Experience architecture
Every other page here follows a signal from a console to a nervous system. This one is about what you do with that chain when the job is not "make the signal arrive" but "make something happen to a person over ninety minutes" — which is a different discipline, and the one all the rest of it is for.
The premise
You cannot design a feeling. You can design the conditions and the timing.
A designer cannot put an emotion into somebody. As the emotion page sets out, a feeling is constructed from a stirred-up body plus a reading of what the stirring means — and the reading is theirs, not yours.
What you can do is entirely material. You can decide where attention goes and when. You can raise and lower how activated a body is. You can set up an expectation and then meet it, delay it, or break it. Everything else — the fixture choice, the delay time, the subnet, the render budget — is a means of doing one of those three things reliably.
This reframe is not a philosophical nicety. It changes what a production meeting is about. "Make it more emotional" is unactionable. "Raise arousal here, and give them a reading for it, and this is what it is a contrast against" is a plan.
The structure
An experience is a sequence, not a set of moments
The most common failure in ambitious work is not a weak moment. It is a strong moment with nothing around it — a peak with no approach, a reveal with nothing established to reveal against, an ending that arrives while the audience is still catching up.
The shape below is not a formula and plenty of good work deliberately breaks it. It is a checklist of stages that exist whether or not anyone designed them, because a person walks through all of them regardless.
The lever
Shape the running order, and see what survives it
The peak-end rule says that what gets remembered is roughly the average of the most intense moment and the final one — and that duration barely counts. Shape a running order below and watch the difference between what happened and what is kept.
The technical decisions that are actually experience decisions
Where the rest of this site lands
Almost every number elsewhere on this site is an experience choice wearing engineering clothes. These are the ones that come up most.
Latency is authorship. Above a very short threshold a visitor stops feeling they caused the thing that happened, whatever the causal chain actually is. This is why interactive work is a determinism problem before it is a content problem.
Where a sound comes from is presence. Localisation is fast and involuntary, and a voice arriving from a PA rather than a body is one of the strongest breaking cues there is — the whole reason object audio and tracking exist.
Darkness is a resource with a recovery time. Dark adaptation takes twenty to thirty minutes and any bright cue spends it. The blackout in act two is not the same blackout as the one in act one.
Sightlines and seams are attention. Anything with contrast takes the spotlight, so a visible edge, a spill, or a technician crossing upstage is competing directly with whatever you were pointing at.
Comfort is not hospitality, it is capacity. Cold, cramped, needing the toilet, unable to see — every one consumes attention that the piece needed.
Level is arousal and exposure. Both at once, always. Check it against the dose tool and DIN 15905-5.
Designing for it going wrong
Graceful degradation is an experience discipline
Something will fail. The question that separates a good production from a lucky one is whether the failure modes were designed or discovered.
The useful move is to decide, in advance and for each system, what the rungs are: what fails invisibly, what fails visibly but survivably, and what stops the show. Write it down, tell the crew, and rehearse the middle rung — because that is the one people actually meet and the one nobody practises.
Two principles are worth stating plainly. A failure that is acknowledged is far less damaging than one that is pretended away — an audience forgives a hold and remembers ten minutes of visible flailing. And anything that can hurt somebody fails to safe regardless of the experience cost: that is the arming chain, the safety channel, and it is never a trade-off.
Everyone, not most people
Accessibility as architecture
Treated as compliance, accessibility is a set of things bolted on in week eleven that satisfy a checklist and delight nobody. Treated as architecture, it is the same design problem as everything else on this page: deliver the experience through a channel the person actually has — which is precisely the sensory substitution argument.
Hearing loops are specified by IEC 60118-4 and venue obligations by legislation such as the ADA standards. Captions and surtitles are not a transcript; they are the text design of the piece, and where they sit decides whether somebody watches the show or watches the words. Audio description is dramaturgy — somebody is deciding what matters in each moment. Haptic vests and tactile transducers deliver a mix through the body rather than the ear, and are increasingly used by hearing audiences too because they are simply good. Relaxed performances change the arousal design, not the content. And the sightline from the wheelchair space is a design decision that gets made by whoever puts the desk in, deliberately or not.
Done early it is cheap and it improves the piece for everyone. Done late it is expensive and it feels like an apology.
Knowing whether it worked
What can and cannot be measured
You can measure some real things: where people went and how long they stayed, what they touched, whether a room fell silent, physiological synchrony across an audience, structured self-report at scale. All of that tells you about a distribution, not about a person, and it is noisier than anyone selling it admits.
What you cannot do is ask people whether they were present, because asking makes them notice the room. Behaviour is far better evidence than a questionnaire — flinching, leaning in, going quiet, staying afterwards, what they describe first to whoever meets them outside.
And the honest measure that survives everything: what do people say about it a week later, and which moment do they say first. That is the peak-end rule reporting back, and it is worth more than most instrumentation.
- The most common mistake is a peak with no approach. Ambition spent on one moment, and nothing built to make it land.
- The second most common is an under-designed ending. Half the memory, and the part everyone is too tired to work on.
- Novelty is not a strategy. It works on press night and is inert by the third week, for the crew if not the audience.
- Comfort is capacity. Every discomfort is attention removed from the thing you built.
- You are not the audience. You have seen it a hundred times, you know where to look, and you cannot un-know it. Watch somebody else watch it.
The whole thing, as a practice
Where each stage of this site does its work
This is the argument the section is arranged to make. Read from the bottom of the chain rather than the top and the stages stop being topics and start being a method.
Foundations
How something physical becomes a signal, how that signal is written down as numbers, how it survives a wire, and what the plug on the end of it actually carries.
The signal
What happens on the control bus, on the network, in the air and on the crew's headsets — and why each one fails in its own characteristic way.
The room
Turning a signal into something an audience experiences, and drawing it accurately enough that other departments can trust you.
The system
Several technologies behaving as one thing. A shared clock, a shared coordinate system, and what happens when either is assumed rather than agreed.
What you build with
Software talking to software, the languages and platforms it is written in, and the engines that draw a world sixty times a second.
The person
Every threshold in this industry is a measurement of a nervous system. What it can detect, how it works, how a feeling is built, and what makes somebody believe they are somewhere.
Machines that copy us
Each of these is an attempt to do in a tool what the previous stage does in a person: read the world, decide something about it, and act. Seeing them that way is what stops them being magic.
It draws on perception and emotion research, presence research, and a lot of accumulated production practice, and it takes positions. If your working method differs in a way that has survived real audiences, open an issue — that is exactly the kind of knowledge that normally stays in people's heads.
Where to go next
These pages are a chain rather than a library — the next one along usually answers the question this one raised.
Everything on this page, in the index
Each of these has its own entry with sources, ports, gotchas and an open gap where a field is missing.