Concepts
The irreducible epistemic atoms underlying the curriculum. 4,828 atoms across 8 types
Steel-man both sides of a contradiction before resolving…
Before attempting to resolve any contradiction between beliefs, construct the strongest possible version of each side that the most informed advocate would recognize and endorse.
Count only independent evidence lines for confidence — ten…
When evaluating confidence in a belief, count only genuinely independent lines of evidence—sources that do not share origins, methods, or assumptions—rather than total source count, because correlated sources compound confidence on a single foundation.
Ground every abstraction with 3+ examples from different…
Connect each abstract concept in your knowledge system to at least three concrete examples from different domains, because single examples invite surface-feature overgeneralization while multiple examples force attention to shared structural patterns.
Check transitivity before inferring through chains…
When a relationship type is mathematically transitive (like 'is greater than' or 'is ancestor of'), you can safely infer the endpoint connection from a chain; when it is not transitive (like 'is friend of' or 'is close to'), you must verify endpoint relationships directly rather than inferring them.
Reporting hierarchy does not make priorities transitive…
Before assuming organizational hierarchy makes strategic priorities transitive, verify alignment at each reporting level independently, as 'reports to' relationships do not make 'shares priorities with' transitive.
Identify the deepest assumption in each decision domain…
For each major domain where you make decisions, explicitly write down the single deepest assumption everything else depends on, then list 5-10 decisions that would change if that root were different to verify you've found an actual root.
Negate your root assumption and explore the implications…
State the negation of any root concept you identify and ask what you would do differently if the opposite were true, not to believe the negation but to break the structural lock that makes the original feel inevitable.
Override traits, not identity — if you are overriding…
Override properties rather than identities—preserve the child's relationship to its parent category while modifying specific inherited traits, as overriding identity signals you need a different category, not an override.
Defensiveness about testing a schema IS the signal…
When a schema triggers defensiveness at the suggestion of testing it, treat that emotional response as a diagnostic signal of high psychological investment requiring especially rigorous validation.
Schema revisions must generate new predictions, not just…
When adjusting a schema after disconfirming evidence, require the adjustment to generate new testable predictions rather than merely explaining away the original failure.
Edge cases that break your schema reveal boundary…
When an edge case breaks your schema, extract the implicit boundary condition that the edge case revealed rather than dismissing the edge case as an irrelevant exception.
Ask 'tell me where this breaks' not 'do you agree' — get…
Before explaining your schema to another person, frame your request as 'tell me where this breaks' rather than 'do you agree' to shift the conversation from validation theater to genuine testing.
Test each component of a compound schema independently…
Validate each atomic component of a compound schema independently before trusting the complete structure, because compound failures provide no diagnostic information about which component broke.
Check if indirect indicators could agree for reasons other…
When using indirect evidence, assess whether indicators are genuinely independent by checking if they could agree for reasons other than the schema being true—if all evidence shares a common causal source, it counts as single evidence despite multiple data points.
Pre-mortem your critical schemas
Run a pre-mortem on each critical schema by specifying what the early warning signs would look like if that model is becoming obsolete, then check whether you have already seen some of those signs.
Before finalizing a schema update, verify the new version…
Before finalizing any schema update, list five situations where the old schema produced accurate results and verify the new schema handles all five to ensure backwards compatibility.
Sit with cognitive dissonance for 60 seconds before…
Set a tolerance window of at least sixty seconds to sit with cognitive dissonance before attempting resolution, as premature resolution systematically favors existing schemas over new evidence.
Pre-commit to prediction failure thresholds (X out of Y)…
Set prediction failure thresholds as numeric ratios (X failures out of Y recent predictions) for each schema before observing prediction outcomes to trigger review when the threshold is crossed.
When your framework can't even formulate the right…
When your schema can no longer formulate the questions you need to ask about a domain, treat this incommensurability as a signal that the framework itself has become a cage requiring replacement.
An unfalsifiable schema is a belief system, not a model…
When a schema cannot specify any observation that would falsify it, classify it as a belief system rather than a testable model and flag it for replacement or constraint.
Track where each schema came from — prioritize…
For each schema you operate on, document source provenance in a single field—specific person, book, cultural norm, direct experience, or unknown—then prioritize verification effort by source weakness.
Evaluate sources by searching about them, not by reading…
Apply lateral reading by immediately opening new tabs to search for independent information about a source rather than evaluating the source by reading the source itself, because external assessment outperforms internal coherence checking.
When AI suggests a framework, ask for sources, boundary…
When AI assistants suggest frameworks or schemas, respond by asking for original research sources, boundary conditions, and strongest counterarguments rather than accepting or rejecting the claim directly.
A bridge node must generate insights in BOTH connected…
Test each bridge node by verifying it generates novel predictions or actionable insights in both connected domains—if it only produces a sense of similarity without bidirectional inference, demote it to metaphor status or delete it.