Source Prioritization

Overview

Given limited time to extract procedures from sources, prioritize which sources to process for maximum procedure value.

Steps

Step 1: Gather source metadata

For each candidate source, collect or estimate:

1. Basic information:

   • Title, creator, type, URL
   • Length (duration/pages)
   • Publication date
   • Initial reason for adding to queue

2. Quick assessment (don’t deep-dive yet):

   • Skim description/abstract/table of contents
   • Check creator credentials
   • Note topic areas covered
   • Estimate procedure density (HIGH/MEDIUM/LOW)

3. Volume metrics:

   • Total videos/items (for channels)
   • Total hours of content
   • Estimated transcript characters (hours x 9000)

Create standardized metadata for each source.

Step 2: Score procedure density

Rate how rich each source is in extractable procedures (1-5):

5 - Almost entirely procedural (tutorials, courses, how-tos) 4 - Mostly procedural with some context 3 - Mix of procedural and informational 2 - Mostly informational with some procedures 1 - Almost entirely informational/entertainment

High density signals:

• Tutorial channels, course creators
• “How to” in most titles
• Step-by-step format common
• Practical demonstrations

Low density signals:

• News/commentary
• Entertainment focus
• Opinion/reaction content
• Abstract theory without application

Step 3: Score uniqueness

Assess whether this source has knowledge unavailable elsewhere (1-5):

5 - Completely unique - only source for this knowledge 4 - Rare - few others have this perspective/access 3 - Somewhat unique - different angle on common topics 2 - Common - similar to many other sources 1 - Commodity - same as everyone else

High uniqueness signals:

• Original research/data
• Unique professional access
• Contrarian successful approaches
• Proprietary methods
• Decades of specialized experience

Low uniqueness signals:

• Summarizes others’ work
• Common knowledge packaged
• Follows trends
• No original insight

Step 4: Score relevance

Assess how relevant to YOUR specific goals (1-5):

5 - Directly applicable to current goals 4 - Highly relevant to goals 3 - Moderately relevant 2 - Tangentially relevant 1 - Interesting but not relevant

NOTE: This is PERSONAL - depends on what you’re trying to achieve. Cross-reference with extraction_goals and library_gaps.

Step 5: Score credibility

Rate creator’s track record of producing results (1-5):

5 - Proven exceptional results, widely recognized 4 - Strong track record, credible in field 3 - Some evidence of results 2 - Claims results but limited evidence 1 - No evidence, unverified claims

Credibility signals:

• Verifiable achievements
• Peer recognition
• Students/followers with results
• Published/cited work
• Professional credentials used

Low credibility signals:

• Only self-reported success
• No verifiable outcomes
• Sells without substance
• Contradicted by evidence

Step 6: Score extractability

Rate how easy it is to extract procedures (1-5, higher = easier):

5 - Very easy - clear explanations, transcripts, structured 4 - Easy - good explanations, some structure 3 - Moderate - requires inference 2 - Hard - implicit, scattered, unstructured 1 - Very hard - no transcripts, unclear, heavily visual

Factors that improve extractability:

• Transcripts available
• Clear step-by-step explanations
• Written summaries/notes
• Consistent format
• Explicit about methods

Factors that reduce extractability:

• No transcripts, heavy accent
• Visual demonstrations without explanation
• Scattered across many videos
• Personality-heavy, procedure-light
• Assumes high prior knowledge

Step 7: Calculate composite scores and ROI

Calculate weighted score for each source:

raw_score = (procedure_density x 0.25) + (uniqueness x 0.25) + (relevance x 0.20) + (credibility x 0.15) + (extractability x 0.15)

Then estimate extraction metrics:

estimated_procedures = content_hours x density_factor

• density_factor: score_5=3.0, score_4=2.0, score_3=1.0, score_2=0.5, score_1=0.2

unique_procedures = estimated_procedures x (uniqueness / 5)

extraction_value = unique_procedures x (relevance / 5)

extraction_hours = content_hours x extraction_multiplier

• manual_deep: 3.0 hours per 1 hour content
• manual_quick: 1.5 hours per 1 hour content
• automated_review: 0.5 hours per 1 hour content

adjusted_cost = extraction_hours / extractability_score

ROI = extraction_value / adjusted_cost

ROI interpretation:

• excellent: > 2.0 procedures per hour
• good: 1.0 - 2.0 procedures per hour
• acceptable: 0.5 - 1.0 procedures per hour
• poor: < 0.5 procedures per hour

Step 8: Assign tiers

Assign each source to a tier based on scores:

TIER 1 - EXTRACT NOW:

• Criteria: ROI > 2.0 OR (raw_score > 4.0 AND relevance = 5)
• Action: Extract immediately, high priority

TIER 2 - EXTRACT SOON:

• Criteria: ROI 1.0-2.0 OR raw_score 3.5-4.0
• Action: Extract when Tier 1 complete

TIER 3 - EXTRACT LATER:

• Criteria: ROI 0.5-1.0 OR raw_score 3.0-3.5
• Action: Extract if time permits

TIER 4 - MAYBE:

• Criteria: ROI 0.25-0.5 OR raw_score 2.5-3.0
• Action: Only if specifically needed

TIER 5 - SKIP:

• Criteria: ROI < 0.25 OR raw_score < 2.5
• Action: Do not extract

Special considerations:

• Foundational sources: bump to Tier 1 if Tier 2-3
• Time-sensitive: bump priority if may become unavailable
• Bundled value: extract early if unlocks other sources
• Diminishing returns: after 10 procedures from one source, reassess

Step 9: Create extraction schedule

Build ordered extraction schedule:

1. Within Tier 1, sequence by:

   • Foundational sources first
   • Quick wins early (high ROI, low volume)
   • Batch similar sources for efficiency

2. Allocate time based on budget:

   • Sum hours needed for each tier
   • If budget limited, may not reach lower tiers
   • Leave buffer for unexpected difficulty

3. Set checkpoints:

   • After every 10 procedures extracted
   • After completing each source
   • Weekly review of priorities

4. Create schedule document with:

   • Rank, source name, ROI, hours, expected procedures
   • Cumulative totals
   • Summary statistics

Step 10: Document skip list

For Tier 5 sources and any dropped from lower tiers:

1. Record why skipped:

   • Below ROI threshold
   • Insufficient time
   • Redundant with existing library
   • Low credibility

2. Decide retention:

   • KEEP FOR LATER: Still valuable, review next quarter
   • DROP: Remove from queue entirely
   • CONDITIONAL: Keep if specific need arises

3. Create skip list table: | Source | Raw Score | ROI | Reason | Disposition |

When to Use

• When you have multiple sources waiting for extraction
• Before starting a procedure extraction sprint
• When building extraction queue from content backlog
• When deciding whether to process new source vs queued sources
• During quarterly library planning
• When extraction time is limited and must be optimized
• After accumulating “watch later” or “read later” items
• When onboarding to GOSM and choosing initial extraction targets

Verification

• All sources scored on all dimensions with evidence
• Weights reflect current priorities
• ROI calculation is consistent across sources
• Top priorities align with actual needs
• Time allocation is realistic and complete
• Skipped sources have clear rationale

Input: $ARGUMENTS

Apply this procedure to the input provided.