Skip to main content
Ship Name Lab
Ship Name Lab

Structured, privacy-minimized product research

Human Review Study for Generated Pair Names

Software can verify that a result contains fragments from two inputs. It cannot decide whether a person recognizes the source, can read the result aloud, or would keep it as a draft. This fixed-sample study records those three human judgments separately.

Fixed cases
10
Protocol
2026.07.29-1
Engine
2026.07.29-1
Retention
180 days

Why this study exists

The public engine regression checks are maintained by this project. They are useful for catching code changes, but a passing self-test is not independent evidence that a generated candidate is understandable or useful to another person. The review below starts with a blind source-choice question before revealing the fixed input pair.

Readability and “keep for review” are deliberately subjective. A keep response means only that the reviewer would retain the candidate for a later discussion. It does not establish cultural suitability, pronunciation in every language, consent, trademark status, or a recommendation to use the name.

Fixed sample · one response per case and browser

Review one generated candidate

0/10 reviewed here

This browser reviewed every case

Thank you. The completion list stays only in this browser. Please do not clear it to submit the same cases again.

Protocol and safeguards

  1. One of ten fixed candidates is selected from cases this browser has not reviewed.
  2. The reviewer chooses one of four source pairs before the fixed source is revealed after the full response is submitted.
  3. The reviewer separately rates reading clarity and whether to keep or reject the draft.
  4. A local completion list discourages accidental repeats, and obvious crawlers are excluded from public totals.
  5. Per-case results are withheld until at least five browser reviews exist for that case.

What is and is not collected

The study sends a random review ID, fixed case ID, chosen option, readability choice, keep/reject choice, protocol and engine versions, broad referrer category, and a broad browser/bot classification. The application database does not store an entered name, generated free text, email address, raw referrer URL, or IP address for this study.

The browser safeguard is not identity verification, so these results cannot be treated as a representative public-opinion poll. No demographic or linguistic-population claim is made. Read the privacy policy for the complete data-flow description.

Publication is not validation

Aggregate percentages describe only the responses received under this protocol. They are not scientific validation. Internally, this project will not treat the collection as a monetization-readiness signal until it has at least 100 genuine browser reviews, all ten cases have at least ten reviews, and responses span at least 14 active days. Traffic quality and abnormal response velocity must also be reviewed.

The name-pair limits report remains the controlling guidance for individual use.