Backward Reasoning
Input: $ARGUMENTS
Interpretations
Before executing, identify which interpretation matches the user’s input:
Interpretation 1 — Reverse-engineer an outcome: Something happened and you want to reconstruct the chain of decisions, causes, or conditions that led to it (e.g., “how did we end up here?” or “what had to be true for this to happen?”). Interpretation 2 — Goal-backward planning: You have a desired end state and want to work backward from the goal to figure out what steps are needed (e.g., “work backward from launch day” or “what do I need to have in place?”). Interpretation 3 — Understand the purpose behind a claim: Someone has stated a position, belief, or conclusion, and you want to trace the reasoning journey that would lead a person to arrive there (e.g., “why would someone believe this?” or “what’s the logic behind this position?”).
If ambiguous, ask: “I can help with reconstructing how an outcome happened, planning backward from a goal, or tracing the reasoning behind a claim — which fits?” If clear from context, proceed with the matching interpretation.
Overview
Typical reasoning goes “forward” - from premises to conclusions. This procedure goes “backward” - from conclusions to premises.
Given a conclusion, what came before? What journey led here?
This is harder than forward reasoning but reveals PURPOSE and enables evaluation of the journey, not just the endpoint.
Depth Scaling
Default: 2x. Parse depth from $ARGUMENTS if specified (e.g., “/br 4x [input]”).
| Depth | Min Backward Steps | Min Premise Chains | Min Alternative Paths | Min Validation Checks |
|---|---|---|---|---|
| 1x | 3 | 1 | 1 | 1 |
| 2x | 5 | 2 | 2 | 2 |
| 4x | 7 | 3 | 3 | 3 |
| 8x | 10 | 5 | 5 | 5 |
| 16x | 14 | 7 | 7 | 7 |
These are floors. Go deeper where insight is dense. Compress where it’s not.
Steps
Step 1: Start with the endpoint
Take the conclusion, statement, or position as given. Don’t question it yet - treat it as the END of a journey.
Ask: “Someone arrived at this. What journey brought them here?”
Output: endpoint_statement
Step 2: Ask “What goal does this serve?”
This conclusion was reached for a REASON. What is the statement trying to achieve?
Not: What are its logical implications? But: What PURPOSE does it serve?
Output: immediate_purpose
Step 3: Ask “What problem generated that goal?”
Goals arise from problems/needs. What problem would make this goal relevant?
Example: Goal: Establish something certain Problem: Everything seems doubtable
Output: generating_problem
Step 4: Ask “What context created that problem?”
Problems arise in contexts. What situation would make this problem salient?
Example: Problem: Everything seems doubtable Context: Skepticism is challenging previous beliefs
Output: generating_context
Step 5: Continue tracing until reaching foundational goals
Keep asking backward:
- What led to this context?
- What underlying values are at play?
- What would have to be true/desired for this journey to make sense?
Stop when reaching intrinsic goals (apply intrinsic_goal_termination_gate).
Output: full_backward_trace
Step 6: Construct the forward story
Now reverse the trace to construct the “story”:
Chapter 1: Foundational goal/value Chapter 2: Context that made it relevant Chapter 3: Problem that arose Chapter 4: Goal that addressed the problem … Final: The conclusion
This is the journey that (hypothetically) led to the endpoint.
Output: reconstructed_story
Step 7: Evaluate the story
Apply story_coherence_gate:
- Does the story cohere?
- Are the goals legitimate?
- Does the conclusion serve the goals?
- Was the journey necessary?
- Is this a valid path?
If coherent: You’ve understood WHY this conclusion exists. If not coherent: Either the conclusion is confused OR your reconstruction is wrong.
Output: story_evaluation
When to Use
- Evaluating a philosophical claim or position
- Understanding why someone believes something
- Analyzing a criticism or objection
- Making sense of an unfamiliar argument
- Finding the PURPOSE behind any statement