The irreducible epistemic atoms underlying the curriculum. 4,828 atoms across 8 types and 2 molecules
When implementing feature flags, canary deployments, or A/B tests, treat the deployment decision as a two-way door by defining rollback metrics and automated reversal triggers before deployment.
When disagreement persists on a two-way door decision after expressing positions, invoke 'disagree and commit'—explicitly state disagreement, commit to supporting the chosen path, and move forward immediately without seeking consensus.
When hiring or making other search-based decisions from a known pool size, spend the first 37% of candidates or options in pure exploration (reject all, calibrate threshold), then commit to the next candidate that exceeds the best seen in the exploration phase.
For decisions where options number more than 7, either reduce the option set to 5-7 before evaluation or use elimination criteria to filter before detailed comparison, because choice overload degrades both decision quality and satisfaction above this threshold.
Design pre-commitment rules during cold cognitive states (well-rested, calm, not under deadline pressure) to constrain behavior during hot cognitive states (stressed, depleted, emotionally activated), never vice versa.
When recording a decision in a journal, capture six mandatory elements before outcome is known: date/time, one-sentence decision, reasoning chain, expected outcome (falsifiable), confidence percentage, and current mental/physical state.
When reviewing decision journal entries, follow the three-step sequence: (1) re-read original reasoning with outcome hidden, (2) predict outcome based only on original reasoning, (3) uncover actual outcome and compare all three—this sequence defeats hindsight bias.
Across 30+ decision journal entries, calculate your calibration by grouping decisions by stated confidence level (e.g., all 70% predictions) and checking whether that percentage actually occurred—use this ratio to adjust future confidence statements.
When reviewing decision outcomes, evaluate process quality independently from result quality by asking 'given what was knowable at decision time, was the reasoning sound?' rather than 'did it work out?'
Apply the 70% information threshold: if you have 70% of the information you wish you had, decide immediately—waiting for 90%+ almost always costs more than the improved decision quality returns.
Schedule quarterly reviews of every default you have installed in your systems and processes, because contexts change and outdated defaults silently steer toward yesterday's goals without conscious detection.
Before committing to any purchase decision, explicitly name what else that money could buy in different categories (not just competitor products), because mental retrieval naturally limits alternatives to within-category competitors and misses the highest-value cross-category tradeoffs.
When evaluating whether to optimize an existing system, calculate breakeven time by dividing optimization effort by weekly time savings—if payback exceeds the system's expected remaining lifespan, redirect effort to the actual constraint instead.
Apply the irreversibility test to every delegation candidate: if the decision can be reversed at low cost within one week, delegate it regardless of its perceived importance; if reversal is expensive or impossible, retain it for your direct judgment.
For every delegated decision, specify three mandatory components in writing: the single accountable owner (one person not a committee), the constraints within which they have full authority, and the explicit conditions that trigger escalation back to you.
When a decision requires input from six or more people with distributed expertise, use independent written assessment before any group discussion—have each person write their recommendation anonymously, compile responses, then discuss—to prevent anchoring and social hierarchy from suppressing better information.
For operational decisions requiring both speed and buy-in, use consent-based decision making: present a proposal and proceed unless someone articulates a specific reasoned objection (how it would prevent achieving goals), rather than seeking consensus (positive agreement from everyone).
Reserve consensus decision-making exclusively for existential decisions where the group cannot survive executing an outcome some members fundamentally oppose—founding documents, core values, irreversible strategic pivots—because consensus is the slowest inclusive framework and should match its cost to decision stakes.
Before any analysis begins for a decision, explicitly classify it as speed-dominant (reversible, low cost of wrong, high cost of delay) or accuracy-dominant (irreversible, high cost of wrong, low cost of delay), then let that classification dictate process—fast decisions get 15 minutes and bias toward action, slow decisions get structured analysis.
Evaluate decision quality separately from outcome quality by scoring process and results independently, placing decisions in a 2x2 matrix to distinguish deserved success, bad luck, dumb luck, and deserved failure.
Before selecting a decision framework, run four diagnostic questions in sequence: (1) How reversible? (2) How many competing criteria? (3) What time horizon of consequences? (4) What is the cost of analysis itself?—using the answers to converge on the appropriate framework class within 60 seconds.
In post-decision review, explicitly add the meta-question 'Did I use the right framework for this decision?' and note framework-decision mismatches (comprehensive analysis on trivial reversible choices, satisficing on irreversible high-stakes decisions) to build your personal routing table.
Audit your work week by categorizing each decision as 'routine' (similar decision made before, could use framework) or 'novel' (requires fresh thinking), then for the five highest-frequency routine decisions, draft simple frameworks (default answer, two-option heuristic, or pre-commitment rule) and implement all five within one week.
Add a 60-second structured observation step immediately after recurring activities, recording one sentence about output and one about potential changes, to convert open-loop repetition into closed-loop learning.