Journal / Interactivity
02 · Designing interactionThe user of a public installation did not read a manual, will give it about two seconds, and is being jostled. Design for that person or design for nobody.
The first failure of interactive installations is silent: people walk past because nothing told them they could touch, move or respond. A screen showing finished content reads as a video, and a video is something you glance at, not something you engage. Successful interactive pieces signal their interactivity unmistakably — an attract loop that demonstrates the interaction, a visible prompt, motion that responds to a passer-by before they have committed. Discoverability is not decoration; it is the difference between an installation that engages a steady stream of people and one that engineers a clever interaction nobody ever discovers. The cleverer and more novel the interaction, the more work the signposting has to do, because novelty means no one arrives knowing what to do.
A public audience grants almost no patience. If the first interaction is not understood and rewarded within a couple of seconds, most people abandon it — and worse, their visible confusion teaches the next person that the thing is broken or boring. This forces a brutal simplicity: the entry-level interaction must be instantly legible and instantly rewarding, with any depth layered underneath for the few who stay. Designers who build the rich, multi-step experience first and the obvious first move last get it backwards. Start from the single, dead-simple action that works in two seconds, then earn the right to complexity. Depth that nobody reaches because the doorway was confusing is depth wasted.
Sensor-driven interaction — gesture, position, presence — demos beautifully because the demo has one cooperative person in known light. The crowd is the opposite: many bodies confusing a tracker, variable lighting defeating a camera, people standing exactly where the sensor's assumptions break. The reliability you measured in an empty room is not the reliability you will get on a busy Saturday, and the gap is largest for exactly the camera-and-gesture technologies that look most impressive. The defensive move is to choose the most robust sensing that still delivers the idea, test it deliberately under crowd and lighting stress before launch, and never trust a single quiet-room demo as evidence it will work in public. A sensor that works ninety per cent of the time fails one visitor in ten — and that visitor tells the next one.
Because interaction will sometimes fail, the only safe design is one that fails into something acceptable. When the touch layer dies, does the piece become a still-attractive display, or a frozen error? When the sensor is overwhelmed, does it fall back to an attract loop, or freeze? The installations that earn long lives are the ones whose worst day is merely a non-interactive but presentable screen, not a public advertisement of malfunction. Designing the failure state is not pessimism; it is the single highest-leverage decision in interactive work, because the failure state is what real visitors will eventually see, and it is what they will remember.
03 — Spatial content design →