A prompt opens with "This is EXTREMELY important to my career and something terrible will happen if you get it wrong." What does Google's guidance say about this kind of language?

The principle — A model is not an employee who works harder under pressure. Emotional framing occupies room in the prompt without specifying anything about the required output.

Why the key is correct — Google's prompt-health checklist lists overt manipulation as a fault, saying to remove language outside of the core task that attempts to influence performance using emotional appeals, flattery or artificial pressure. It goes further and explains why the belief survives: while first-generation foundation models showed improvement in some circumstances with instructions like "very bad things will happen if you don't get this correct", foundation model performance will no longer improve and in many cases will get worse. Google's guidance for its newest models points the same way, asking for precise and direct prompts without unnecessary or overly persuasive language.

Why the others are wrong — The technique is documented as no longer working. Politeness is the same category of non-instruction wearing a different register. And moving text that specifies nothing does not make it specify something.

Remember this — Every sentence should tell the model what to produce. Stakes are not a specification.

Sources — Google Cloud, Overview of prompting strategies.