when recall falls below threshold, locate the failure layer (position and size vs content readability).

Related

• Same group: Z8.03.2 Notice must state the purpose of collection, not merely the fact · Z8.03.3 The lack of an exit is what separates public monitoring from personal devices · Z8.03.4 The duty to inform lies with the venue operator, not the individual
• Nearby: Z6.06 Awareness of passive collection (home setting) · Z6.08 Physical indicators of sensing state
• Search terms: CCTV signage · surveillance notice · public sensing · transparency