Tier 4

ld - Learning Discovery Orderings

Learning Discovery Orderings

Input: $ARGUMENTS

Overview

When the goal is understanding (not just completion), order work to maximize learning. Test assumptions early, validate incrementally, pursue the most informative experiments first.

Core Principle

Do the thing that teaches you the most, not the thing that progresses the most. Early learning compounds.

Ordering Rules

Rule 1: Most Uncertain First

  • Identify what you know least about
  • Address the biggest knowledge gaps first
  • Why: reducing uncertainty early prevents building on wrong assumptions

Rule 2: Test Assumptions Before Building

  • List assumptions your plan rests on
  • Validate the riskiest assumptions before investing in execution
  • When: new domain, unproven approach, high uncertainty

Rule 3: Smallest Viable Experiment

  • Design the cheapest test that gives the most information
  • Prefer quick probes over thorough studies early on
  • Increase rigor as confidence grows
  • When: resource-constrained learning

Rule 4: Alternate Theory and Practice

  • Alternate between: learn concept → apply it → reflect on what happened
  • Don’t front-load all theory or all practice
  • When: skill acquisition, learning by doing

Rule 5: Breadth Then Depth

  • Survey the whole landscape before diving deep into any area
  • Prevents premature specialization in the wrong sub-topic
  • When: entering a new field, starting research

Application Procedure

Step 1: Map What You Know and Don’t Know

  • What’s certain? What’s uncertain? What’s unknown?

Step 2: Prioritize by Information Value

  • Which unknown, if resolved, changes the most about your plan?
  • Address that first.

Step 3: Learn in Cycles

  • Each cycle: hypothesis → test → update → new hypothesis
  • Get faster at cycles rather than trying to get it right the first time

When to Use

  • Entering new domains, research, exploration
  • When requirements are unclear
  • Skill acquisition, R&D

Verification

  • Knowledge gaps identified
  • Most uncertain items addressed first
  • Assumptions tested before relied upon
  • Learning captured and applied to next steps