Metric System Index
Metric System Index
LDI
LDI
LDI measures whether the vocabulary engine is broad, sustained, and disciplined without using entropy in the formula.
LDI measures whether the vocabulary engine is broad, sustained, and disciplined without using entropy in the formula.
Lexical discipline: tracks vocabulary breadth and word-choice control outside the entropy formula.
Lexical discipline: tracks vocabulary breadth and word-choice control outside the entropy formula.
Formula
Formula
Mean: average of lexical diversity, MTLD, inverted Yule’s K, and inverted Maas. Balance: weakest component.
Mean: average of lexical diversity, MTLD, inverted Yule’s K, and inverted Maas. Balance: weakest component.
Signals: lexical diversity, MTLD, repetition control, and lexical concentration control.
Signals: lexical diversity, MTLD, repetition control, and lexical concentration control.
What It Means
What It Means
LDI tests lexical discipline. It shows whether a segment has broad vocabulary behavior that stays sustained over length while keeping repetition and lexical narrowness under control.
LDI tests lexical discipline. It shows whether a segment has broad vocabulary behavior that stays sustained over length while keeping repetition and lexical narrowness under control.
Scoring Context
Scoring Context
Scores are standardized within the declared comparison pool, tokenizer, transform, and slice-size run before index calculation. Index values are therefore run-context-specific and should not be interpreted outside the stated configuration.
Scores are standardized within the declared comparison pool, tokenizer, transform, and slice-size run before index calculation. Index values are therefore run-context-specific and should not be interpreted outside the stated configuration.
Evidence Boundary
Evidence Boundary
The public LDI page describes aggregate metric behavior. It does not publish protected corpus text, third-party lyrics, artist names, album titles, song titles, or source-reconstructable mappings.
This page supports a metric-specific aggregate claim: LDI can show whether vocabulary range stays broad, sustained, and controlled without using entropy in the formula. It does not publish source text, identify protected reference material, or prove universal artistic ranking. The public result is strongest inside the tested comparison pool, and deeper source-level verification belongs in controlled review.
The public LDI page describes aggregate metric behavior. It does not publish protected corpus text, third-party lyrics, artist names, album titles, song titles, or source-reconstructable mappings.
This page supports a metric-specific aggregate claim: LDI can show whether vocabulary range stays broad, sustained, and controlled without using entropy in the formula. It does not publish source text, identify protected reference material, or prove universal artistic ranking. The public result is strongest inside the tested comparison pool, and deeper source-level verification belongs in controlled review.
Evidence Charts
Evidence Charts
Start with the LDI leaderboard to see the direct lexical-discipline ranking. Then use the slice-size trend to check whether that ranking is stable across measurement windows.
Start with the LDI leaderboard to see the direct lexical-discipline ranking. Then use the slice-size trend to check whether that ranking is stable across measurement windows.
LDI leaderboard
LDI leaderboard
Ranks the tested segments by LDI, the lexical-discipline lens.
Ranks the tested segments by LDI, the lexical-discipline lens.
LDI slice-size trend across top segments
LDI slice-size trend across top segments
Tracks LDI across slice sizes for the top segments.
Tracks LDI across slice sizes for the top segments.
EURE / LDI / RACS leaderboard comparison
EURE / LDI / RACS leaderboard comparison
Compares EURE, LDI, and RACS without merging the three lenses.
Compares EURE, LDI, and RACS without merging the three lenses.
Reference Comparison Boundary
Reference comparisons are baseline coordinates for interpreting metric behavior. They are not presented as artistic rankings or claims of superiority over other writers.