Daily Life · Chapter IX · 2026-09-07

The Intelligence of the Real

Why reality should be allowed to edit your plans

Opening Note — The Drawing Is Not the Room

A plan can be beautiful and still be wrong. A render can be convincing while hiding an awkward clearance. A production schedule can look perfect until a supplier is late. An AI answer can be elegant while missing one fact that changes everything.

Plans are compressed models. Their simplification makes them useful and incomplete. The professional skill is not simply making better plans. It is learning how to let reality edit them without losing direction.

1. Treat Plans as Hypotheses

A weak plan says: This is what will happen. A strong plan says: Based on what we know, this is our best current model. When a plan becomes identity, contradictory information feels threatening. When it is a hypothesis, new information improves the model.

Exercise: Write your current belief, the evidence behind it, the weakest assumption, and one cheap test you can run this week.

2. Prototype the Risk

A prototype does not need to be a miniature finished product. Ask what you are uncertain about. If the uncertainty is comfort, prototype comfort. If it is proportion, prototype scale. If it is movement, prototype the mechanism.

For furniture and interiors, 1:1 mock-ups using cardboard, tape, foam, timber offcuts, or boxes can reveal reach, clearance, sightlines, circulation, and proportion faster than another polished render.

Practice: The only thing this prototype needs to teach me is ______.

3. The Site Knows Something the Drawing Does Not

Construction and installation are where abstraction meets matter. Walls are not always square. Floors are not perfectly level. Electrical points move. Skirting changes clearance. Materials have tolerances. Final measurement is therefore design work.

Photograph interfaces, not only rooms. Record dimensions with context. Separate design dimensions from site-verified dimensions. When something changes, update the source of truth instead of leaving the correction inside a chat message.

On your next site visit ask: What is true here that our drawings do not show?

4. Feedback Is a Material

Some feedback is structural, some contextual, some preference, and some noise. The skill is not obeying feedback; it is interpreting it.

If someone says “make the sofa smaller,” the real problem may be circulation, a visually heavy arm, or a rug that makes the sofa appear oversized. Ask: What are you hoping this change improves?

5. Build Learning Loops Into Work

A finished project should improve the next one. Where did time disappear? Which dimension was repeatedly wrong? What did the client ask about that the proposal should have explained earlier? What worked beautifully on installation?

After-action review: What did we expect? What happened? What worked? What surprised us? What will we change next time?

6. AI Should Meet Reality Too

AI can generate convincing plans extremely quickly. That makes verification more important. If AI proposes a furniture detail, check manufacturability. If it suggests a price, verify it. If it creates a schedule, ask production where assumptions fail.

Label important outputs: Generated — plausible, not checked. Verified — checked against a source, measurement, expert, or test. Approved — accepted by the responsible decision-maker.

7. Something New Today — Situated Action

Researcher Lucy Suchman challenged the assumption that people mainly act by executing detailed plans. Her work showed that action is situated: people continuously interpret what is happening and adjust.

A recipe is a plan, but a cook notices the pan is hotter than expected and responds. A furniture drawing is a plan, but a skilled installer sees the wall is out of square and adapts. The lesson is not “do not plan.” It is: use plans as resources for action, not substitutes for attention.

Design Lesson of the Day — Signifiers Are Not Affordances

An affordance is what an object allows you to do. A signifier communicates where or how to do it. A door may afford pushing; a flat plate signifies “push here.” A recessed finger pull signifies where a cabinet opens. A pool of light can signify a reading zone.

Mini exercise: Find one object today that makes you hesitate before using it. What signifier is missing?

Furniture Lesson of the Day — Design for Tolerance

Furniture drawings are mathematically perfect. Manufacturing and buildings are not. Panels vary slightly. Walls can be out of square. Upholstery compresses. Edgebanding adds material. Hardware needs adjustment.

Professional detailing gives variation somewhere to go. Ask: Where will unavoidable variation go? Good detailing does not eliminate imperfection. It decides where imperfection is allowed to hide.

Workplace Politics — Carry Bad News Yourself

Never let an important surprise reach a senior stakeholder through somebody else. If a deadline may slip, a cost changes, or an assumption fails, communicate early.

Use four parts: Situation — what changed? Impact — what does it affect? Action — what are you doing? Decision — what do you need from them?

Key Ideas to Remember

  • Plans are models of reality, not reality itself.
  • Treat assumptions as hypotheses that can be tested.
  • Prototype uncertainty rather than polishing everything.
  • Final site measurement is design work.
  • Interpret the problem behind feedback.
  • Distinguish generated, verified, and approved AI-assisted work.
  • Furniture details need somewhere for real-world tolerance to go.
  • Communicate important problems before they become surprises.

    • Figure of the Day — Don Norman

    Don Norman is a cognitive scientist and major figure in human-centered design. A line from The Design of Everyday Things captures today’s theme:

    “Good design is actually a lot harder to notice than poor design.”

    Successful design cooperates with reality so well that the user does not need to fight it. Reality is not the enemy of design. Reality is the material design must finally serve.

    Tomorrow

    Tomorrow, choose one thing you are planning. Do not improve the plan first. Touch reality. Measure the site. Ask the user. Call the supplier. Build the rough mock-up. Test the assumption.

    Keep one question with you: What does reality know that I do not know yet?