Audience
Data researchers, post-training teams, infra leads
Chris Royse field notes
Why One Input Can Produce Up To 91 Signals
The paper's core accounting move: N embedders create N single-lens signals plus pairwise interactions from the same fixed input.
Signal / PAPER / 11:49
Core idea
At N=13, the constructive ceiling is N plus C(N,2), or 91 structured signals per input before measuring redundancy.
DDA Counting Identity
Watch on YouTube· 11:49
Why One Input Can Produce Up To 91 Signals
The exact effective count still needs mutual-information auditing, but the direction matters: fixed data can be decomposed before it is replaced.
Watch videoOpen the full video on YouTubeWhat to take from it
The videos are raw build context. These notes translate them into the shortest useful frame for creators, companies, and AI lab readers.
The 91 number is a ceiling, not a measured realized floor.
The audit question is pairwise redundancy.
The business question is whether the extra signals help a real task.
Continue this thread.
Related notes stay inside the same problem area first, then move to the next useful context.
Watch + read / 11:23
The Data Wall Is A Meaning Extraction Problem
Why frontier labs should look for more signal inside existing data before defaulting to synthetic data loops.
Watch + read / 11:34
The 13-Embedder Context Graph
Semantic, temporal, causal, code, graph, typo-tolerant, paraphrase, entity, and late-interaction lenses in one memory system.
Watch + read / 9:25
Teleological Constellation Training In Plain English
A target identity or style can be defined as frozen centroid vectors, then checked at generation time instead of trusted by vibe.
Make it concrete.
Send the audience, data type, target task, proof bar, and sharing limits.