The Building Needs a Harness

A South China Morning Post story this week showed an office scene that looks slightly absurd at first: people sitting under parasols indoors, wearing face coverings, applying sunscreen, and turning empty drink cups into makeshift humidifiers.

The article described a Chinese social-media conversation around “office ageing.” Some workers attributed skin changes, eye strain, dryness, and discomfort to harsh lighting or poor office conditions. One person complained about second-hand smoke. Others responded with personal equipment at their desks.

I do not think the interesting design question is whether every claim inside the trend is medically correct.

The interesting part is the workaround.

The parasol is an interface

A parasol inside an office is a strange object because it makes a hidden relationship visible.

The building produces a condition. The person does not feel able to inspect or change the building. So the person inserts a new layer between the environment and the body.

Glare becomes a parasol.

Dry air becomes a humidifier.

Noise becomes noise-cancelling headphones.

Temperature mismatch becomes a personal fan, blanket, or heater.

Air-quality anxiety becomes a purifier or mask.

The building did not design these interfaces. The occupants did.

This is useful product-research evidence. Repeated workarounds are often more revealing than asking people which feature they want. A workaround shows where the official system stops being sufficient.

It is a revealed requirement.

The design question becomes: why does the occupant have to adapt to the building rather than the building adapting to the occupant?

Most of the building is invisible

An office looks legible because we can see the room, windows, lights, vents, thermostat, furniture, and people.

But much of the state that affects experience is hidden.

Carbon dioxide concentration is invisible. Fine particles are mostly invisible. Humidity is difficult to judge precisely. Airflow can be local and uneven. Temperature varies between zones. Glare depends on direction, screens, windows, and time of day. Noise changes through the day. Ventilation may be working, partially working, or designed for a different occupancy than the one currently using the room.

The US Environmental Protection Agency's Building Assessment Survey and Evaluation study is interesting here because it did not treat indoor air quality as one isolated measurement. The study collected three kinds of information together: environmental and comfort measurements, building and HVAC characteristics, and occupant symptoms and perceptions.

That combination matters.

The building has a state.

The person has an experience.

The useful system connects the two.

That is already close to the logic behind The Product Needs a Harness. In software, a harness makes an important system more observable and testable. It helps us see whether the thing still behaves as intended instead of assuming that a successful build means a successful product.

A building has the same problem at a different scale.

The lights are on. The air conditioner is running. The room is technically open. None of those facts tells us whether the environment is actually working for the people inside it.

The building needs a harness

I would describe a building harness as three connected layers.

Building state: CO₂, particulate matter, humidity, temperature, light, noise, ventilation, occupancy, and change over time.

Human experience: comfort, fatigue, concentration, irritation, perceived control, and self-reported problems.

Interpretation: correlate, explain, recommend, verify.

The point is not to turn every room into a medical device or every employee into a quantified-self project.

The point is to make an otherwise opaque environment legible enough that people can understand what is happening and decide what to do next.

A useful interface might say:

This space is dry. Noise has been elevated for 47 minutes. Temperature is stable. CO₂ is rising. The ventilation data is unavailable.

That last sentence matters as much as the measurements. A trustworthy environmental interface should distinguish what is measured, what is estimated, and what is unknown.

A phone can contribute some of this context. It can provide proxies for sound and light, location, duration, motion, and self-reported experience. But a phone does not directly measure CO₂, PM2.5, VOCs, or ventilation rate. Those require external sensors, building systems, or other trusted data.

The interface should not pretend otherwise.

The workaround is not a personal failure

There is a wider systems pattern hiding inside these desk hacks.

Shared infrastructure is increasingly being supplemented by personal infrastructure.

The office provides lighting, but the worker brings a parasol.

The office provides ventilation, but the worker brings a purifier.

The office provides climate control, but the worker brings a fan.

The office provides a shared acoustic environment, but the worker wears active noise cancellation for eight hours.

The institution provides the workplace. The individual supplies the missing protective layer.

I think of this as the individualisation of environmental infrastructure.

It resembles a pattern I keep noticing in digital systems. When the shared system cannot handle variation, uncertainty, or edge cases, the burden moves to the individual. People build spreadsheets beside the official software. They copy data between tools. They create private reminders because the workflow cannot be trusted. They keep screenshots because the system has no usable memory.

Physical environments produce their own versions of the same thing.

The personal workaround can be clever and effective. But if everyone has to construct a private micro-environment in order to tolerate the shared one, that is information about the shared system.

Sustainability needs a human constraint

This also connects to a problem in sustainable design.

Energy efficiency is important, but a building cannot be evaluated only by how little energy it consumes.

The EPA's office guidance describes indoor air quality as an interaction among pollutant sources, ventilation, moisture, humidity, building operation, occupant perceptions, and other comfort factors. It also explicitly lists temperature, air movement, sunlight glare, ceiling-light glare, noise, and vibration among factors that affect comfort and productivity.

So the optimization target cannot simply be:

use less energy

It has to be closer to:

reduce environmental cost while maintaining conditions in which people can actually live and work

That is a multi-objective design problem.

A system can be efficient and still transfer its costs somewhere else. It can save energy while producing discomfort. It can reduce centralized services while expecting every person to buy a private device. It can hit a building-performance target while making the occupants responsible for compensating for the result.

This is one reason I keep returning to Human-Centered Design Is Not Enough. Environmental responsibility cannot mean optimizing the building while treating the human experience as an externality. But it also cannot mean maximizing immediate comfort while ignoring energy, materials, climate, and the wider environment.

The interesting design work is in the relationship.

Measure the environment instead of reasoning from vibes

There is another reason the SCMP story is useful.

It shows what happens when an environment is hard to inspect: people fill the gap with stories.

Some of those stories will be correct. Some will be partially correct. Some will confuse correlation with cause. Some will spread because they are vivid enough to photograph.

The sunscreen example is a good case. The FDA notes that typical compact fluorescent lamps emit very low levels of ultraviolet radiation. That does not mean every lighting environment is comfortable, or that glare and visual fatigue are imaginary. It means we should be careful not to turn discomfort into an unsupported explanation.

The better response is not to mock the person under the parasol.

It is to make the environment easier to inspect.

If the complaint is glare, measure light and direction.

If the complaint is dryness, measure humidity.

If the room feels stuffy, inspect ventilation and CO₂ as useful context.

If people report headaches or fatigue, do not assume a single environmental cause. The EPA itself notes that symptoms such as headaches can have causes unrelated to the building.

Good environmental UX should reduce uncertainty without manufacturing certainty.

From the screen to the room around the screen

For years, interaction design has treated the screen as the obvious site of intervention.

But the person using the screen is also sitting in a chair, breathing air, hearing equipment, receiving light, moving through a building, and depending on infrastructure that usually reveals very little about itself.

This is why I think Human Limits should extend beyond software ergonomics.

The body is already part of the interface.

The room is already part of the interface.

The building is already making decisions about what people can perceive, tolerate, control, and understand.

The only question is whether those decisions remain hidden.

A Good Building Can Still Make a Bad City argues that a building cannot be understood only as an object inside its property boundary. The same expansion can happen inward. A building cannot be understood only from drawings, systems, energy metrics, and equipment schedules either. Its performance also exists in the lived relationship between those systems and the people inside them.

That relationship needs an interface.

Not another dashboard full of numbers.

A harness.

Something that senses what can be sensed, admits what cannot, connects system state to human experience, and helps people act before the workaround becomes permanent infrastructure.

Important systems should expose their actual state to the people inside them.

That principle applies to software.

It applies to organizations.

And increasingly, I think it applies to the room around the screen.

AI authorship note

This essay began with Kris Haamer connecting a September 2026 South China Morning Post story about office workarounds to HAAM's work on product harnesses, human limits, environmental sensing, and life-centered design. AI supported source verification, drafting, editing, and implementation. Kris directed the argument and asked for it to be published on HAAM.