# A system that can revise its own representation language

Can a working system create new meta-representational concepts and revise its language while reasoning?

The proposal explicitly extends beyond memory efficiency and merging existing tokens. New semantic or meta-description units could develop while the system operates; sharing between systems is a possible use rather than a defining requirement.

Status: open problem. Formulation and comparisons: reconstruction. Development: active.

## Alternative representations

### A fixed vocabulary with reusable macros

Affords: Stable interpretation and composition.

Limits or losses: Restricts the forms of distinction a new situation can introduce.

### An evolving language of concepts and transformations

Affords: Lets the system introduce a previously unavailable abstraction.

Limits or losses: Needs tests of interpretation, backward compatibility, and the meaning of the new unit.

## Tension

A new unit may express a new conceptual distinction or merely abbreviate a longer phrase.

## Challenge

What evidence would show a new representational capability rather than a renamed old operation?

This challenge is an editorial prompt, not a claim that the evidence already resolved it.

## Proposed next move

Introduce an unfamiliar problem, require explicit expansion and counterexamples for a new concept, and compare with a fixed-vocabulary baseline. Count planning and execution costs separately.

Status: new model-proposed experiment or extension; not a reported result.

## Related problems

- [A reusable concept for similar inputs with divergent outcomes](concepts-for-divergence.md) — supplies a candidate kind of new unit (reconstruction). A parameterized concept for outcome divergence is one concrete candidate for an evolving meta-description language.
- [Changing the grain of a conceptual token](adaptive-token-granularity.md) — extends beyond compression (corpus_evidence). The explicit correction says new meta-representational units and change during operation matter beyond combining two existing units.
- [Memory represented through adaptable parameters](memory-as-adaptation.md) — extends beyond a memory objective (corpus_evidence). The explicit correction rejects treating the project only as memory efficiency. Memory remains one branch.
- [Learning to propose parameter changes](models-editing-models.md) — raises implementation questions (retrospective_connection). A changing representational language invites questions about corresponding parameter changes, but the mechanisms are not established as identical.

## Evidence anchors

- basis-evolving-representation-language: The material explicitly rejects narrowing the project to memory efficiency or combining existing tokens. It asks for new semantic and meta-description units that can develop while a model operates and states that sharing between systems is not a necessary condition of the idea. [Editorial paraphrase; presence of a proposal does not verify its truth.]

Root evidence and private recovery mappings are not part of this public release.

## Recorded revisions

- [From memory and token compression to a changing representation language](../REVISIONS.md#revision-representation-scope)

[Open in the reader](../index.html#node=evolving-representation-language) · [All problems](../CATALOG.md)

## Research editions

- [An Evolving Semantic Representation Language](../documents/evolving-semantic-representation-language.md) (reconstruction).