← Living Museum · 100 × 26 Census · Name Lab
Is a Name a Compressed Code of a Life?
A popular claim becomes scientifically useful only when we specify what it predicts, prevent retrospective storytelling, compare against baselines, and allow the claim to fail.
STATUS · PILOTS 1–2 COMPLETE / TWO SIMPLE NAME-CODE TESTS NOT SUPPORTED
Claim Passport
A person's name contains information predictive of later life trajectory or contribution.
Any apparent correspondence is explained by chance, flexible retrospective interpretation, culture/language, demographic signals or other confounders.
Name-derived features add reproducible predictive information beyond preregistered baselines and background variables.
If name-only predictions do not beat baselines under blind evaluation, the tested version is not supported.
Existing Name Space · already built
Life Space · keep independent
Experiment pipeline
Freeze hypotheses, features, outcomes, baselines and scoring rules before reveal.
Hide identity/life information from the name-side prediction process.
Generate fixed predictions from name features only.
Timestamp and preserve predictions before outcomes are revealed.
Compare with predefined Life Space outcomes and baseline models.
Shuffle names among citizens and repeat the same pipeline.
Test whether the interpretive method still produces convincing stories after names are deliberately reassigned.
Remove culture/language, name length, phonetic, etymological and demographic cues in separate analyses.
Repeat on a new population and, separately, a Chinese-character representation.
What counts as evidence?
| Observation | Interpretation |
|---|---|
| Name-only model ≈ random / baseline | Tested compressed-code hypothesis not supported. |
| Effect disappears after controlling culture/language | Likely representation/cultural signal, not an independent name-code effect. |
| Retrospective stories work equally after shuffled names | Strong evidence of narrative flexibility / post-hoc fitting. |
| Preregistered name features predict held-out outcomes above controls | Interesting association; requires mechanism testing and independent replication. |
| Effect replicates across representations/populations | Stronger evidence, but still not proof of mystical or deterministic causation. |
Designed interpretation stress test
Example: ALAN TURING can be retrospectively read as Attention → Link → Network → Test → Relation → Information → Generate. Great Experiment 001 asks a harder question:
Could the same rule have predicted anything useful before we knew it was Alan Turing—and would it still look persuasive if “ALAN TURING” were assigned to a different citizen?
Planned replications
Objective letter structure first. Etymology and phonetics are separate feature families.
Character, documented meaning, pronunciation, homophone family and naming history must remain distinct layers.
Ask whether any effect survives when the same person is represented differently.