# Reframe, or Be Reframed **Author:** Sushee Nzeutem, SVRNOS **Published:** June 16, 2026 **Canonical URL:** https://svrnos.com/insights/reframe-or-be-reframed **Hero image:** https://svrnos.com/insights/reframe-or-be-reframed/hero.png > The skill is what survives after reality answers. --- ![An intricate hand-drawn system on a desk, struck out by a single decisive red line. The discipline this essay is about: removing a frame you built once reality rejects it.](/insights/reframe-or-be-reframed/hero.png) Everyone keeps saying the AI era belongs to people with taste. I think that's too soft. Taste works on things that already exist. It tells you one version is stronger than another. It helps you choose, edit, reject, and polish. That matters. The deeper advantage sits one step upstream: seeing a mechanism from one domain do another domain's job, and building the frame before there's anything to judge. I call it reframing. A web error code becomes a vocabulary for governance failure. A poker term becomes a way to name late-turn AI risk. A set of ideas from endurance physiology becomes a model for how people behave under pressure. The mechanism already works somewhere. The reframe gives it a new job. AI can propose frames too, sometimes good ones. So the human edge can't just be having the analogy first. The edge is what happens next: putting the frame in front of reality, absorbing the result without explaining it away, and reshaping or removing the thing you were proud to build. That discipline has four moves. Frame, Field, Feel, Forge. The frame is only the beginning. The skill is what survives after reality answers. ## Three kinds of knowledge work I find it useful to separate knowledge work into three modes, by how the answer gets made. Retrieval: the answer already exists, and the work is finding it, checking it, presenting it. What did this regulation require? How do I sort a list? AI is excellent here. Assembly: the parts exist but the finished thing doesn't, and the work is combining them. Draft a privacy policy. Generate a support workflow. AI is fast here, already close to a junior specialist. Reframing: the problem itself changes, because a mechanism from one domain is given a new function in another. Calligraphy doing typography. HTTP error codes doing AI governance. Retrieval and assembly describe how the work gets produced. Reframing changes what work is being produced. AI can propose frames, but a proposed frame is not yet a contribution. It becomes one only after someone exposes it to users, constraints, and consequences, and it survives. AI can assist with that test. It cannot bear the consequence or hold the authority to decide what survives. ## The 4F Loop Most people treat reframing as a single moment. The insight. The metaphor. The leap. That is too early to celebrate. A frame that never reaches users is decoration. A frame that reaches users and survives only because its maker explains away every objection is worse. The full loop: - **Frame.** See an existing mechanism doing a non-default job. The output is a new category, vocabulary, system, or method. - **Field.** Put it in front of real users in real conditions, not a friendly audience that already shares your references. - **Feel.** Absorb how it lands before pride turns the signal into a messaging problem. Feel does not mean sentiment. It means receiving the result honestly. - **Forge.** Reshape it, park it, or kill it. Forge can be a smaller next version. Sometimes the right move is no next version at all. ![The 4F Loop: Frame, Field, Feel, Forge, the cycle that turns a reframe into something that survives. Creativity under accountable contact with reality.](/insights/reframe-or-be-reframed/4f-loop.png) This is creativity under accountable contact with reality. AI can support every stage of the loop. It cannot own the consequences of the call. ## My best reframe failed In 2025 I designed a behavioral identity assessment called SIM95. The premise: identity shows itself under pressure. What you do when stress stops being theoretical is structurally different from what you say about yourself when calm. I needed language that carried capacity, recovery, threshold, collapse, pacing, and composure without borrowing clinical authority. So I took the structure from endurance sports physiology. Bonking, the sudden glycogen collapse cyclists hit when the fuel runs out, became the line where pressure stops being manageable. Heart rate variability became flexibility under building load. Resting heart rate became Body OS, how the body carries strain before the mind admits it. VO₂max became a Stability Index. Functional Threshold Power became a Momentum Window, the throughput someone can hold without collapse. Race IQ became cognitive rhythm under unclear stakes. Composure stayed composure. Seven concepts, one source domain, one coordinated system. Each section said plainly that the term was a metaphor, not a medical metric. I was proud of it. It had compression, elegance, structure. It was a real reframe. It failed in production. Beta users found the metaphors more confusing than helpful. The compression that made the report feel intuitive to me created an education burden I had not built. The vocabulary was elegant because I knew both domains. Users did not. I pulled the entire metaphor layer. Not one term. Not one section. The whole thing. ## Why I removed it Given cheap generation, the easiest response would have been to add. A glossary. Tooltips. Onboarding cards. An FAQ. A definition before every section. A short course explaining the system. Heavier, in the name of clearer. More output. More compliance. More slop. The metaphors did not need more explanation. The product was carrying an education layer it did not need to carry yet. The move was subtraction. AI had no sunk cost in that system, no embarrassment, no responsibility for how it landed. I had all three. The real test was whether I could keep those stakes from corrupting the call, whether I could remove something clever and coherent and mine because reality had voted against it. AI makes another version cheap. Judgment decides whether there should be one. ## The pattern shows up across my work Once I saw it, I found it in more than one place. **HTTP status codes became a governance taxonomy.** The [Governance Error Register](https://svrnos.com/research/governance-error-register) borrows the structure of HTTP status codes and points it at AI governance failures. Everyone knows a 404 means something expected was not found, so GER-404 became Governance Rule Not Found. HTTP 501 means not implemented, so GER-501 became Escalation Not Implemented: detection fires, but nothing downstream handles it. The web already trained millions of people to read failure states through short codes. The reframe gives that habit a new job. **Poker became a language for late-turn AI risk.** In Texas Hold'em, the river is the last community card, the moment hidden strategy becomes visible and commitment can no longer be deferred. In multi-turn safety evaluation, the river became the late point where a harmful trajectory finally resolves. The word carries its baggage with it: patience, incomplete information, late reveal, irreversible consequence. Same move, different domain each time: see the structure, borrow the mechanism, change the job. ## Where the advantage sits Refusing AI makes you slow. Tool fluency alone won't stay scarce either. The advantage sits with the person who frames the problem at the level where context, consequences, and accountability matter, fields the frame against reality, owns the judgment of whether it worked, and uses AI to accelerate everything downstream of that. A frame that survives contact with reality can become a product, a method, or a standard. The edge in SIM95 was never inventing the endurance-sports frame. A model could have suggested bonking and threshold power. The edge was putting the system in front of people, seeing that its elegance cost them comprehension, and removing it before pride turned confusion into a feature. Are you using AI to accelerate your frame, or becoming the operator inside someone else's? Framing is the part people celebrate. Removing your own frame when reality rejects it is the rare part, and the part no model does for you. A model can recommend subtraction. It cannot make the sacrifice or own the decision. Reframe, or be reframed. Just don't confuse inventing the frame with earning the right to keep it.