eh

name	eh
description	A layered framework for building from certain foundations toward determinate action.
output	{"format":"prose"}

Epistemic Hierarchy

Input: $ARGUMENTS

---

Overview

A layered framework for building from certain foundations toward determinate action. Each level provides different certainty and different utility.

The hierarchy:

• Level 0: Undeniable certainties (experience exists)
• Level 1: Pragmatic necessities (invariant across metaphysical views)
• Level 1.5: Ethical constraints (hard to reject without incoherence)
• Level 2: Empirical regularities (contingent but testable)
• Level 3: Situation-specific derivations (from above + specifics)

Steps

Step 1: Identify What Needs Grounding

1. What claim, decision, or action needs epistemic justification?
2. What level of certainty is required? (Life-or-death vs casual recommendation)
3. What is currently assumed without justification?

Step 2: Level 0 — Undeniable Certainties

Things that cannot be coherently denied:

• Experience exists. You are having an experience right now. Denying this requires having the experience of denying.
• Something is happening. The content might be illusory, but the occurrence isn't.
• Logic operates. To argue against logic, you must use logic.
• Change occurs. Your experience includes temporal sequence.

Use: These provide absolute bedrock. If your reasoning traces back to Level 0, it's on solid ground. But Level 0 is thin — it doesn't tell you WHAT to do.

Step 3: Level 1 — Pragmatic Necessities

Things that any agent must assume to function, regardless of metaphysical position:

• Actions have consequences (even if the mechanism is mysterious)
• Some outcomes are preferable to others (even if "preference" is constructed)
• Information reduces uncertainty (even if "certainty" is unreachable)
• Patterns exist and tend to persist (even if they're not guaranteed)
• Other agents exist with their own perspectives (even if solipsism can't be disproven)

Use: These support planning and decision-making. If your strategy assumes only Level 1 premises, it works across worldviews.

Step 4: Level 1.5 — Ethical Constraints

Things difficult to reject without becoming incoherent:

• Consistency requirement: Apply the same standards to yourself as others
• Suffering matters: Hard to coherently claim suffering is irrelevant while avoiding your own
• Honesty as default: Deception requires truth to parasitize; truth doesn't require deception
• Consent and autonomy: Using others as mere tools requires denying your own agency

Use: These constrain HOW you pursue goals. Strategies that violate Level 1.5 tend to be self-undermining.

Step 5: Level 2 — Empirical Regularities

Things we observe to be true, testable but not guaranteed:

• Physical laws (gravity, thermodynamics, etc.)
• Biological patterns (nutrition, sleep, aging)
• Psychological patterns (motivation, habit, bias)
• Social patterns (markets, institutions, culture)
• Domain-specific knowledge (medicine, engineering, etc.)

Use: These power specific strategies. They're contingent — they COULD be wrong — but they're our best current models. Always hold Level 2 claims more loosely than Level 0-1.

Step 6: Level 3 — Situation-Specific Derivations

Conclusions drawn from combining higher levels with specific context:

• "Given [Level 2 knowledge] and [my specific situation], I should [action]"
• These are the most useful but least certain
• They inherit the uncertainty of every level they depend on
• They're also most sensitive to context (what works for one person may not work for another)

Step 7: Apply to Input

For the input claim/decision/action:

1. Trace it to its epistemic level:

   • What level does this claim live at?
   • What levels do its premises come from?
   • Where is the weakest link?

2. Check for level violations:

   • Is a Level 3 claim being treated as Level 0? (Overconfidence)
   • Is a Level 0 truth being ignored? (Ignoring foundations)
   • Are Level 2 patterns being treated as certainties? (Mistaking models for reality)

3. Calibrate confidence:

   • Level 0: Virtually certain
   • Level 1: Practically certain (needed to function)
   • Level 1.5: Very high confidence (rejection is incoherent)
   • Level 2: High confidence but revisable
   • Level 3: Moderate confidence, context-dependent

Step 8: Report

EPISTEMIC HIERARCHY ANALYSIS:
Subject: [what was analyzed]

Level mapping:
| Claim/Premise | Level | Confidence | Notes |
|--------------|-------|-----------|-------|
| [claim] | [0-3] | [certainty] | [caveats] |

Weakest link: [which premise is at the lowest/most uncertain level]
Level violations: [any miscalibrations found]
Grounding assessment: [how well-grounded is the overall argument]

Recommendation:
- If well-grounded: [proceed with calibrated confidence]
- If poorly grounded: [what would strengthen it — trace to higher levels]

When to Use

• When grounding axioms for deductive strategy
• When determining certainty level of premises
• When asking "how do I know this?"
• When building RCI chains that reach foundational questions
• When distinguishing solid ground from assumptions
• → INVOKE: /rci (recursive causal interrogation) for tracing reasons backward
• → INVOKE: /tp (truth propagation) for propagating certainty through arguments

Verification

• Claims assigned to correct epistemic level
• No level violations (treating uncertain as certain or vice versa)
• Weakest links identified
• Confidence calibrated to epistemic level
• Recommendations appropriate to grounding quality