Normalized Telegram record
Retrieval and long context do not create company memory
Finding a relevant fragment does not establish whether it is current, authorized, trustworthy, or applicable to the present task.
Signal contract
- Finding a relevant fragment does not establish whether it is current, authorized, trustworthy, or applicable to the present task.
- Organizational memory needs provenance, scope, temporal validity, permissions, and an explicit connection to changed action, not only better search.
- Novelty: structural. Verification: source-inspected.
Source ledger
Publishable sources attached to this record.
| # | Source | Role | Public status |
|---|---|---|---|
| 1 | arxiv.orgpaper | primary receipt | source_urls |
| 2 | aclanthology.orgsource | supporting receipt | source_urls |
Observation
Finding a relevant fragment does not establish whether it is current, authorized, trustworthy, or applicable to the present task.
Why it matters
Organizational memory needs provenance, scope, temporal validity, permissions, and an explicit connection to changed action, not only better search.
Entities
RAG, long-context models
Provenance
This public record is an English normalization of QWG AI Telegram message 2695. The complete original-language post remains the canonical raw message.