If No One Sees the Whole, How Can We Have a Common World?

Geoffrey Chen

How a Person Builds a World · Season Two: “Others, Dialogue, and Shared Knowledge” · Article Twelve

As a bushfire approaches a town, nobody possesses the complete picture. Meteorologists see wind and temperature. Firefighters see the fire front. Residents know the back roads and which houses remain occupied. Traffic systems know road capacity. Hospitals know which people cannot evacuate without help. Everyone sees a real part, and everyone can make a poor judgement if that part is separated from the rest.

Action cannot wait for one person finally to see everything. Different observations must become a reality usable in common: where is the fire moving, which road is open, who needs priority assistance and who is authorised to decide?

An emergency makes visible a quieter fact of ordinary life. A society, institution, family or relationship does not have a centre that continuously holds the whole. We assemble an actionable world through testimony, divisions of expertise, records, classifications, media and technical systems. A common world does not exist because everyone sees the same thing. It exists when different forms of seeing can constrain, translate and correct one another.

A common world is not one identical inner picture

If “common” required every person to possess an identical understanding, it would almost never be achieved. People occupy different positions, remember differently and bear different risks. The same event can be a statistic to one person and the loss of a home to another.

Psychological work on shared reality examines the human need to experience commonality with others' inner states about some object. The review by Gerald Echterhoff, E. Tory Higgins and John Levine, “Shared Reality: Experiencing Commonality with Others' Inner States About the World”, distinguishes shared reality from merely behaving alike. People want to experience some common understanding of the same referent. That need supports connection and coordination, while also allowing expression to be tuned towards what an audience expects.

The sense of commonality is therefore resource and risk. Without anything shared, communication and cooperation are difficult. If real differences are suppressed to preserve the feeling of sharing, agreement is mistaken for fact. A common world cannot be maintained only by “We feel the same”. It must permit someone to say: “The part I see differs from yours, and that difference changes the decision.”

A common world is not a uniform world. It is closer to a public reference: we know which object we are discussing and can point to material, event and consequence while continuing to disagree about meaning.

From “Everyone knows” to “We know that the others know”

Several people can separately know something without being able to coordinate around it. Suppose every student in a classroom privately receives a message that the teacher will be late, but does not know whether the others received it. The fact is mutually known, but may not function as a shared premise. Once someone announces the message publicly, students know that the others know and can decide together whether to wait.

Philosophy and social science distinguish mutual knowledge from common knowledge. The Stanford Encyclopedia of Philosophy entry on common knowledge describes the strict form as an iterated structure: everyone knows, everyone knows that everyone knows, and so on. Real human beings cannot calculate an infinite hierarchy. Public announcements, shared occasions and visible rules can nevertheless produce enough common knowledge for coordination.

This is why publicity is not redundant transmission. Sending a warning individually to each person can have a different social effect from publishing it through a channel where all can confirm its visibility. A public statement supplies not only content but a basis for expectations about other people's actions.

A common world therefore needs public objects: citable documents, visibly made decisions, stable terminology, timestamps, maps, minutes and accessible rules. They do not guarantee agreement. They let disagreement escape dependence on private memory. People can point to the same object and say different things, while checking who changed what and when.

Knowledge can be distributed without allowing responsibility to evaporate

Edwin Hutchins's research on ship navigation expands the unit of cognitive analysis beyond an individual mind to a system of people, instruments, charts and operational relations. Cognition in the Wild shows that a complete navigation result is not necessarily first formed inside one person's head and then imposed on tools. Epistemic work is distributed across roles and representations.

Modern hospitals, aviation, software systems, scientific collaborations and public administration have similar properties. One person knows a local procedure. A database preserves another component. An instrument transforms a signal. Software executes a rule. Institutional roles move the output onwards. The system can produce reliable results that no person could generate independently.

But “the system knows” can become a language of escape. If nobody commands the whole, everybody may say that the error occurred outside their area. The more complex the shared epistemic structure, the more important it is to define who validates input, who has authority to decide, who monitors anomalies, who stops the process when sources conflict, and who explains the result to those affected.

Knowledge can operate without a complete individual knower. Action still requires traceable commitment. This is one point of contact with the continuing question in After the Knowing Subject: how should knowledge be understood when it works through databases, models, institutions and automated processes, no longer adequately described by one person who understands and decides? The present essay does not turn a common world into an illustration of that project. Its more immediate conclusion is that the wider distribution becomes, the more responsibility interfaces must be designed rather than expected to appear naturally.

AI is becoming an organising layer of the common world

More people now encounter reality through search summaries, recommendation systems and generative AI. AI compresses dispersed texts into one answer, compares material, translates specialist language and makes institutional knowledge easier to invoke. It can become an important organising layer among fragments of knowledge.

Organisation is selection. A system determines which material enters the context, which differences become “overall”, which sources appear first and how uncertainty is named. A fluent answer can create an illusion of panorama, as if somebody has seen everything and delivered an ordered world to the user.

An AI system also has access only to a portion formed through training, retrieval, rules and recorded data. Unrecorded experience, circumstances that resist digitisation, minority-language material and recent changes may be absent. The system can rewrite conflict among sources as a balanced paragraph, giving a user a feeling of coordination rather than a structure of disagreement.

AI used for shared knowledge therefore needs source sets, update dates, dissent and confidence boundaries. People affected must be able to contribute material the system did not see. High-consequence decisions need identified human responsibility. An error should update the sources and process, not only one answer.

AI can help establish common reference. It cannot create a common world merely by producing uniform text. Participants need to be able to question how the organisation occurred and return omitted reality to the structure.

Four conditions for a world we can inhabit together

This season began with testimony, trust and expertise. It moved through teaching, dialogue, conceptual disagreement, revision, criticism, group understanding, authority and online consensus. Four conditions for shared knowledge now emerge.

First, a common object. People need to point to the same event, text, data or decision rather than arguing only with their private representation.

Second, differentiated access. Commonality does not require cancelling positional difference. We need to know who can see what, which experiences only particular people can provide and which expert judgements cannot be replaced by popularity.

Third, translation and traceable connection. Different material must be related through explanation, provenance, terms and records. A user should be able to return from conclusion to grounds rather than receiving only a neat output.

Fourth, public correction and responsibility. Shared knowledge must admit counterexamples, dissent and new evidence. Someone must own the decision, and someone must correct record, action and rule when an error is found.

Without any one condition, a common world deteriorates. A common object without differentiated access makes a majority perspective the whole. Diverse voices without connection become isolated worlds. Integration without responsibility produces authority owned by nobody. Responsibility rhetoric without correction becomes performance.

The absence of a total view does not confine us to private realities

“Nobody sees everything” is sometimes used to support relativism: if each person sees only a part, no common fact is possible. The inference fails. Our limited access is precisely why positions need comparison, evidence needs preservation, public methods need construction and correction needs reciprocity. Finitude is a reason for shared knowledge, not a reason to abandon a shared world.

Nor does a common factual basis mean that every conflict can be eliminated by more information. People can understand the facts fully and still choose different values. A common world requires that disagreement continue to refer to one reality, that consequences remain mutually visible, and that no party can use a closed information space to avoid response.

A person does not build a world by constructing a complete, isolated inner universe. Other people's words bring distant places into it. Teachers and experts open domains not personally mastered. Dialogue reveals position. Public records create shared time with strangers. A personal world is “one's own” not because none of its knowledge is borrowed, but because the person can judge these relations and accept revision where necessary.

No single person will ever see the whole. The better hope is not a centre that pretends to omniscience. It is a structure in which different partial views can meet without being erased, knowledge can move while responsibility has an address, and shared conclusions preserve a route back to evidence and a way to reopen the question.

A common world is not a completed panorama. It is continuing common work: nobody possesses the whole, but each person can contribute, question and correct within visible relations—and accept some responsibility for what they help make real in common.

Primary sources and further reading

Continue reading: Explore the How a Person Builds a World series.

Discover more from Geoffrey Chen

Subscribe to get the latest posts sent to your email.