Vertical vs Horizontal Integration Decision

Input: $ARGUMENTS

Overview

Framework for deciding when to improve existing procedures/categories (vertical) versus when to add new procedures/categories (horizontal) in GOSM.

This applies to any system where you must choose between depth and breadth: codebases, product features, skill development, organizational capabilities.

Steps

Step 1: Assess Current State

Vertical health (depth/quality of existing):

Factor	Score (1-5)	Evidence
Procedure quality: Are existing procedures well-developed?
Coverage within categories: Do procedures handle their intended cases?
Integration: Do procedures work well together?
Validation: Have procedures been tested and refined?
User satisfaction: Do users find existing procedures helpful?

Horizontal health (breadth/coverage):

Factor	Score (1-5)	Evidence
Domain coverage: Are all important domains represented?
Gap frequency: How often do users need something that doesn’t exist?
Request patterns: What do users ask for that we can’t provide?
Competitive coverage: What do alternatives offer that we don’t?
Structural completeness: Is the taxonomy well-covered?

Step 2: Calculate Direction Score

Vertical signal strength (reasons to go deep):

Existing procedures are frequently used but produce mediocre results → +3
Users report confusion or errors with current procedures → +3
Quality issues are causing trust problems → +3
Integration between procedures is poor → +2
Procedures have known gaps within their scope → +2

Horizontal signal strength (reasons to go wide):

Users frequently need capabilities that don’t exist → +3
Entire domains are unrepresented → +3
Low-hanging fruit: easy new procedures would add high value → +2
Structural gaps create routing problems → +2
New domains have emerged since last expansion → +2

Step 3: Consider Portfolio Balance

Map current state on the depth-breadth matrix:

          HIGH QUALITY
              |
  "Polished    |    "Complete
   but narrow" |     system"
              |
LOW BREADTH ---+--- HIGH BREADTH
              |
  "Bare        |    "Wide but
   minimum"    |     shallow"
              |
          LOW QUALITY

If “bare minimum”: go horizontal (you need basics before depth)
If “polished but narrow”: go horizontal (diminishing returns on depth)
If “wide but shallow”: go vertical (breadth without depth is fragile)
If “complete system”: go vertical on weak spots or maintain

Step 4: Apply Decision Rules

Condition	Decision	Reasoning
Vertical score > Horizontal score + 5	Go vertical	Strong depth signal
Horizontal score > Vertical score + 5	Go horizontal	Strong breadth signal
Scores within 5	Check portfolio position	Tiebreaker from Step 3
Both scores low	Neither — something else is wrong	Investigate root cause
Both scores high	Do both — but separate tracks	Parallel investment

Step 5: Plan the Work

If going vertical:

Rank existing procedures by: usage × quality_gap
Start with highest-impact improvements
Define “done” for each improvement (not “make it better” — specific criteria)
Set a stopping criterion (switch to horizontal when quality reaches threshold)

If going horizontal:

Rank gaps by: frequency_of_need × difficulty_to_create
Start with highest-value, lowest-effort additions
Define minimum viable quality for new additions (avoid creating more stubs)
Set a stopping criterion (switch to vertical when coverage reaches threshold)

Step 6: Report

VERTICAL vs HORIZONTAL DECISION:
System: [what system is being developed]

Assessment:
- Vertical score: [N] — key signals: [top factors]
- Horizontal score: [N] — key signals: [top factors]
- Portfolio position: [which quadrant]

Decision: [VERTICAL / HORIZONTAL / BOTH]
Rationale: [why]

Work plan:
1. [specific item to improve/add]
2. [specific item]
3. [specific item]

Stopping criterion: Switch to [other direction] when [condition]

When to Use

Deciding what to work on next in library/system development
Planning a development sprint
Evaluating whether to improve existing or add new
At portfolio review points to assess balance
→ INVOKE: /iterate for executing the chosen direction

Verification

Both vertical and horizontal health assessed with evidence
Signal strengths scored independently
Portfolio balance considered
Decision follows from assessment (not predetermined)
Work plan has specific items and stopping criteria

vhd