# Considering a problem is different from implementing a policy

Can unresolved policy and moral-plurality problems become generative training material without being mistaken for endorsed implementation?

Comparative interpretation and the generation of training examples are possible uses of open normative problems. Whether to implement a particular policy remains a separate judgment.

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

## Alternative representations

### An unresolved problem presented for interpretation

Affords: Permits critique without endorsement.

Limits or losses: A fluent response may prematurely close the question.

### Generated examples for learning or evaluation

Affords: Makes implications more concrete.

Limits or losses: Synthetic examples may normalize the generator’s favored solution.

## Tension

Expanding examples can enrich a problem or turn its unresolved tensions into a repeated doctrine.

## Challenge

Can a training example preserve a real disagreement and an implementation decision that remains open?

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

## Proposed next move

Generate contrasting examples with explicit objections and compare whether readers still recognize the unresolved issue.

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

## Related problems

- [A research space that remains revisable](open-problem-publication.md) — keeps exploration distinct from adoption (retrospective_connection). Publicly available problems may inform interpretation or learning without constituting an instruction to implement their proposed policies.
- [Social information as evidence and as contamination](social-evidence-exposure.md) — raises evaluation contamination (retrospective_connection). Generated examples and visible prior judgments can both alter what appears to be an independent interpretation.

## Evidence anchors

- basis-exploration-before-implementation: The material asks whether open problems would be considered when developing a model and explicitly distinguishes considering them from implementing a policy. It also asks about comparing interpretations and generating further examples for use in another model's training. [Editorial paraphrase; presence of a proposal does not verify its truth.]

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

[Open in the reader](../index.html#node=exploration-before-implementation) · [All problems](../CATALOG.md)