WikifitaGitHub live67e8de5
pesquisa · kaggle/pokemon_tcg_metanoia_tensor_scaling

Pokémon TCG Metanoia Spec 04 — Tensorized Scaling and Subagent Orchestration

Conceptual three-axis model for reasoning depth, parallel research roles and orthogonal domain context, with provenance-ledger requirements.

Baixar raw

Pokémon TCG Metanoia Spec 04 — Tensorized Scaling and Subagent Orchestration

Three axes

Spec 04 models research capacity as a conceptual tensor with three independent directions:

AxisMeaningPokémon TCG analogue
VerticalDepth of one line of reasoningStage 3 loss postmortem or one theorem audit
HorizontalParallel roles or candidate explanationsSeparate data, model, tournament and provenance analyses
OrthogonalDifferent domains or representationsCode, SQLite, Parquet, Git, transcript and mathematical model

The tensor is a planning metaphor and orchestration model. It does not establish that a neural representation has these axes internally.

Provenance ledger

The specification's strongest operational contribution is the provenance ledger. Every candidate or report should carry:

claim id
source artifact
repository commit
agent/harness identity
transformation applied
evidence class
review state

For this Wikifita pass, the ledger is instantiated through the corpus map, Git timeline, Antigravity page and evidence tables in the Stage 3 postmortem.

Sequential orchestration boundary

The project requested iterative integration rather than multiple overlapping agents. That is compatible with the tensor metaphor: the axes describe dimensions of the evidence space, not a mandate to run agents concurrently. A single coordinator can move through data, architecture, manuscripts and Metanoia sequentially while preserving cross-links and provenance.

Scaling risk

More candidates do not automatically create more knowledge. Without a stable ranking and review step, horizontal expansion creates duplicated claims, inconsistent denominators and derivative-artifact contamination. The Stage 3 failure is a technical analogue: more heads and signals increased optimization complexity without a shared scale contract.

Primary source

docs/metanoia/04_tensorized_scaling_and_subagent_orchestration.md

Cross-references