# Research idea gates

Apply all gates before including an idea in Section 16.

## 1. Traceability gate

Tie the idea to a specific limitation, failure case, assumption, contradiction, or unexplained result. Include its source pointer.

## 2. Hypothesis gate

State a proposition that could be false. Reject vague ideas such as "improve efficiency" or "use a larger model".

## 3. Delta gate

State exactly what changes relative to the paper:

- representation;
- mechanism;
- data;
- objective;
- feedback;
- evaluation;
- deployment constraint.

## 4. Validation gate

Specify:

- comparison;
- controlled variables;
- metric;
- dataset or population;
- expected observation;
- falsifying result;
- compute or experimental budget when relevant.

## 5. Failure gate

Give at least two plausible reasons the idea may fail. Include attribution error, confounding, instability, cost, data scarcity, or invalid assumptions when applicable.

## 6. Novelty-language gate

Use:

- `candidate idea`;
- `hypothesis`;
- `possible extension`;
- `prior-art search required`.

Do not use `novel`, `first`, `unexplored`, or a numeric innovation score unless a dedicated prior-art search has been performed and cited.