Domain Research & 360° Grades | JR’S

THE SCALE OF THE WORK

NOT A HANDFUL OF NAMES. AN ENTIRE RESEARCH LANDSCAPE.

JR’S combines registry-derived observations, lexical research, reported transactions and direct human judgments. Each source answers a different question. None answers all of them.

A frozen full-zone .com research extract contained 20,609,488 rows. A map of how words appear together—not 20.6 million sales.

Nearly two million stored appraisal records gave us a large operational landscape to inspect, re-grade and challenge.

Our sales database spans multiple domain extensions and venues. Reported transactions are evidence, not a complete market census.

A documented lexical reference for studying language, identity and construction. Knowing a word exists is not knowing its value.

Research and database snapshot: October 6, 2026. Counts describe different datasets and must not be added together as a total of unique domains. Approximate database counts are marked with “~”; the full-zone extract and dictionary are documented research artifacts.

ACROSS THE DOMAIN SPACE

SIX QUESTIONS. NO SINGLE SHORTCUT.

THE REGISTERED .COM LANDSCAPE

We studied word combinations found in a full-zone snapshot, then widened the language coverage. The resulting research vocabulary grew from 9,392 to 27,140 words. That lets us examine more ordinary constructions without treating unfamiliar words as automatically bad.

A registration is evidence of a choice. It is not proof of quality, a sale or years of survival.

A MARKET THAT CHANGES EVERY DAY

Dated registration and drop observations let us study demand, turnover and persistence separately from a static snapshot. Our database holds approximately 3.47 million dated word-pair rows. Repeated observations describe change; they are not all different names.

A historical test must use only evidence available before the event it is testing.

LANGUAGE, BRANDS AND COMMERCIAL FIT

We examined frequency, pronunciation, word relationships, business usage and name structure. Single words, compounds and coined brands need different readings. Short is not automatically strong; an uncommon word is not automatically weak.

Word familiarity, coherent meaning and commercial usefulness are related—but not interchangeable.

HUMAN JUDGMENT, PRESERVED

We retained 2,321 direct human word-pair judgments, separate from generated labels and sale-derived evidence. Human supervision helped test whether a model was learning useful combinations instead of simply rewarding popular words.

Personal judgments are valuable training evidence, not universal truth about every domain.