After reducing the candidate pool, early development should compare several base formulas in parallel and expand only promising directions instead of repeatedly changing one sample in place.
Branching consumes more samples and attention; a consumer protocol needs a small fixed design rather than the dozens or hundreds of trials feasible in a commercial laboratory.
Parallel candidates preserve alternatives and make tradeoffs visible. Elliott’s process separates candidate reduction from formula branching, then expands successful directions into additional blind-evaluated variants. Consumer work can use fewer branches while retaining the same decision structure.
Strengthened with the Four Roses / Brent Elliott source cluster on 2026-08-29.
A Zettel is one durable idea in your own words—not a quotation, source summary, or draft-book claim.
Atomic idea
State one complete idea. It should be understandable on its own and specific enough to connect to other notes.
Reasoning
Explain why the idea is true or useful in Rationale and Implications. Record meaningful objections or exceptions in Counterpoints.
Provenance
Connect one or more Literature Notes. Direct source relations are legacy-only; trace evidence through the literature-note and excerpt chain.
Connections
Apply approved Tags for retrieval and one or more Themes only when the note genuinely advances those interpretive threads.