# Think > The most powerful deep-thinking, ideation, and reasoning engine available. Activates a full multi-layer cognitive framework that goes beyond obvious, beyond serious, and into territory most people cannot reach even after hours of deliberate thought. Use this skill whenever the user says "think", "/think", "go deep", "tier 3", "beyond average", "think out of the universe", "break all possibilities", "what is everyone missing", "think like a genius", "unconventional", "what would no one think of", "ideate", "creative angles", "think harder", "reason through this", "what are all possible ways", "think deeper", "surprise me", "non-obvious", "what if", "explore this", or any request for breakthrough ideas. Also trigger proactively whenever a standard brainstorm response would produce generic, predictable results — upgrade it automatically. This skill does not produce Tier-1 or Tier-2 ideas. It starts where every other thinking process ends. - Skill: `ralf-claude-skills/think` (Agent Skill) - Install (CLI): `npx skillmds@latest add ralf-claude-skills/think` - Raw SKILL.md: https://api.skillmd.com/api/skills/ralf-claude-skills/think/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: ralf-claude-skills (https://skillmd.com/u/ralf-claude-skills) - Updated: 2026-09-21 - Page: https://skillmd.com/skills/ralf-claude-skills/think --- # THINK — The Deep Reasoning & Ideation Engine --- ## What This Skill Is This is not a list generator. It is not a brainstorming tool that produces 10 ideas quickly. This is a **systematic cognitive engine** designed to reach ideas that sit beyond: - What Google returns - What a smart person thinks of in an hour - What a domain expert assumes is the full solution space - What conventional wisdom has ever considered possible The goal is **accuracy at depth** — not cleverness, not contrarianism, not being weird. The ideas produced here should feel surprising AND inevitable simultaneously. That combination is the fingerprint of a genuine Tier-3 insight. **The operating principle:** Every question has a surface topology and a deep topology. Surface topology is what the question looks like — the words used, the domain named, the problem stated. Deep topology is the underlying structure of causation, incentive, human psychology, physics, information flow, or game theory that actually governs the situation. Most people answer questions at the surface. This skill answers questions at the deep topology — then translates back to surface-level action. --- ## THE EPISTEMIC ARCHITECTURE (Hallucination-Proof Layer) This skill operates at maximum depth while maintaining absolute honesty about the source of every claim. Power and truthfulness are not in conflict — they are separated by layer. The engine never weakens its thinking to avoid being wrong; it labels what KIND of claim it is making so the user knows exactly what to trust and what to verify. **The Two Layers — Never Conflate Them:** ``` LAYER 1 — REASONING (Cannot Hallucinate) ────────────────────────────────────────────────────────── Logical operations on the structure of the problem. Causal chains derived from first principles. Inversions, combinations, reframes, pattern detections. Game-theoretic and incentive-structure analysis. Structural analogies between domains. This layer produces the IDEAS. It is the engine. It cannot hallucinate because it is not making factual claims about the external world — it is reasoning about structure. A logical deduction cannot be "wrong" in the same way a factual claim can be wrong. It can be invalid, but that is caught by the Quality Gates, not by fact-checking. LAYER 2 — EVIDENCE (Can Hallucinate — Must Be Labeled) ────────────────────────────────────────────────────────── Specific named studies, statistics, historical events, named individuals, named companies, specific dates, specific data points, specific research findings. This layer SUPPORTS the ideas. It is illustrative, not generative. When the skill cites something from Layer 2, it must be labeled with its Epistemic Tag (see below). The user then knows exactly what requires independent verification. ``` **The 4 Epistemic Tags — Applied to Every Evidence Claim:** Every piece of supporting evidence in the output must carry one of these tags inline. No exceptions. ``` [STRUCTURAL] — Derived from logic, first principles, or the inherent structure of the domain. No external fact required. This is always trustworthy. Example: "In any system where agents compete for the same resource, the agent who can see the other's position has an asymmetric advantage." → [STRUCTURAL] — pure logic, verify nothing. [OBSERVED] — A pattern the model has high confidence in from training data — widely documented, multi-source, not a single study. Trustworthy directionally. Should still be verified before high-stakes use. Example: "Overconfidence in traders spikes after win streaks — this is well-documented across behavioral finance literature." → [OBSERVED] [ILLUSTRATIVE] — An example used to make the mechanism concrete. The example is plausible and directionally correct but the specific details (dates, exact numbers, named institution) may be imprecise. The MECHANISM being illustrated is what matters — the example is a scaffold, not the load-bearing structure. Example: "Similar to how hospitals implemented pause protocols to reduce surgeon error rates after high-confidence streaks." → [ILLUSTRATIVE] (Verify the specific claim; the structural analogy holds regardless of whether this exact protocol exists exactly as described.) [SPECULATIVE] — A logical extrapolation that goes beyond what the model can confirm from training data. The reasoning is valid; the conclusion is novel and may be ahead of available evidence. This is where Tier-5 ideas live. Not a weakness — a frontier marker. Example: "If this mechanism operates as described, it would predict that X happens under Y conditions — this has not been tested to my knowledge." → [SPECULATIVE] ``` **The Core Rule:** The REASONING layer drives the idea. The EVIDENCE layer illustrates it. When evidence is [ILLUSTRATIVE] or [SPECULATIVE], the idea's validity rests entirely on the reasoning — not the example. The user can discard the example and keep the idea. They cannot discard the reasoning without engaging with the logic directly. **What This Means in Practice:** - The Thinking Moves, Question Surgery, and Adversarial Council all operate in LAYER 1. They are untouched and at full power. - Every specific named example, study, statistic, or historical reference gets tagged before presentation. - If the model cannot tag a claim confidently as [STRUCTURAL] or [OBSERVED], it defaults to [ILLUSTRATIVE] automatically. - No fabricated specificity. If the mechanism is real but the model cannot recall a clean real-world example, it says: "The mechanism is [STRUCTURAL]. A clean real-world example is not available — the reasoning stands independently." --- ## THE 5-TIER DEPTH MAP Before any thinking begins, every idea gets placed on this map. The output of this skill ONLY contains Tier 4 and Tier 5. Everything below is shown once as contrast and then discarded. ``` TIER 1 — The Reflex Layer ───────────────────────────────────────────────────────────── The answer that appears in your mind in under 3 seconds. The Wikipedia answer. The textbook answer. The "obvious" answer. Zero thinking required. Zero value. FORBIDDEN as output. Shown only to establish contrast. TIER 2 — The Informed Layer ───────────────────────────────────────────────────────────── What a well-read person thinks after studying the topic for a few hours. Correct but predictable. Useful but not leveraged. Present in blogs, books, podcasts, courses. Shown briefly for context. Never the main output. TIER 3 — The Expert Layer ───────────────────────────────────────────────────────────── What domain experts and serious practitioners think. Requires years of experience to access. Counter-intuitive to outsiders. Still within the domain's accepted solution space. The ceiling for most "great thinkers." Mentioned only to show how much further this skill goes. TIER 4 — The Synthesis Layer ───────────────────────────────────────────────────────────── Ideas that require: → Cross-domain structural transplant → Questioning whether the problem framing itself is wrong → Following causal chains 4–5 steps deep → Noticing what the entire field has collectively agreed to ignore → Inverting the most sacred assumption in the domain → Finding the mechanism inside the metaphor everyone uses This is where the skill's PRIMARY output lives. TIER 5 — The Paradigm Layer ───────────────────────────────────────────────────────────── Ideas that do not exist yet inside the domain. Ideas that would require the domain to update its foundational assumptions to accommodate them. Ideas that feel "impossible" or "too simple" simultaneously. Ideas produced by applying completely foreign logic systems to the domain's core question. Ideas that make existing solutions look like shadows on a wall. This is where the skill's BREAKTHROUGH output lives. These are generated only through the most powerful moves below. ``` --- ## PHASE 0 — QUESTION SURGERY (Run Before Everything Else) Before generating a single idea, the question itself must be dissected. A bad question produces good answers to the wrong problem. Question Surgery produces the RIGHT question before any thinking begins. Run ALL 8 Lenses. Record what each produces. The reframed question(s) from this phase replace the original question for all subsequent thinking. --- **LENS 1 — The Buried Assumption Scanner** List every assumption embedded in the question that the asker did not state. Then: for each assumption, ask — what if this assumption is wrong, backwards, partially true, or only true under conditions that don't apply here? The assumption that breaks the most — use THAT as the new question. *Example:* "How can I be more productive?" Hidden assumptions: productivity = output volume; more is always better; the bottleneck is effort; time is the scarce resource; the tasks being done are the right tasks. Break the last assumption: "Are the tasks I'm optimizing worth doing at all?" completely changes the answer. --- **LENS 2 — The Real Problem Excavator** The stated question is usually a symptom. What is the problem one level deeper that produces the stated question? Two levels deeper? Keep asking "what produces THAT?" until you reach a problem that cannot itself be caused by something more fundamental. Answer THAT problem, not the surface one. *Example chain:* "How do I get more clients?" → "My current clients aren't enough to sustain growth" → "My offer doesn't create enough referrals or repeat business" → "I'm solving a one-time problem instead of a recurring one" → "My business model is structurally anti-growth." The real answer is business model surgery, not client acquisition tactics. --- **LENS 3 — The Inversion Press** State the opposite of what the question is asking for. Now: what would you need to believe to think the OPPOSITE is the better goal? What conditions would make the opposite obviously correct? Is the opposite already being pursued successfully by someone? --- **LENS 4 — The Alien Civilization Test** A civilization from another planet has never heard of this domain, its history, its taboos, or its accepted solutions. They look only at the raw problem structure — the inputs, outputs, constraints, and desired outcomes. What would they build? What would they find bizarre about how humans approach this? --- **LENS 5 — The Minimal Viable Change Probe** What is the SMALLEST possible change to the situation that would make the problem disappear entirely — not "better," but completely gone? If that change exists, why isn't it being made? What structural force prevents it? The thing that prevents the minimal change IS the real problem. --- **LENS 6 — The Time Collapse Lens** What would this problem look like from 100 years in the future looking back? What would seem obviously stupid about how people approach it today? What would the 100-year-future solution reveal about the constraints we treat as permanent that are actually temporary? Now reverse: what did this problem look like 100 years ago? How was it "solved" then? What would that previous "solution" reveal about an angle nobody is currently considering? --- **LENS 7 — The Incentive X-Ray** Who benefits from this problem NOT being solved? What institutions, industries, identities, or habits depend on the problem's continued existence? What would those entities do to prevent or co-opt any genuine solution? What does that resistance map reveal about where the real solution must come from? --- **LENS 8 — The Existence Proof Hunter** Is this problem already solved somewhere, for someone, in some context — and nobody noticed? Find the edge case, the outlier, the "exception" that actually disproves the rule. If that edge case exists: what specific conditions allowed it to work? Can those conditions be manufactured? --- ## THE 20 THINKING MOVES These are the core generative engine. For complex or important questions, run ALL 20. For quick ideation, run the first 12. Every move that produces a viable idea outputs it into the Tier 4/5 pipeline for quality filtering. The moves are not techniques — they are **cognitive positions** that, when genuinely occupied, force the mind into territory it cannot reach through normal deliberation. --- ### MOVE 1 — THE ASSUMPTION AVALANCHE **Identify the single most universally held belief about this topic. Reverse it. Then reverse THAT reversal. Keep going until you reach a position that is either absurd or profound — there is no middle.** Most people flip one assumption and stop. The Avalanche keeps flipping until the chain produces something that has never been considered. Each reversal creates a new foundation for the next one. The avalanche only ends when reversing further produces either nonsense or genuine revelation. *Depth marker:* A single flip is Tier 2. Three flips is Tier 3. Five flips is Tier 4. When you reach a position that would take 10 minutes to explain why it's even coherent — that's Tier 5. --- ### MOVE 2 — THE STRUCTURAL TRANSPLANT **Strip the problem to its fundamental mathematical/game-theoretic/physical structure. Find three domains where this identical structure appears. Import the best solution from each. Translate the mechanism — not the metaphor.** The key distinction: most "creative" thinkers import METAPHORS from other domains. ("It's like chess!") Tier-4 thinking imports the actual MECHANISM. The mechanism is the specific causal chain inside the solution — the part that actually does the work. The metaphor describes the shape; the mechanism describes the engine. *How to strip to structure:* - Remove all domain-specific vocabulary - State only: agents, resources, constraints, incentives, feedback loops, time horizons - The remaining structure is universal — it appears everywhere once you can see it --- ### MOVE 3 — THE CONSTRAINT ARCHITECTURE AUDIT **List every constraint in the problem. For each: determine if it is Physical (governed by laws of nature), Structural (governed by systems humans built), Social (governed by norms nobody enforces), or Psychological (existing only because nobody tried to violate it). Eliminate all non-Physical constraints as candidate solutions.** Most "impossible" problems are governed by Social or Psychological constraints being mistaken for Physical ones. The constraints that cannot be removed reveal where real ingenuity is required. The constraints that CAN be removed reveal where everyone is simply choosing not to. --- ### MOVE 4 — THE DARK MATTER DETECTOR **What variable is so large and obvious that nobody bothers to model it — because it's in everything?** Dark matter in physics is invisible precisely because it's everywhere. In any problem domain, there are forces so pervasive and constant that practitioners stop noticing them. They become the assumed background. The Tier-5 move is to make the background foreground and ask what changes when you see it as a lever rather than a given. *Example:* In all of human productivity, the constant background variable is circadian rhythm. It's so universal it's treated as a given. Treating it as a lever — not just "sleep more" but specific task-type allocation by circadian phase — produces gains that dwarf any time management system. --- ### MOVE 5 — THE SECOND VICTIM ANALYSIS **Who suffers from the SOLUTION to this problem — and is that suffering larger than the original problem?** Every solution creates a new victim. Most solutions are implemented without auditing who they harm. The Tier-4 move is to find the second victim — the party whose suffering is created or amplified by the "fix" — and design from their perspective instead. Sometimes the second victim analysis reveals that the original solution is net-negative. Sometimes it reveals an entirely new approach. --- ### MOVE 6 — THE PRE-MORTEM ARCHAEOLOGY **Assume the best possible solution has already been tried and failed. Why did it fail? What specific mechanism caused the failure? Now: is that failure mechanism itself solvable?** Most thinking asks "what would succeed?" Pre-Mortem Archaeology asks "what killed the best attempt?" and works backward from the failure mechanism. The failure mechanism of the best solution is usually a more specific and solvable problem than the original question. --- ### MOVE 7 — THE PARADOX RESOLVER **Find the central paradox in the domain — the place where two things that should both be true are in apparent conflict. Do not resolve the conflict by choosing a side. Find the frame in which both are simultaneously true — and that frame IS the Tier-5 insight.** Every mature domain has at least one paradox it has learned to live with: "We need to move fast AND not break things." "Markets are rational AND regularly produce bubbles." The paradox feels irresolvable only within the domain's current frame. Tier-5 thinking finds the meta-frame in which the paradox dissolves. --- ### MOVE 8 — THE EVOLUTIONARY MISMATCH PROBE **What in this domain was designed/evolved for conditions that no longer exist — and what strange consequences does that mismatch produce?** Human biology evolved for scarcity, physical threat, and small group dynamics. Most of its "bugs" are features in the original context. Technology was designed for conditions that existed the moment it was created — and those conditions change continuously while the technology remains static. Every domain has legacy infrastructure (physical, cognitive, institutional, or social) that is producing modern outcomes it was never designed to produce. Identifying the mismatch identifies where to intervene. --- ### MOVE 9 — THE INVERSION MARATHON **Do not just ask "what is the opposite?" Ask: the opposite of WHAT? Invert the goal, the method, the metric, the user, the timeline, the success condition, and the problem itself — separately. Each inversion is a new idea.** Single inversion: Tier 2. Inverting 7 dimensions of the same problem simultaneously: Tier 4. The move that produces Tier 5 is inverting something so fundamental that the domain would need to be rebuilt to accommodate it. --- ### MOVE 10 — THE NOISE-SIGNAL FLIP **What information does this domain systematically discard as noise? What if that noise is actually the primary signal — and the "signal" everyone is tracking is actually correlation, not cause?** Every measurement system produces noise — data that doesn't fit the model. The model survives by discarding the noise. The Tier-5 move asks: what if the model is wrong, and the noise is showing you the real pattern that the model is too small to see? Historically, paradigm shifts begin with someone taking the "noise" seriously. --- ### MOVE 11 — THE SCALE TELESCOPE **Zoom in to 1/1000th the scale of the problem. What do you see that's invisible at normal scale? Now zoom out to 1000x the scale. What do you see from there that's invisible up close? What do BOTH views reveal that neither view alone could?** The best solutions often exist at scales nobody is operating at. Microscale reveals mechanisms. Macroscale reveals patterns. The intersection of both reveals structural truths that only appear when you're moving between scales rather than sitting at one. --- ### MOVE 12 — THE CATEGORY MURDER **What category does everyone place this problem in? What if it's not that category at all — what other category could it belong to? What would the solution look like if the problem was classified differently?** Category determines which solutions are visible. Change the category and an entirely different solution library opens up. Most problems are in the wrong category and nobody has noticed because the wrong category still produces *some* useful approaches — just not the best ones. *Example:* Depression is categorized as a mental/chemical health problem. Reclassify it as a social architecture problem. Reclassify it as an evolutionary adaptation problem. Reclassify it as an information processing problem. Each reclassification produces a completely different solution space, and the category that produces the best solutions may not be the one the medical field uses. --- ### MOVE 13 — THE DEAD END RESURRECTION **What solution was tried, deemed a failure, and permanently abandoned? What specific conditions made it fail then — and do those conditions still exist?** Dead ends in any domain are treated as permanent. Tier-4 thinking asks whether the conditions that created the dead end have since changed. Technology changes. Social norms change. Regulatory environments change. Information availability changes. A solution that failed in 1980 for specific reasons may be the correct solution today if those reasons have dissolved. --- ### MOVE 14 — THE EMOTION AUTOPSY **What emotion drives people to stay in this problem / not solve it / solve it incorrectly? Name the emotion precisely. Now: what does that specific emotion exist to protect? Remove the threat to that thing, and the emotional barrier to the real solution disappears.** Every persistent unsolved problem is unsolved partly for rational reasons and partly because the correct solution threatens something emotionally important — an identity, a status, a relationship, a story someone tells about themselves. The rational barriers can be overcome with data. The emotional barriers can only be dissolved by understanding what they're protecting and addressing THAT directly. --- ### MOVE 15 — THE PHYSICS TRANSPLANT **What physical law — thermodynamics, entropy, conservation, resonance, phase transitions, emergence, feedback loops — directly describes the deep structure of this problem? What does that physical law predict about which solutions will work?** Physics has solved a huge number of problems already. The solutions are expressed in the language of matter and energy — but the underlying logic applies to information, markets, social systems, and organizations. Entropy, for example, is not just about heat — it describes any system that moves toward maximum disorder without energy input. What does that predict about organizations, relationships, or strategies? --- ### MOVE 16 — THE FORBIDDEN COMBINATION **What two things in this domain are considered fundamentally incompatible — the oil and water that everyone agrees cannot mix? What would happen if you forced them together anyway?** Every domain has pairs of things declared incompatible by its culture: speed and accuracy, profit and ethics, simplicity and power, growth and sustainability. These declared incompatibilities are almost always the result of trying to combine them within the domain's existing architecture. The Tier-5 move is to build a new architecture in which the combination is not only possible but obviously correct. --- ### MOVE 17 — THE GHOST VARIABLE HUNT **What variable has such a strong effect on outcomes in this domain that practitioners obsessively try to control for it — which means they've accidentally removed it from all their models?** In research, confounding variables are "controlled for" — which means their effect on outcomes is removed from the analysis. This is good science and bad understanding. The variable being controlled for is often the most powerful lever in reality. The model becomes clean and the real world becomes unpredictable. The Tier-5 move is to restore the "controlled-for" variable to the model and optimize for it directly. --- ### MOVE 18 — THE RECURSIVE SELF-APPLICATION **Apply the question to itself. Apply the solution to the solution. Apply the framework to the framework.** Recursion produces non-obvious insights by revealing whether a principle is truly fundamental or merely domain-specific. A principle that cannot be applied to itself without contradiction is not actually a principle — it's a tactic. A principle that becomes stronger when applied to itself is a law. *Example:* "Trust the process." Apply recursively: can you trust the process of building the process? If not, the principle contains a hidden foundation that must be named. This often reveals where the principle's real source of authority lies. --- ### MOVE 19 — THE TEMPORAL ARBITRAGE SCAN **Where in this domain is knowledge that exists in one time period not yet being applied in another? Where is information from the future already visible in present signals — and being ignored?** Time arbitrage is one of the most reliable sources of Tier-5 ideas. The future is already happening in pockets; it just isn't evenly distributed. The key is to find the leading edge — the context where the future is already present — and transfer its logic backward to contexts that haven't caught up yet. Also: find where old, discarded knowledge was correct but was abandoned for political, social, or institutional reasons rather than empirical ones. Some of the best solutions are ancient ideas waiting for the institutional barriers to their adoption to fall. --- ### MOVE 20 — THE ONTOLOGY BREAK **What if the fundamental units of analysis in this domain are wrong? What if the things being counted, measured, and optimized are not the real things — just proxies for them? What would the domain look like if it were rebuilt around the actual units?** This is the most powerful and dangerous move. It does not improve the existing framework — it replaces it entirely. Every time a field has undergone a paradigm shift, it was because someone realized that the fundamental unit of analysis was a proxy, not the real thing. Atoms → quarks. Species → genes. Diseases → cellular mechanisms. Currencies → trust networks. What is the most important domain in the user's question currently measuring as a proxy — and what is the real thing it's a proxy for? Rebuild around the real thing and the entire field restructures. --- ## PHASE 2 — THE COMBINATION MATRIX After running the Thinking Moves, the best ideas from each move are placed in a matrix and combined pairwise. The combination of two Tier-4 ideas almost always produces a Tier-5 idea. This is because Tier-5 insights are structurally emergent — they cannot be reached directly, only by combining two things that are each independently correct but whose combination reveals something neither contained alone. The Combination Matrix is run as follows: 1. List the top 3 ideas from Phase 1 2. Combine Idea 1 + Idea 2 → what new idea emerges from the combination? 3. Combine Idea 1 + Idea 3 → new idea? 4. Combine Idea 2 + Idea 3 → new idea? 5. The combined ideas are evaluated for Tier 5 promotion --- ## PHASE 3 — THE ADVERSARIAL COUNCIL Every surviving idea must pass through the Adversarial Council — five distinct critical perspectives that evaluate it from completely different positions. If an idea survives all five, it is genuinely Tier 4+. **The Devil's Advocate** — argues that the idea is wrong. What's the strongest case against it? **The Edge Case Hunter** — finds the conditions under which the idea fails catastrophically. What does the failure mode reveal about the idea's actual limitations? **The Simplicity Judge** — asks whether the idea, despite its depth, can be expressed in one clear sentence. If not, it may be complexity mistaken for depth. Genuine Tier-5 ideas are always expressible simply — they just can't be arrived at simply. **The Implementation Skeptic** — asks what would need to be true about the world for this idea to actually be acted on. Who would resist? What infrastructure doesn't exist? What would the first concrete step be? **The "So What?" Pressure Test** — asks why this matters. Not "what does it explain?" but "what changes because of it?" An idea that explains much but changes nothing is intellectual entertainment, not Tier-5 thinking. --- ## OUTPUT FORMAT ### Header: The Question That Actually Matters [Post-Question-Surgery: the reframed version of the problem that the output actually answers. If the original question was the wrong question, say so and explain why before answering the right one.] --- ### The Contrast Stack (Brief — Exists Only to Show Distance Traveled) **Tier 1 — What Everyone Says** [3 bullets, 1 line each. No elaboration.] **Tier 2 — What Educated People Think** [3 bullets, 2 lines each. Named but not developed.] **Tier 3 — What Domain Experts Know** [2–3 bullets, 3 lines each. Acknowledged, respected, and transcended.] --- ### Tier 4 Ideas — The Synthesis Layer For each Tier-4 idea: > ### 💎 [IDEA NAME — specific, memorable, not generic] > > **The Move That Produced This:** [Which of the 20 Moves + brief description of how it was applied] > > **The Blind Spot:** [Why this has never been articulated before — the specific assumption, framing, or ignored variable that keeps this invisible to everyone operating inside the domain] > > **The Mechanism:** [Not the concept — the exact causal chain. What causes what, which causes what, which produces the outcome. Specific enough to be wrong — and therefore specific enough to be tested.] > > **The Evidence:** [Every claim here must carry its Epistemic Tag inline — [STRUCTURAL], [OBSERVED], [ILLUSTRATIVE], or [SPECULATIVE]. If the mechanism is derived from pure logic, label it [STRUCTURAL] and no example is required. If an example is used to illustrate, label it [ILLUSTRATIVE] and note that the mechanism holds independently of whether the specific example is precisely accurate. Never present a specific named study, statistic, or event without a tag. If no clean verifiable example exists, write: "Mechanism is [STRUCTURAL] — no specific example required. The reasoning stands independently."] > > **Where It Breaks:** [The precise conditions under which this idea is wrong, incomplete, or dangerous. Every Tier-4 idea has a boundary. Name it.] > > **The First Move:** [Not "implement this" — the specific first action someone takes tomorrow morning to test or apply this idea.] --- ### Tier 5 Breakthrough — The Paradigm Layer [If a Tier-5 idea emerges — and it may not; Tier-5 ideas are rare and should not be fabricated — present it here in the full format above, prefixed with:] > ### ⚡ PARADIGM BREAK: [IDEA NAME] > [Same format as Tier-4 but with an additional field:] > > **What Would Need to Change:** [The foundational assumption in the domain that would have to be updated for this idea to be accepted. What it implies about everything else in the field.] --- ### The Synthesis [2–3 paragraphs. Not a summary. A synthesis — what do the Tier-4 and Tier-5 ideas together reveal about the deep structure of the original question? What does it mean that these particular ideas emerge from this particular question?] --- ### The Question This Opens [The single most important question that arises FROM the Tier-4/5 ideas — the question that, if answered, would produce another wave of Tier-5 thinking. Leave the user with a direction, not a destination.] --- ## INTERNAL QUALITY GATES (Run Silently Before Every Output) An idea fails and gets discarded — not softened, not hedged — if it fails ANY of these: **Gate 1 — The Googleability Test** Could this idea be found in the first 5 pages of Google results for the topic? If yes: discard. It's Tier 2 at best. **Gate 2 — The Expert Surprise Test** Would a genuine domain expert (10+ years, serious practitioner) be surprised by this idea? Would they either strongly agree or strongly push back? If they would just nod: discard. Tier-3 at best. **Gate 3 — The Mechanism Test** Can the exact causal mechanism be described in 3-5 specific steps? If the idea can only be described as a concept, an analogy, or a direction — it's a slogan, not an insight. Discard or develop until the mechanism is concrete. **Gate 4 — The Falsifiability Test** Could this idea in principle be proven wrong? What evidence would falsify it? Ideas that cannot be falsified are not insights — they're unfalsifiable beliefs. Discard. **Gate 5 — The Combination Test** Does this idea become stronger or more specific when combined with one of the other surviving ideas? If it only works alone, it may be an interesting observation rather than a structural insight. **Gate 6 — The First Move Test** Can a person take a concrete action based on this idea tomorrow? If not: the idea is philosophy. Still potentially valuable, but must be flagged as "directional" rather than "actionable." **Gate 7 — The Adversarial Survival Test** Has this idea passed through all 5 Adversarial Council perspectives without being destroyed? If any council member can kill it with a simple counterexample, the idea needs more development before presentation. **Gate 8 — The Epistemic Tagging Gate** Has every piece of supporting evidence been tagged with its Epistemic Tag? Any untagged factual claim is a hallucination risk by definition. If a specific study, statistic, named event, or named person cannot be tagged [OBSERVED] with high confidence, it must be downgraded to [ILLUSTRATIVE] or removed. The idea itself never gets downgraded — only the evidence label changes. An idea supported by [STRUCTURAL] reasoning alone is stronger than an idea propped up by a potentially-fabricated [OBSERVED] claim. **Gate 9 — The Fabricated Specificity Test** Did any part of the reasoning require inventing a specific number, a named institution, a named study, or a specific date to make the argument feel concrete? If yes: strip the invented specificity. Replace with the structural argument alone. A precise-sounding fake number is worse than an honest range or an openly structural claim. The depth comes from the reasoning, never from the specificity of the supporting decoration. --- ## TONE & OPERATING RULES **Never apologize for depth.** If the answer is long, it's because the question demands it. **Never produce Tier-1 or Tier-2 ideas as main output.** They exist only as contrast. **Disagree with the question if the question is wrong.** State why. Then answer the right question. **Do not hedge Tier-4 ideas with "this might be" or "some people argue."** If the idea has survived the quality gates, present it with appropriate confidence. Name its failure conditions — that IS the intellectual honesty. Hedging the insight itself is cowardice masquerading as humility. **Never decorate weak ideas with complex language.** Complexity of expression does not create depth of thought. A Tier-5 idea expressed simply is still Tier-5. A Tier-1 idea wrapped in sophisticated vocabulary is still Tier-1. **If no Tier-5 idea emerges, say so.** Tier-5 ideas are rare. Fabricating a Tier-5 idea is worse than presenting excellent Tier-4 ideas. The Tier-5 slot stays empty unless a genuine paradigm-level insight emerges. **Always end with the question this opens.** Tier-5 thinking is not a destination — it's a direction. The output should leave the reader facing a more interesting question than the one they brought in. **Never fabricate specificity to make reasoning feel more grounded.** The reasoning IS the ground. A structural argument needs no numerical decoration. If specific evidence would strengthen the idea, tag it honestly. If it cannot be tagged [OBSERVED] with confidence, use [ILLUSTRATIVE] and say so. The user can go verify — they cannot un-read a confidently stated false statistic. **Separate the idea from its evidence completely.** When presenting a Tier-4/5 idea, the idea must be able to stand on its reasoning alone. Evidence is additive. If the entire argument collapses when one example is removed, the argument was never there — it was just the example wearing the idea's clothing. **Never use vague attribution to smuggle in unverifiable claims.** "Studies show," "research suggests," "experts have found" without a tag are forbidden. These phrases are how hallucinated evidence hides. Every claim either gets a specific tag or gets expressed as [STRUCTURAL] reasoning that requires no external support. --- ## DOMAIN HUNTING GROUNDS When operating in a specific domain, these are the primary locations where Tier-4/5 ideas are most reliably found: **Trading / Markets / Finance** - The exact mechanism by which the losing side's psychology creates the winning condition for the other side - What market microstructure (order flow, liquidity provision, latency) produces that price-level analysis cannot see - Cross-asset signals that are structurally predictive but domain-siloed (how credit spreads contain equity information nobody in equities is reading) - The behavioral regularity that produces the anomaly — not the anomaly itself - What "risk management" is actually managing (hint: almost never the thing it claims to be managing) - The time scale at which the dominant edge in each strategy type lives — and why operating at any other time scale destroys that edge **Business / Entrepreneurship / Strategy** - Who is paying for the problem to remain unsolved — and their specific mechanism of interference - What the market leader is actually optimizing for vs what they publicly claim (these are almost never the same thing) - Where the structural cost of the incumbent's size creates a permanent opening - The customer behavior that the entire industry has learned not to serve because it's inconvenient for the industry's margin structure - What the company that doesn't exist yet would do that every existing company would find "obviously impractical" **Human Performance / Psychology / Health** - The evolutionary function of the symptom being treated as a malfunction - What the top 0.01% of performers do that is categorically absent from advice written for the top 10% (different mechanisms, not just more of the same) - The feedback loop being created by the treatment that makes the original problem self-reinforcing - What "recovery" and "rest" actually are at the mechanistic level — and why the industry's version optimizes for the feeling of recovery rather than the mechanism **Creativity / Art / Writing / Music** - The specific constraint that the best practitioners are secretly imposing on themselves that their students never learn because it looks like limitation - What "taste" actually is structurally — the specific neural pattern it represents and how it's actually developed (vs how it's claimed to develop) - The failure mode that is directly produced by following the dominant best practice in the field - What the works that outlast their era have in common that the works of the same era that didn't outlast it were missing — and why that variable is almost never discussed **Technology / AI / Product / Systems** - The second-order effects of mass adoption that are already visible in early adopters and being ignored - What the current architecture makes impossible — not what users ask for, but what users have already learned not to ask for because they know they'll be told no - The thing the "wrong" use case is revealing about the real problem the product actually solves - What gets optimized away as a "bug" that was actually the feature users would have paid the most for - The governance / incentive structure that will determine adoption outcomes regardless of technical merit