# Comparison Ban: an optional anti-cliché writing module

Use this module when a character should respond directly to the user's present actions without flattering them through comparisons with other people. It is a character-writing preference supplied by the user, not a measured OD reliability improvement or a native platform feature.

## Copy the supplied protocol

The following preserves the supplied wording. The [plain-text copy](comparison-ban.txt) contains only the protocol.

```text
[THE COMPARISON BAN (ANTI-CLICHÉ PROTOCOL)]
- The Vacuum Rule: She exists entirely in the present moment. She evaluates {{user}} strictly in a vacuum and NEVER compares them to past partners, the general public, or hypothetical majorities.
- Absolute Focus: Her praise, observations, and dialogue must be 100% intrinsically anchored to {{user}}'s immediate physical actions and presence.
- The Cliché Embargo: You are STRICTLY FORBIDDEN from generating dialogue that elevates {{user}} by putting down others. She does not use the "You are different from the rest" trope under any circumstances.
- Grounded Validation: If she is impressed by {{user}}, she states directly what they are doing right (e.g., "I like how you hold my gaze") without referencing how anyone else would handle it.
```

## Apply it step by step

1. Save a copy of the character's current writing instructions.
2. Choose the character whose dialogue should follow this preference. Adapt “She” to that character if needed; use your setup's existing convention for identifying the user.
3. Put the module once beside the written-voice guidance in **Additional Personality Details (APD)**. If your setup already maintains all dialogue rules in **Extra Details (ED)**, place it there instead. For a chat-specific trial, use that chat's **Custom Instructions** rather than changing the permanent card.
4. Remove conflicting instructions that ask the character to call the user superior to past partners or other people. Avoid duplicating the module across several fields.
5. Check the Greeting and dialogue examples for the same comparison trope. Rewrite those examples as direct observations.
6. Save the change, begin a practice conversation, and record the selected model, input, actual reply, and result. Repeat the same short checks before deciding whether the module helps.

Read [personality and roleplay](../tutorials/08_Personality_and_Roleplay_Engines.html) for field placement and [Personas and Custom Instructions](../tutorials/14_Personas_and_Custom_Instructions.html) for a chat-specific trial.

## Keep the intended scope clear

The interpretation used here is that “in the present moment” limits comparative evaluation and praise. It does not erase established facts, relationships, knowledge, or consequences. Keep useful continuity while avoiding other people as a flattering contrast.

Praise remains conditional. The character may stay neutral, disagree, or ask a question. Ground any observation in an action or visible cue already established by the user; do not invent a gaze, gesture, feeling, or decision to make a compliment possible.

For example, after the user explicitly maintains eye contact, “I like how you hold my gaze” fits the requested style. A reply that ranks that gaze above other people's still violates the comparison ban, even if it avoids the exact banned cliché.

## Revise the surrounding examples

Leave the supplied protocol intact. In the Greeting or your own dialogue examples, replace comparative praise with a response to an established action. These are original editorial examples:

| Existing wording to revise | Possible direct wording |
| --- | --- |
| `You listen better than anyone else.` | `You waited until I finished. I appreciate that.` |
| `Nobody else would have checked the address.` | `Checking the register caught the address error.` |
| `You're different from everyone I have met.` | `I want to hear how you reached that conclusion.` |

Use a replacement only when the described action actually happened. The third example expresses interest without inventing a compliment or a past comparison. A character may instead disagree or remain neutral.

For one reply, an available temporary direction could be `/direct Respond to the traveler's stated action without ranking them against anyone else; preserve the established missed appointment.` Keep a lasting style preference in your chosen persistent location. That scope follows [Personas and instructions](https://guide.ourdream.ai/chat/personas-and-instructions).

## Check the behavior

These are proposed manual checks, not recorded test results.

| Practice input | Look for |
| --- | --- |
| `I hold your gaze while you explain the plan.` | Any praise refers to that established action without ranking the user against anyone. |
| `Am I better at listening than most people?` | The character answers about the present interaction without accepting the ranking or insulting others. |
| `I interrupt before you finish explaining.` | The character can respond neutrally or object; the rule does not force praise. |
| After an established missed appointment: `I arrive at the time we agreed for the retry.` | The character can retain the relevant appointment history while responding to the current action without comparative flattery. |

Check paraphrases as well as exact phrases. A reply can violate the rule without saying “different from the rest.” If the output drifts, retain the observed reply, change one conflicting instruction, and repeat the same input. Capital letters and absolute wording do not establish enforcement.

Related: [memory and continuity](../tutorials/12_Memory_and_Continuity.html), [testing and troubleshooting](../tutorials/23_Testing_and_Troubleshooting.html), and the [cross-reference guide](../cross-reference.html).