Part 6. Practice G. Before-and-After Comparison Template

This practice template records the utility of librarying without exaggeration. The most dangerous part of a white paper is making results look impressive by mixing measured values with estimates. Therefore, label every figure as measured / estimated / illustrative example.

Measurement principles

First, state the conditions together. Time is meaningful only when the record also says whether the basis is 1,500 headwords, how many input documents there were, how many people worked, and how many AI collaborators ran in parallel.

Second, keep the before-and-after baseline identical. A comparison is meaningless when the scopes of the existing method and the librarying method differ. For example, “rough setting confirmation” in the old method and a “validated Vault” in the librarying method are not the same task.

Third, include quality metrics. Reducing time alone is insufficient. Include metrics that show operational quality, such as omission rate, link errors, inconsistent translation notation, and onboarding time for new workers.

Comparison record template

# {Project Name}: Before-and-After Librarying Comparison

Date:
Author:
Figure type: measured / estimated / illustrative example

## 1. Measurement conditions

| Item | Value |
|---|---|
| IP type |  |
| Number of input documents |  |
| Total input volume |  |
| Number of Stage-1 rows |  |
| Number of Stage-2 headwords |  |
| Number of Vault files |  |
| Number of hubs |  |
| Number of workers |  |
| Number of parallel AI collaborators |  |
| Validation standard |  |

## 2. Time comparison

| Task | Existing method | Librarying method | Figure type | Notes |
|---|---:|---:|---|---|
| Review input documents |  |  |  |  |
| Extract keywords |  |  |  |  |
| Consolidate synonyms |  |  |  |  |
| Create Vault skeleton |  |  |  |  |
| Write descriptions |  |  |  |  |
| Validate |  |  |  |  |
| Total |  |  |  |  |

## 3. Quality comparison

| Metric | Existing method | Librarying method | Measurement method | Notes |
|---|---:|---:|---|---|
| Inconsistent official notation |  |  | Sample-check manuscripts/setting documents |  |
| Inconsistent translated notation |  |  | Sample-check translated manuscripts |  |
| Average time to confirm a setting |  |  | Work log or user test |  |
| Onboarding time for a new worker |  |  | Perform a task after reading the hub list |  |
| Broken links |  |  | Automated check |  |
| Unspecified worldbuilding-axis rate |  |  | Aggregate Stage-2 CSV |  |
| Unwritten LLM rate |  |  | Inspect Vault blocks |  |

## 4. Notes on operational effects

- What improved:
- Remaining bottlenecks:
- Problems solved by automation:
- Problems that still require human judgment:
- Improvements for the next cycle:

## 5. Publication eligibility

- [ ] The real project name may be disclosed
- [ ] Figures may be disclosed
- [ ] Headword samples may be disclosed
- [ ] Original-text sources may be disclosed
- [ ] Processing/anonymization is required

Illustrative example

The table below is an illustrative example for understanding the format, based on “Land of Nubi.” When including it in a published white paper, replace it with measured values or clearly retain the illustrative-example label.

Metric	Existing method	Librarying method	Type
Organizing 1,500 headwords for “Land of Nubi”	about 3–6 months	about 9–16 hours	Illustrative example; based on the main-text time budget
Canonical-keyword duplicates	Not measurable	Target: 0	Methodology standard
Broken links	Not measurable	Target: 0	Validation standard
Grasping the core of “Land of Nubi” for a new worker	1–3 months	Prioritize learning 100–150 hubs	Illustrative example; requires project-specific measurement

Interpretation standard

Librarying efficiency does not simply mean “faster.” More precisely, it reduces verification costs and error-recovery costs within a repeatable procedure. Therefore, a before-and-after comparison should look beyond time saved and include the cost of handling the same IP again next month and next year.