Skip to content

Factor Diagram Naming: Research Report

📋Page Status
Page Type:ContentStyle Guide →Standard knowledge base article
Quality:31 (Draft)⚠️
Importance:0 (Peripheral)
Words:1.6k
Structure:
📊 18📈 0🔗 0📚 5521%Score: 10/15
LLM Summary:Internal research report comparing terminology for factor diagrams across 8 frameworks (influence diagrams, causal loop diagrams, crux maps, etc.), recommending 'Crux Map' as most appropriate for the EA/rationalist audience. Provides implementation checklist for renaming component throughout codebase.
Issues (1):
  • QualityRated 31 but structure suggests 67 (underrated by 36 points)
FindingKey InsightRecommendation
”Cause-effect” is misleadingOur diagrams aren’t strictly causal—they include uncertainties, considerations, and influence relationshipsRename to something more accurate
Multiple established frameworks existInfluence diagrams, driver diagrams, argument maps, crux maps all map similar territoryBorrow terminology strategically
Uncertainty types matterAleatoric (inherent randomness) vs epistemic (knowledge gaps) distinction is well-establishedConsider incorporating uncertainty typology
”Crux” has traction in target audienceRationalist/EA community already uses “crux” terminology from CFARStrong candidate for naming
Recommended name”Crux Map” or “Factor Map”Balances clarity, accuracy, and audience fit

This report surveys similar conceptual frameworks across multiple fields to identify appropriate terminology for LongtermWiki’s factor visualization system. We examined:

  • Decision analysis: Influence diagrams, decision trees, tornado diagrams
  • Systems thinking: Causal loop diagrams, stock-flow models
  • Philosophy/Argumentation: Argument maps, IBIS, crucial considerations
  • Rationalist community: Crux mapping, double crux
  • Quality improvement: Driver diagrams, theory of change
  • Knowledge representation: Concept maps, knowledge graphs

Established 1981

Origin: Introduced by Howard and Matheson in 1981, building on Howard Raiffa’s decision analysis work.

Definition: A graph representing a Bayesian decision problem with three node types:

  • Circles: Random variables (uncertainties)
  • Rectangles: Decision variables (choices)
  • Diamonds: Utility/value variables (outcomes)
StrengthWeakness
Distinguishes uncertainty from decisionsTechnical/academic connotation
Well-established in decision theoryImplies quantitative Bayesian framework
Supports automatic model generationLess familiar outside specialized fields

Sources: Decision Analysis Wikipedia, INFORMS Decision Analysis


2. Causal Loop Diagrams (Systems Thinking)

Section titled “2. Causal Loop Diagrams (Systems Thinking)”
Systems Dynamics

Origin: Emerged from System Dynamics practice, popularized by Jay Forrester and the MIT school.

Definition: A diagram showing causal relationships between variables, with emphasis on feedback loops:

  • Reinforcing loops (R): Change compounds (e.g., compound interest)
  • Balancing loops (B): Change counteracts (e.g., thermostat)
  • Polarity (+/-): Whether variables move together or opposite
StrengthWeakness
Handles feedback loops nativelyOur diagrams are DAGs (no cycles)
Clear visual conventions”Causal” in the name has same problem
Popular in organizational contextsEmphasizes dynamics over static factors

Sources: Causal Loop Diagram Wikipedia, Systems Thinker, Cascade Institute Handbook


Philosophy/Design

Origin: IBIS (Issue-Based Information System) invented by Werner Kunz and Horst Rittel in the 1960s for tackling “wicked problems.”

Core Elements:

  • Issues: Questions to be addressed
  • Positions: Proposed answers/solutions
  • Arguments: Pro/con rationales supporting or opposing positions
StrengthWeakness
Designed for complex, ill-defined problemsFocused on argumentation, not factors
Captures disagreement structureThree-type ontology is restrictive
Software tools exist (Compendium)Less familiar than other terms

Tools: Compendium, DRed, designVUE

Sources: IBIS Wikipedia, Kunz & Rittel 1970, Cognexus


Bostrom 2007/2014

Origin: Concept introduced by Nick Bostrom in 2007, elaborated in a 2014 talk.

Definition: “A consideration such that if it were taken into account, it would overturn the conclusions we would otherwise reach about how we should direct our efforts.”

Key concepts:

  • Crucial consideration: Overturns conclusions
  • Crucial consideration component: Becomes crucial when combined with other considerations
  • Deliberation ladder: Sequence of successive reassessments
StrengthWeakness
Directly relevant to LongtermWiki’s mission”Consideration” is abstract
Familiar to EA/rationalist audienceMore about ideas than structure
Captures the “changes your mind” aspectNot specifically a diagram type

Sources: Bostrom’s website, EA Forum, Effective Altruism article


CFAR

Origin: Developed by CFAR (Center for Applied Rationality) as part of their curriculum.

Definitions:

  • Crux: A belief C that, if changed, would change your belief in B
  • Double crux: A shared crux where two disagreeing people would both change their minds

Victory condition: “Agreement on a shared causal model of the world—you’ve won when both you and your partner agree to the same if-then statements.”

StrengthWeakness
Familiar to target audienceJargon outside rationalist circles
Emphasizes “what would change your mind”Originally about dyadic disagreement
”Crux” is memorable and specificSome critique it as overly idealized

Sources: LessWrong Double Crux, CFAR Resources, Basic Double Crux Pattern


Healthcare QI

Origin: Widely used in healthcare quality improvement, particularly with Plan-Do-Study-Act (PDSA) cycles.

Structure:

  • Aim: What you’re trying to achieve
  • Primary drivers: High-level factors that must be influenced
  • Secondary drivers: Specific factors/interventions
  • Change ideas: Concrete actions
StrengthWeakness
Clear hierarchical structureAssociated with healthcare/QI
”Driver” is intuitiveImplies more agency than we model
Links factors to interventionsMay feel too operational

Sources: IHI Driver Diagram, AHRQ Key Driver Diagram


7. Theory of Change / Logic Models (Nonprofit)

Section titled “7. Theory of Change / Logic Models (Nonprofit)”
Philanthropy

Origin: Common in nonprofit and philanthropic contexts.

Key distinction:

  • Theory of Change: Explains why change will occur, works backward from impact
  • Logic Model: Describes what will happen, works forward from inputs
StrengthWeakness
Familiar to funders/nonprofitsAssociated with program evaluation
Emphasizes causal pathwaysLess about uncertainty
Links to impact measurementMay feel too outcome-focused

Sources: NPC Theory of Change, La Piana comparison, TOC Toolkit


Knowledge Representation
FeatureMind MapConcept MapKnowledge Graph
StructureTree (single center)Network (multiple hubs)Graph/Network
RelationshipsImplicitLabeled connectionsTyped relationships
OriginTony Buzan (1974)Joseph Novak (1970s)Computer science

Our diagrams are most similar to concept maps: network structure with labeled relationships between nodes.

Sources: NN/g Cognitive Maps, Gloow comparison, Concept Map Wikipedia


TypeDefinitionCan be reduced?Example
AleatoricInherent randomness in the systemNo—only quantifiedCoin flip outcome
EpistemicUncertainty due to lack of knowledgeYes—with more information”Will this policy pass?”

Key insight: “Epistemic uncertainty tends to be expressed using phrases like ‘I am 90% sure’ whereas aleatory uncertainty tends to be expressed using phrases like ‘I think there’s a 90% chance.’”

Our diagrams mostly represent epistemic uncertainty—things we don’t know but could learn—rather than inherent randomness.

Sources: Springer Machine Learning, Berkeley Statistics, Towards Data Science


Based on this research, here are the main naming candidates:

DimensionAssessment
AccuracyHigh—captures “what would change your mind”
FamiliarityHigh for EA/rationalist audience, low outside
DistinctivenessHigh—not easily confused with other diagram types
Fit with LongtermWikiExcellent—aligns with Crux Graph component
DimensionAssessment
AccuracyHigh—we are mapping factors
FamiliarityMedium—generic but clear
DistinctivenessMedium—could be confused with factor analysis
Fit with LongtermWikiGood—neutral, accessible
DimensionAssessment
AccuracyHigh—captures influence relationships
FamiliarityMedium—some association with decision analysis
DistinctivenessMedium—could be confused with influence diagrams
Fit with LongtermWikiGood—clear meaning
DimensionAssessment
AccuracyHigh—we’re mapping considerations
FamiliarityLow—not established terminology
DistinctivenessHigh—distinctive
Fit with LongtermWikiGood—connects to “crucial considerations”
DimensionAssessment
AccuracyMedium—implies more agency
FamiliarityMedium—established in QI
DistinctivenessMedium—associated with healthcare
Fit with LongtermWikiModerate—less about uncertainty
DimensionAssessment
AccuracyMedium—we map factors, not just priorities
FamiliarityLow—not established
DistinctivenessMedium
Fit with LongtermWikiModerate—captures decision-relevance
DimensionAssessment
AccuracyMedium—emphasizes tractability
FamiliarityLow—not established
DistinctivenessHigh
Fit with LongtermWikiModerate—overemphasizes changeability

Rationale:

  1. Directly aligns with LongtermWiki’s mission (“Crux Graph” is already a core component)
  2. Familiar to the target audience (EA/rationalist community)
  3. Captures the key insight: “what would change your mind?”
  4. Distinctive—won’t be confused with other diagram types
  5. The word “crux” conveys importance and decision-relevance

Use if: You want broader accessibility outside the rationalist community. “Factor” is neutral, clear, and doesn’t require explaining jargon.

Our diagrams include four scoring dimensions. Consider renaming to clarify uncertainty types:

Current NameSuggested RenameRationale
noveltyKeep as isClear meaning
sensitivityKeep as isClear meaning
changeabilitytractabilityMore common in EA
certaintyepistemic confidenceDistinguishes from aleatoric

If renaming to “Crux Map”:

  • Update cause-effect-diagrams.mdxcrux-maps.mdx
  • Update component name: CauseEffectGraphCruxMap
  • Update YAML property: causeEffectGraphcruxMap
  • Update style guide references
  • Update skill name: /cause-effect-diagram/crux-map
  • Update documentation throughout