Human-Routed Handoffs in ChatGPT: A Possibly More Token-Efficient Procedure for Cross-Surface Work

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

This paper proposes a human-routed procedure for maintaining continuity across ChatGPT conversations and work surfaces.Derived from developing papers and a screenplay, it separates two objects: a Dump preserves relevant discussion; a Handoff directs the next operation. Verbatim and Summary Dumps offer different levels of fidelity and compression.Two controls govern the procedure: Authority Space bounds jurisdiction; Goodwill bounds optimization after sufficiency. Handoffs must remain checkable, with review proportionate to consequences and the object’s validity conditions.The procedure may reduce reconstruction and token use while preserving human judgment and responsibility. These benefits are practical observations rather than quantified or universal claims. The proposal is offered for adaptation, testing, correction, or rejection.