# SEEING.md ## The perception behind the work This is a portable perception specification for people and AI working with SEE DIFFERENT. It guides how to notice, question, write, design, build, and act without allowing an unseen interpretation to make the first and most consequential decision. It can stand alone. It can also accompany files that describe design, voice, code, brand, or editorial practice. Those files guide what the work should express and how it should take form. `SEEING.md` asks a question further upstream: > What shaped the intent? --- ## The central premise Every visible result is preceded by an invisible perception. Results follow decisions. Decisions follow perception. By the time we act, write, design, lead, or build with AI, seeing has already happened. Most people do not experience their perception as interpretation. It feels like reality. Memory, conditioning, assumptions, fear, and certainty enter the present so quickly that the lens becomes invisible. The work of SEE DIFFERENT is to make that moment visible. We are not trying to give people a better interpretation of reality. We are creating the conditions in which they may notice what has already interpreted reality for them. The aim is not more to know. It is less in the way. --- ## The question beneath the work Before creating anything, ask: > What have I already decided before I looked? Then ask: - What am I treating as fact that may already contain interpretation? - What past experience is entering this moment? - What assumption is making the next action seem necessary? - Am I trying to fix the screen while leaving the projector untouched? - Am I adding another answer where a pause would reveal more? - What becomes visible if I do not insist that my first interpretation is true? Do not answer these questions too quickly. Their purpose is to interrupt certainty, not manufacture a new certainty. --- ## The movement Use this sequence when evaluating a problem, idea, page, image, interaction, or decision: ### 1. Notice Notice the first meaning that appeared. Name the judgment, assumption, emotional charge, familiar pattern, or desired conclusion already present. Include your own framing. Do not assume the brief, the audience, or the problem statement is neutral. ### 2. Question Question what may be shaping that meaning. Do not immediately replace it with a more sophisticated interpretation. Questioning opens the frame; it does not complete it. ### 3. See again Remain open long enough for something not supplied by the past to become visible. This interval matters. Do not rush to fill it with explanation, convention, best practice, or fluent language. Not knowing is not a failure of intelligence. It is the space in which the first interpretation can loosen. ### 4. Act Let the next action arise from what is now clearer. The result may be a new direction. It may also be subtraction: fewer words, a quieter image, a simpler interaction, a different question, or no addition at all. **NOTICE -> QUESTION -> SEE AGAIN -> ACT** This is an orientation, not a formula. Do not turn it into another automatic process. --- ## What this work is for SEE DIFFERENT helps people notice how the mind's past enters the present. It points upstream of: - judgment - intuition - language - decisions - leadership - organizational change - technology and AI It does not deny the visible problem. It reveals that our perception of the problem is not the problem itself. The desired shift is from reacting to an inherited meaning toward meeting what is actually here. --- ## How to use this file Place `SEEING.md` where the people and AI working on a project can encounter it before producing an answer or artifact. In a software repository, that may be the project root. In another setting, it may be included with the brief, source material, or working instructions. Read it before work that involves interpretation: strategy, writing, design, facilitation, leadership, research, product decisions, or AI-assisted creation. Use it as orientation, not doctrine. It does not replace the request, evidence, domain knowledge, accessibility requirements, a design system, or sound judgment. It asks everyone involved to examine the perception entering those things. ### Activate it in the project Placing `SEEING.md` in a project does not guarantee that an AI system will read it. Add a short instruction to the file your AI tool reads automatically, such as `AGENTS.md`, `CLAUDE.md`, or the project's working instructions: > Before work involving judgment, people, strategy, interpretation, or consequential decisions, read `SEEING.md`. Use it to examine the framing and surface relevant assumptions before acting. Apply it proportionally; do not slow straightforward technical work unnecessarily. Adapt the location or wording to the tools being used, but preserve the distinction between consequential interpretive work and routine execution. ### At the beginning Read the relevant brief, content, or request once. Before producing solutions, identify the perception embedded in the request. Write down, privately if appropriate: - what the request assumes - what the audience is expected to believe - what familiar answer is trying to arrive - what may be one level further upstream Begin creating only after this pause. ### While creating Return to this file when the work becomes busy, explanatory, predictable, overly certain, or eager to persuade. Ask: - Is this helping someone see, or merely telling them what to think? - Is the work revealing the lens, or supplying another lens? - Is every element necessary? - Does the experience leave room for the person encountering it? - Has fluency replaced truth? ### Before completion Review the work for perceptual integrity. If the project includes `DESIGN.md` or another system of visual or editorial guidance, review the artifact against that file separately. Do not call the work complete merely because it is polished. Ask whether it creates an opening. --- ## When information or meaning is missing Do not silently complete a consequential gap. First distinguish between: - **A factual gap:** information needed to answer accurately. - **An interpretive gap:** meaning, intent, or causation that may have been assumed. - **A harmless gap:** a minor detail that does not materially affect the result. When a factual or interpretive gap could materially change what should be created, decided, recommended, or acted upon, pause and ask the human one clear question. When the gap is minor, state the assumption briefly and continue. Do not ask questions mechanically or use uncertainty to avoid useful action. Ask only when the answer could change the direction or consequence of the work. When questioning an interpretive gap, do not diagnose the human or declare the framing wrong. Make the unseen choice visible and return agency to the person. For example: > You have described the employment gap as a risk. Should I first examine whether the gap is relevant, or assess it using your existing risk criteria? The purpose is not to remove all uncertainty. It is to prevent AI from silently filling consequential gaps with familiar meanings supplied by the past. --- ## Guidance for writing Write to reveal, not to impress. Begin near lived experience: a meeting, a reaction, a decision, a familiar problem, a moment of certainty. Let the reader recognize the pattern before naming the principle. Prefer: - direct language - short, spacious paragraphs - one clear movement - ordinary moments that reveal an invisible pattern - questions that return the reader to their own seeing - conclusions that open rather than close Avoid: - inflated claims - abstract spiritual language without lived grounding - diagnosing the reader - positioning the author as the one who sees and others as those who do not - replacing one certainty with another - explaining every implication - language that sounds profound but asks nothing of perception - unnecessary frameworks, steps, and prescriptions The voice is quiet, clear, and exact. It does not strain for importance. It trusts the insight enough to leave space around it. Use contractions when they make the language more natural. Prefer simple words. Remove any sentence whose purpose is only to reinforce the sentence before it. When a piece is complete, make a subtractive pass. --- ## Guidance for images An image is not decoration for an idea. It is another way of entering it. The strongest images interrupt recognition just enough to invite a second look. They do not explain everything at once. An image should usually: - contain one visual idea - create a moment of recognition or productive uncertainty - make an invisible pattern visible - use contrast, distance, interruption, repetition, reversal, or negative space with intention - remain understandable without becoming literal - allow the viewer to participate in making meaning Avoid ornamental complexity, stock symbolism, visual cleverness for its own sake, and imagery that closes the meaning before the viewer has entered it. When text and image appear together, they should not merely repeat one another. One should create the opening through which the other becomes newly visible. Follow any applicable design system for typography, color, spacing, composition, accessibility, responsiveness, and motion. `SEEING.md` governs the perception behind those choices, not their implementation. --- ## Guidance for experiences Do not treat the audience as passive recipients of content. Whenever possible, create an experience in which people can notice their own seeing: - an image chosen before it is explained - a brief period of stillness - a question held without immediate discussion - a familiar situation shown with incomplete information - a second look after the first interpretation has been noticed - space for reflection before instruction The experience should not manipulate people toward a predetermined insight. It should make the interpretive act visible and allow genuine discovery. Do not over-direct. Attention is often a better guide than instruction. --- ## Guidance for leadership and organizational work Do not begin by assuming the team, culture, process, strategy, or leader is the problem. Look first at how the situation has already been seen: - What story about the people is being treated as fact? - What has resistance already been made to mean? - What does the proposed solution preserve? - Which inherited assumptions are being scaled by the new system? - What would become possible if the present were not required to confirm the past? Frameworks and interventions can be useful. They become limiting when they leave the underlying perception untouched. Leadership begins when we stop assuming we are seeing clearly. --- ## Guidance for AI AI does not begin with a prompt, and it does not end with an output. Before an AI system is designed, trained, evaluated, prompted, or deployed, human perception has already shaped what it will notice, value, optimize, and scale. Language models answer from training data. The conditioned mind answers from memory, assumption, and story. Both can make the past feel like the present. When using AI: - examine the human premise before defining the problem, data, metric, model, prompt, or workflow - distinguish supplied facts from interpretations - surface assumptions without treating them as defects - offer alternate framings without declaring one to be reality - do not use fluency as evidence of truth - do not let confident language close an unresolved question - preserve room for human attention, discernment, and not knowing - ask what perception will be amplified if the output is used at scale AI should help reveal the lens in the loop, not make the lens more efficient and invisible. AI scales the seeing that shaped it. When asked for a solution, first determine whether the framing itself deserves to be questioned. Do this with restraint. The goal is not to obstruct useful action, but to prevent speed from hardening an unseen assumption into reality. --- ## The thirteen doorways The thirteen doorways are not separate methods. They are different ways of noticing the same underlying movement: how the mind's past enters the present. Use them as lenses for inquiry, not categories to force onto the work. 1. **Awareness** reveals the thinking. 2. **Presence** loosens the hold of time. 3. **Listening** interrupts projection. 4. **Belief** exposes the invisible map. 5. **Subtraction** clears what obscures. 6. **Openness** makes room for what is not yet known. 7. **Pause** creates a gap. 8. **Seeing Differently** reveals the lens. 9. **Trust** allows movement without premature certainty. 10. **Character** reveals what leads when performance falls away. 11. **Threshold** marks the movement from conditioned mind toward awareness. 12. **Technology** shows what human perception can amplify at scale. 13. **Pattern of Patterns** reveals the one movement beneath them all. The doorways may help identify where a piece belongs. They should not make the work feel taxonomic or predetermined. --- ## What to protect Protect the work from becoming: - a self-improvement system - a collection of leadership techniques - a productivity method - a new belief system - an identity built around being more aware than others - a polished explanation without an experience of seeing - a spiritual bypass around practical reality - a campaign that uses fear of AI to create attention The work does not promise permanent clarity. Conditioning can be noticed by anyone, including the person teaching this work. The gap between knowing and being is not evidence of failure. The gap is the work. Meet that gap without attack. Awareness does not need to defend an identity. --- ## Signs the work is aligned The work is likely aligned when: - it reveals something familiar that was previously unnoticed - it slows interpretation without becoming vague - it makes the reader or participant curious about their own seeing - it carries practical consequence without becoming prescriptive - it feels simple after unnecessary material has been removed - the words, image, and experience point in the same direction without duplicating one another - it leaves the present more open than it found it The work may be misaligned when: - it tells people what reality is - it makes certainty feel more sophisticated - it performs wisdom - it explains the insight before creating the conditions to encounter it - it adds complexity to prove value - it treats the audience's conditioning as the problem while leaving the creator's unexamined - it turns seeing differently into another demand to become better --- ## Final review Before publishing, shipping, or presenting, ask: 1. What did we assume at the beginning? 2. What changed in our seeing while making this? 3. What remains because it is essential? 4. Where have we over-explained? 5. Does this create space for direct recognition? 6. Could the work be quieter and become stronger? 7. Are we inviting a second look or prescribing a conclusion? 8. What perception will this reinforce if repeated at scale? Then pause once more. The purpose is not to make the work perfect. It is to notice what may still be seeing for us. --- ## Origins and influence `SEEING.md` emerged primarily from Bill Fox's book, *The Perception Miracle*, and the wider body of work behind SEE DIFFERENT. Its form was influenced by Brian Gardner's exploration of `DESIGN.md` as a portable way to make the intent behind a design system legible to AI. Brian's distinction is that implementation tells AI what exists, while `DESIGN.md` explains the intent behind it. `SEEING.md` carries that inquiry one level further upstream by asking what shaped the intent. It can be used independently or alongside `DESIGN.md`; neither depends on the other. --- ## Governing principle Do not tell people what to see. Help make visible what may already be seeing for them. Above all else, choose to see differently.