Arrival Had the Right Idea

In Arrival, alien visitors communicate through intricate circles of ink. Humans must learn an unfamiliar visual language before they can understand what the visitors are trying to communicate. The film takes that idea into science fiction, but the question it raises belongs here on Earth: could a different language help us understand things our current ways of communicating make difficult to see?

I think we need to pursue that question seriously.

We explain systems through sentences, documents, diagrams, equations, and code. Each has value. Each captures part of the picture. Connecting those parts into a working understanding is largely left to the person reading them.

Meanwhile, everything we are trying to understand keeps interacting and changing.

We need a universal way to define and follow systems through space and time. Our survival as a species may depend on how quickly we develop it.

Software Is One Participant

Code is a powerful implementation medium and a lousy place to make humans reconstruct an entire system.

We distribute instructions across files and ask people to recover the purpose, responsibilities, relationships, and consequences. AI can generate those instructions faster than ever. That makes the need for a better level of understanding more urgent.

But software is only one participant.

A delivery involves a driver, a vehicle, roads, fuel, warehouses, weather, regulations, software, and someone waiting for a package. All of them affect the outcome. A perfectly functioning application does not guarantee that the package arrives.

Reality contains all kinds of processors. People interpret and decide. Machines move and manufacture. Organisms transform energy and materials. Institutions establish rules. Software processes information and coordinates activity.

These participants operate together whether our models connect them or not.

We need a language capable of representing their interactions together.

Define the Whole, Follow the Connections

A universal systems language should let us describe participants, their behavior, their constraints, and their relationships.

What does each participant do? What does it require? What does it exchange? Where does it operate? How long does its work take? What happens when it fails, changes, or encounters something unexpected?

Those questions apply across disciplines. Their answers need a common foundation.

Medical models would retain medical detail. Engineering models would retain engineering detail. A person’s judgment would remain different from a machine’s programmed response. But we should be able to connect those representations and follow meaning across their boundaries.

Universal does not require one enormous picture containing everything. We need connected views at useful scales. Start with a patient, follow a dependency into a hospital, then into a supplier, a transportation network, or an electrical grid.

We should be able to move through those relationships without rebuilding our understanding from scratch at every step.

The language must also distinguish what we intend, what we assume, and what we observe. Otherwise, we risk confusing a well-organized description with reality.

Walk Through a Hospital

Imagine a hospital operations manager asking why patients are waiting hours for beds while rooms appear to be empty.

In a hypothetical interactive model, she asks the AI, “What is keeping patients in the emergency department today?”

The model connects patient arrivals, room readiness, staffing, equipment, cleaning, and discharge processes. It shows where activities happen and what depends on what.

Some empty rooms need cleaning. Others lack the nursing coverage or equipment required for the patients waiting.

The manager follows the sequence from a discharge decision to a room becoming available. It crosses a physician’s judgment, a transport request, a patient’s physical movement, a cleaning task, and an update in the scheduling software.

The patient experiences one system. The model lets the staff examine it that way.

The manager asks, “What would happen if we moved a cleaning team to this floor for two hours?”

The environment opens a simulated scenario. More rooms could become ready, but nursing coverage would then limit admissions. Travel time between floors also affects how much the cleaning team can accomplish.

She can inspect those assumptions and compare alternatives.

Then a charge nurse spots a problem: the model treats certain beds as interchangeable. Some patients need isolation rooms or specialized equipment. The team adds those constraints and runs the scenario again.

That is useful progress. People with different knowledge can see how their work connects, challenge the representation, and improve it together.

Give AI Something Better to Work With

AI could help us construct these models, connect existing information, expose contradictions, and explore consequences.

A person should be able to ask a question in ordinary language and follow the answer into the relevant relationships and evidence. An explanation should lead somewhere inspectable.

AI could also help translate agreed definitions into action. For software, that might mean implementation. For a physical operation, it might mean revised procedures, staffing, equipment, or layout.

The connection must work both ways. Evidence from the operating system needs to come back into the model. We need to see where actual behavior differs from what we expected.

Otherwise, we have created another document that begins going stale the day it is finished.

The goal is a representation we can keep questioning as the world changes.

Make Uncertainty Visible

A convincing model can still be wrong.

AI can explain an incomplete model beautifully. A simulation can produce precise numbers from bad assumptions. An immersive display can make weak evidence feel substantial.

We need to know which relationships are observed, which are inferred, and which are proposed. We need to see where information is old, where people disagree, and what the model leaves out.

In the hospital example, the nurse’s correction improves the model because the assumptions are available for inspection. That opportunity must be built into the experience.

Understanding includes knowing where our understanding ends.

Teach Children to See the Connections

If we want people to understand systems differently, we need to begin while they are learning how to understand the world.

Children can start with things they know. Where does lunch come from? What does a tree need? How does water reach the school? What happens to a neighborhood when a road closes?

Each question opens a system. There are participants, relationships, resources, boundaries, and changes over time.

Take a school lunch. Children could build a visual model connecting a farm, water, soil, sunlight, workers, transportation, refrigeration, a kitchen, and themselves. They could follow the food, identify what each participant contributes, and explain what happens when something changes.

What if the truck is late? What if the refrigerator stops working? What if there is less rain?

An interactive model could let them explore those questions. AI could help them investigate unfamiliar parts, compare explanations, and find evidence. Teachers would help them examine whether the answers make sense.

The children should also be able to correct the model. Someone notices that it leaves out the people preparing the food. Someone else adds food waste and asks where it goes. The representation improves as their understanding grows.

That is a habit worth teaching: explain your thinking, make your assumptions visible, listen to another perspective, and revise what you believe when you discover something missing.

Reading, writing, mathematics, and scientific knowledge give children essential foundations. A shared systems language could help them connect those foundations across subjects and communicate what they discover.

We should teach them to move between a part and a whole, to distinguish an observation from a prediction, and to recognize that an action can have several consequences. They should learn to ask who benefits, who bears the cost, and what lies outside the boundary they have drawn.

If the next generation is going to inherit responsibility for increasingly powerful and interconnected systems, this needs to become part of its basic education.

We cannot wait until people are responsible for a hospital, an energy network, or a government to teach them how to follow the connections.

Follow the Emerging Tools

This is the territory we’re tracking at diagramming.org.

The directory and newsletter follow tools across human systems, robotics, and software, including digital twins, knowledge graphs, and spatial and immersive visualization. They provide a place to examine the evolving state of the art through documented capabilities and linked sources.

These tools offer parts of the larger capability we need. Knowledge graphs expose relationships. Simulations let us investigate behavior. Spatial environments help us navigate arrangements and scale. AI can help connect those representations and make them accessible through questions.

The next step is making those capabilities work together around a common language.

We should measure progress by what people can understand and do: trace a consequence, identify a constraint, challenge an assumption, or coordinate a decision across disciplines.

Our Ability to Understand Must Catch Up

Climate, food production, energy, public health, economies, and technology already interact at a planetary scale. Our professional languages, institutional boundaries, and disconnected tools make those interactions harder to follow.

Better models will not eliminate conflicting interests or guarantee sound judgment. They can make consequences visible enough to examine and decisions clear enough to challenge.

That is a capacity we urgently need.

Arrival imagined a language that opened a different way of understanding. We need to build our own advance: a universal systems language that helps people define what they mean, connect what they know, and follow what happens.

We are increasing our power to change the world. Our ability to understand those changes must catch up.

For the survival of our species, we cannot afford to keep widening that gap.