Selection

Input: $ARGUMENTS

Overview

Make the final selection from ranked options after all analysis is complete. This procedure transforms human-context filters into LLM-verifiable assessments.

Step 0: Context Assessment

If URGENT: Steps 1, 4, 5 only (skip extensive feasibility) If IMPOSSIBLE reversibility: Include Step 6 (extended rationale)

Step 1: Review Rankings and Context

Examine the input from comparison/optimization.

What to examine:

1. Top-ranked option and why
2. How close are top options in score?
3. Key differentiating factors
4. Any red flags in the analysis?

LLM Execution:

From input/prior analysis, extract:
- Top 3 options with scores
- Score gaps between them
- Primary differentiators
- Any noted uncertainties

Output format:

RANKINGS REVIEW
===============
#1: [option] (score: [X])
#2: [option] (score: [Y]) - gap: [X-Y]
#3: [option] (score: [Z]) - gap: [X-Z]

Close race: [YES if gap < 10% / NO]
Key differentiators: [list]
Red flags: [list or "none"]

Step 2: Feasibility Assessment

Check top options against practical considerations.

LLM Feasibility Assessment Protocol:

For each top candidate (usually top 3), assess using available information:

2.1: Resource Requirements

LLM Execution:

For resource assessment:
1. Scan option description for: cost, price, budget, hours, days, team, equipment
2. Scan context for: available budget, current skills, timeline mentioned
3. Compare: requirements vs. stated availability
4. Flag gaps: What's required but not confirmed available?

2.2: Risk Profile

LLM Execution:

Risk assessment pattern:
1. Identify option category (financial, career, relationship, technical)
2. Apply domain-typical failure modes:
   - Financial: loss, fraud, market change
   - Career: mismatch, burnout, opportunity cost
   - Relationship: conflict, loss, misunderstanding
   - Technical: failure, delay, scope creep
3. Rate each failure mode: P(occurs) × Impact

2.3: Stakeholder Considerations

LLM Execution:

Stakeholder assessment:
1. Explicit stakeholders: names mentioned in input
2. Implicit stakeholders (by option type):
   - Career: employer, family, colleagues
   - Financial: partners, dependents, creditors
   - Technical: users, team, downstream systems
3. For each: What do they want? How does this option affect that?

2.4: Timing Considerations

Timing Assessment Logic:

NOW indicators: urgent, critical, deadline imminent, opportunity expiring
SOON indicators: approaching deadline, dependencies resolving, building pressure
LATER indicators: not ready, prerequisites unmet, other priorities higher
FLEXIBLE indicators: no deadline, reversible, low stakes

Output format:

FEASIBILITY ASSESSMENT: [Option Name]
=====================================
Resource requirements:
- Skills: [required / assumed / USER_VERIFY]
- Resources: [required / assumed / USER_VERIFY]
- Time: [required / assumed / USER_VERIFY]

Risk profile:
- Potential failures: [list]
- Reversibility: [REVERSIBLE / HARD / IRREVERSIBLE]
- Downside: [LOW / MED / HIGH / CATASTROPHIC]

Stakeholders:
- Affected: [list]
- Concerns: [list]
- Approvals: [list or USER_VERIFY]

Timing:
- Prerequisites: [list]
- Constraints: [list]
- Readiness: [NOW / SOON / LATER]

FEASIBILITY VERDICT: [PROCEED / CAUTION / BLOCKED / USER_VERIFY]
Blockers: [list if any]

Step 3: Consider Selection Mode

Based on context, determine selection approach.

Selection modes:

Decision logic:

If top options very close (gap < 10%) AND compatible:
  → Consider MULTIPLE (portfolio approach)

If high uncertainty about key factors:
  → Consider CONDITIONAL (with clear triggers)

If clear winner AND commitment appropriate:
  → SINGLE

Output format:

SELECTION MODE
==============
Mode: [SINGLE / MULTIPLE / CONDITIONAL]
Rationale: [why this mode]

Step 4: Make the Selection

Execute the selection based on mode.

For SINGLE mode:

SELECTION: [option name]
========================
Primary: [option]
Backup: [second option]

Why primary over backup: [key differentiators]

For MULTIPLE mode:

SELECTION: Portfolio Approach
=============================
Options selected: [list in priority order]
- #1: [option] - primary focus
- #2: [option] - secondary
- ...

Resource allocation: [how to divide effort]

For CONDITIONAL mode:

SELECTION: Conditional Approach
===============================
Primary: [option]

Fallback 1: [option]
  Trigger: [what would cause switch]

Fallback 2: [option]
  Trigger: [what would cause switch]

Decision points: [when to evaluate]

IMPORTANT: If top-ranked option is NOT selected, document why:

NOTE: Top-ranked [option] not selected because: [reason]
- Feasibility concern: [specific issue]
- Risk concern: [specific issue]

Step 5: Document Rationale

Create clear documentation of the decision.

Documentation elements:

1. What was selected and why
2. What alternatives were considered
3. Why alternatives were not chosen
4. Key assumptions underlying the choice
5. What would make us reconsider

Output format:

DECISION RATIONALE
==================
Selected: [option(s)]

Why selected:
- [reason 1]
- [reason 2]

Alternatives considered:
- [option]: Not selected because [reason]
- [option]: Not selected because [reason]

Key assumptions:
- [assumption 1]: If wrong, would [impact]
- [assumption 2]: If wrong, would [impact]

Reconsider if:
- [condition 1]
- [condition 2]

Step 6: Define Success Criteria and Reversal Triggers

Establish what happens after selection.

Success Criteria

How to know the selection was good:

SUCCESS CRITERIA
================
Short-term (by [date]):
- [ ] [measurable indicator]
- [ ] [measurable indicator]

Medium-term (by [date]):
- [ ] [measurable indicator]

Validation complete when: [condition]

Reversal Triggers

What conditions would cause a switch:

REVERSAL TRIGGERS
=================
Switch to backup if:
- [ ] [condition] occurs
- [ ] [metric] drops below [threshold]
- [ ] [event] happens

Point of no return: [when commitment becomes irreversible]
Commitment duration: [how long before reconsidering]

Implementation Notes

What implementer needs to know:

IMPLEMENTATION NOTES
====================
Critical first steps:
1. [step]
2. [step]

Dependencies:
- [dependency]

Risks to monitor:
- [risk]

When to Use

• After comparison has produced ranked options
• Making final commitment to course of action
• Strategy selection after analysis complete
• Multiple stakeholders need alignment
• Before resource allocation
• Decision needs documentation

Verification Criteria

Overall verification:

• Selection is from ranked options (not invented)
• Feasibility assessment completed (not just rank accepted)
• Rationale documented and understandable
• Backup identified (unless single option scenario)
• Reversal triggers defined
• USER_VERIFY items flagged for user

Integration Points

• Often invoked from: /comparison (after ranking), /multi_criteria_decision, /procedure_engine
• Routes to: /steps_generation (to plan execution), /project_initiation (for larger initiatives)
• Related: /comparison, /risk_assessment, /criteria_weighting