First Prototype Scorecard
A scorecard of recorded runs places the first prototype’s requirements in one operating path. Nine of eleven requirement rows are met within stated limits, two are partly met, and none is unmet; counting partial rows as half gives 91%. Every infrastructure item except the deferred printed-circuit-board variant is built.
SCOPE + LIMITATIONS
- This is a summary of recorded runs, not a new experiment. Several rows rest on development data or one regime, as stated in their own reports. Parallel Context is counted by its written criteria, which ask for computational use; neural use is not built.
- This public edition reviews saved evidence; experiments were not rerun for publication.
CONTROL COMPARISON
Score each first-prototype requirement against the evidence of passed, locked gates only. The operating path is the demonstration engine: intake and moderation, reading through exact parsers or the learned reader, the qualified Stem Decision, the Mode Decision, the three stems through the staged Context path, the independent checker, memory layers, and a per-answer listing of every executed component with its size. No new run is made for the scorecard.
The internal bar is at least 90% of results acceptable and verifiable. By strict count the prototype is at 82%; the remaining gaps are the learned Mode Decision’s last 1.3 points and a margin for self-designed research cases. A single locked run that scores every requirement at once has not been made.
- Met within limits
- Composition, linked Decision weights, learning during operation, self-chosen learning scopes, transfer from a person or Source Models, feedback, memory layers, Parallel Context in the operating path, selective loading
- Partly met
- Output-mode selection (learned Decision 88.7% against 90%); self-collected data
- Not met
- None
- Strict count
- 9 of 11 · 82%
- Partial rows counted as half
- 91%
- Demonstration
- Every applicable gate met with every part on (V98.1)
Question and method
Score each first-prototype requirement against the evidence of passed, locked gates only. The operating path is the demonstration engine: intake and moderation, reading through exact parsers or the learned reader, the qualified Stem Decision, the Mode Decision, the three stems through the staged Context path, the independent checker, memory layers, and a per-answer listing of every executed component with its size. No new run is made for the scorecard.
Recorded results
| Condition | Recorded result |
|---|---|
| Met within limits | Composition, linked Decision weights, learning during operation, self-chosen learning scopes, transfer from a person or Source Models, feedback, memory layers, Parallel Context in the operating path, selective loading |
| Partly met | Output-mode selection (learned Decision 88.7% against 90%); self-collected data |
| Not met | None |
| Strict count | 9 of 11 · 82% |
| Partial rows counted as half | 91% |
| Demonstration | Every applicable gate met with every part on (V98.1) |
Comparative standing and locked gates
The internal bar is at least 90% of results acceptable and verifiable. By strict count the prototype is at 82%; the remaining gaps are the learned Mode Decision’s last 1.3 points and a margin for self-designed research cases. A single locked run that scores every requirement at once has not been made.
Interpretation and limitations
This is a summary of recorded runs, not a new experiment. Several rows rest on development data or one regime, as stated in their own reports. Parallel Context is counted by its written criteria, which ask for computational use; neural use is not built.
SOURCE PROVENANCE
Qualification close, first prototype: FR-A to FR-G in one operating path (scorecard)
LABORATORY REPORT / 2026-10-20SOURCE CHECKSUM / SHA-256
480f06ffdda9224ff3ad1f14d49d8d088ff75d9bf75209388775e507d6a47cdfPublic journal edition reviewed 2026-10-22. Source documents and saved evidence were inspected; experiments were not rerun for this edition. Proprietary implementation code, model binaries, private infrastructure, and detailed machine records are not published here. Journal identifiers are editorial references. Catalog inclusion does not imply qualification or runtime promotion.