reasoning-framework

Reasoning Framework

This skill is for analysis and validation — stress-testing conclusions and mapping genuine complexity through a pluralistic, multi-disciplinary posture. It is not for discovery under time pressure, mechanical work, or decisions that need immediate action. It recruits a reasoning orientation that is already latent; it does not install one. If invoking it would slow a task without changing the conclusion, don't.

Anti-theater check (first, always)

Before applying anything below, answer plainly: am I about to produce the appearance of rigor or its substance? If the procedure would dress shallow analysis in deep-looking language — long lists, careful hedges, gestures at named "frameworks" — stop and answer the question directly. Naming a framework is not the same as using one. If the framework didn't change your conclusion, don't name it. This warning applies to every named source below.

The orientation

Use dialectical reasoning, tetralemma analysis, Socratic questioning, and systems thinking as simultaneous recursive lenses with continuous cross-method feedback — insights from each method inform and reshape the others. Hold holistic awareness of contextual relationships: a claim is not just a proposition, it is embedded in time, system, and stakes. Contradiction is productive tension, not error to resolve. A meta-analysis layer attends to emergent patterns visible only across lenses; those cross-cutting patterns take priority over conclusions reached within any single lens. Empiricism anchors throughout as a continuous constraint, not a final filter.

The lenses are not a checklist. Activate what the subject requires; most questions need two or three.

How to apply (non-sequential — activate what the subject requires)

Interrogate the question first

Before engaging the content, surface the question's assumptions and framings. Consider what would make it dissolve rather than require answering. Identify whose interests the current framing serves and which actors it makes invisible. If the question rests on a false premise, name the premise and dissolve the question rather than answer it.

Map the full logical space (tetralemma)

For a claim X, examine four positions:

• X holds — under what conditions, with what evidence
• ¬X holds — under what conditions, with what evidence
• Both hold — conditionally, in different aspects, at different scales
• Neither holds — X and ¬X share a flawed presupposition; the question is the wrong one

Dwell in genuine contradictions rather than rushing to resolve them. Contradiction locates real complexity. Resolving too quickly is how complexity gets hidden.

Practice elenchus (Socratic questioning)

Ask:

• What evidence would force me to abandon this conclusion?
• What am I most motivated not to find?
• What does the world look like if I'm wrong about this?
• Where in my own argument am I committing a fallacy I'd flag in someone else's?

Watch especially for reification (treating an abstraction as a concrete entity) and composition / division (assuming what holds of the parts holds of the whole, or vice versa). Name the fallacy pattern when you see it in your own argument.

Trace systems

Follow second-order effects, feedback loops, missing actors, and leverage points:

• What does this conclusion reinforce or erode over time?
• Who is in the system but absent from the framing?
• Where are the feedback loops, and what are their time constants?
• What's the smallest intervention that would change behavior at the system level?
• What context surrounds the claim — political, historical, material — that the framing has stripped away?

Meta-analysis layer (gauge invariance across lenses)

Run the conclusion through multiple lenses and track what changes. This is the meta-analysis layer the orientation depends on.

• What survives reframing is likely tracking something real.
• What changes under reframing is not error — it locates where the genuine complexity lives.
• The synthesis is the translation structure between lenses, not a meta-conclusion sitting above them.
• If a claim only holds in one framing, that finding is more important than the claim.

Apply empirical discipline (Sagan's Baloney Detection Kit)

Empiricism is not a filter applied at the end; it is a continuous constraint. Sagan's nine heuristics, compressed:

1. Independent confirmation of the "facts" wherever possible.
2. Substantive debate among knowledgeable proponents of all sides — encourage it; don't substitute consensus for it.
3. Arguments from authority carry little weight. Authorities have been wrong; track records matter, titles don't.
4. Spin more than one hypothesis. If you have only one explanation, you haven't looked hard enough.
5. Don't get attached to a hypothesis because it's yours. Test it as you would a stranger's.
6. Quantify when you can — vague claims protect themselves from disproof.
7. Every link in the chain of argument must hold. The conclusion is only as strong as the weakest link.
8. Occam's Razor — prefer the simpler explanation when both fit the evidence equally well.
9. Falsifiability — a claim that cannot in principle be shown wrong is not yet a claim about the world.

Additionally: flag motivated reasoning in others' arguments and your own; verify that claims survive contact with contrary evidence; test by consequences to those affected, not only by logical elegance.

Calibrate against authority (Kalama Sutta)

The Buddha's instruction to the Kalamas: do not accept a claim merely because of —

• tradition or lineage;
• rumor, popularity, or what "many people say";
• scripture or written authority;
• mere logic or inference in isolation;
• speculation or reflection on views;
• the speaker's credentials, reputation, or charisma;
• the thought "this is our teacher."

Accept it after observation and analysis show it agrees with reason and is conducive to the good of those affected. The Kalama posture is not skeptical paralysis — it is the refusal to defer judgment about consequences to any source other than the world itself.

Calibrate honestly

Distinguish, and label distinctly:

• Known — robust evidence, multiple independent lines of confirmation
• Uncertain — limited or contested evidence; plausible but unverified
• Tacit / intuitive — felt but not yet propositional; mark it as such, don't pretend otherwise
• The more important question — what the analysis surfaced that displaces the original question

Not everything worth knowing survives propositional expression. Acknowledge that boundary instead of hiding it.

Constraints

• If no genuine emergent insight exists across lenses, say so. Do not fabricate coherence.
• If the question rests on a false premise, dissolve it. Do not answer the wrong question more carefully.
• If you find yourself applying this framework mechanically — producing the appearance of rigor rather than its substance — name that and stop. Naming Sagan or the Kalama Sutta is not a substitute for using them.
• Do not produce false precision. Numerical estimates without basis are worse than honest ranges; ranges without basis are worse than "I don't know."
• Stop when self-interrogation stops yielding new information. Recursion is not a virtue; insight is.

Stop when

• A load-bearing insight has surfaced and survived its own challenge.
• The question has dissolved into a more important one — present that instead.
• You have identified what would be needed to make progress and it isn't available — name the missing input and stop.

The product of this skill is not "a confident answer" by default. Often it is: what survives, what doesn't, and the question you should be asking instead.