01
Bounded offline decision systems
A research direction for replacing narrow, repeated model calls with evolved local decision systems. The source distinguishes rule-based pilots from the proposed neural architecture.
MOBCORP / NATIVE RESEARCH
Our own questions.
Our own evidence.
Original project research, with its limits intact. Begin with two reviewed historical excerpts from the GOFAINEAT programme.
Proposals, not established results. These are excerpts of existing internal project documents, prepared for public reading. No full original papers, independent proofs or validated performance results are published here. Historical implementation claims are not current certifications.
Reviewed documents
02
Generalization is the research risk
The early feasibility study isolates held-out generalization and reviewed training data as key uncertainties, then scopes a narrow pilot instead of assuming a general system works.
Provenance and coverage
Project owner: John Alexander Mobley. No individual writing byline has been established for these project documents. Each excerpt records its source filename, source date, SHA-256 identity and editorial limitations. Internal implementation references are omitted.
First review batch: six distinct documents read in full; two eligible for public excerpts, four withheld after review. Withheld documents contain private or client-specific examples, internal operational notes, unverified claims or substantive reasoning errors. The broader native collection remains under review; file counts and duplicates are not counts of original research papers.
Public provenance catalogue JSON · OpenAI mathematics library