Skip to main content
Ship Name Lab

Human-reviewed research

Name Pair Case Studies

Eight deliberately different pairs, each reviewed beyond its numeric score. These pages preserve the engine output, expose what it misses, and give a repeatable human decision process.

Last reviewed 2026-07-12

Why these are separate pages

The public benchmark answers whether the engine behaves consistently. A case study asks the harder question: should a person actually use the result? Each review below identifies a different boundary—pronunciation, short inputs, punctuation, diacritics, non-Latin text, or source choice.

Taylor + Travis: When a Compact Blend Still Shows Both Sources

A worked review of Tayvis, including why its structural score is strong and what the score cannot establish.

Read the worked review →

Chloe + Mason: Why a High Score Cannot Settle Pronunciation

A case study of Masloe and the gap between visible character structure and a natural spoken name.

Read the worked review →

Bo + Jo: The Hard Limit of Two-Letter Name Blending

A worked example showing why very short inputs leave almost no meaningful blend space.

Read the worked review →

Ann + Lee: When Letter Overlap Improves Score but Hides Meaning

A review of Anee and the difference between a compact written form and a recognizable pairing label.

Read the worked review →

Mary-Jane + O'Connor: What Normalization Removes

A case study of punctuation normalization, compound names, and choosing the right source units before blending.

Read the worked review →

Zoë + Chloé: Preserving Diacritics Without Claiming Pronunciation

A transparent review of Zoloé and why Unicode support is not the same as language competence.

Read the worked review →

小明 + 小红: Why Character Slicing Is Not Language Understanding

A non-Latin benchmark case that distinguishes deterministic character handling from linguistic validation.

Read the worked review →

Elizabeth + Jonathan: Choosing Source Forms Before Scoring

A long-name case study showing why nickname selection can matter more than the final ranking algorithm.

Read the worked review →

Reproduce the software result

Every case points back to an input in the public Name Pair Benchmark. The editorial verdict is deliberately kept separate from the engine's pass threshold.