Direction yes, dictation no

A brief that specifies every line a judge should say is not a detailed request. It is a different job entirely, and a lot of judges will tell you so and decline it.

What a script actually removes

The thing being bought in a human review is a person's judgement applied to what you sent - what they notice, how they phrase it, where they land the register. Hand over a full script and none of that is happening anymore; the judge is reading lines someone else wrote, on camera, which is voiceover work wearing a review's clothes. Presence is supposed to be the product, and a scripted delivery removes the part of presence that made it worth paying a person for in the first place - the response is no longer theirs. Emotional labour research suggests buyers can feel the difference: Hülsheger and Schewe's 2011 meta-analysis found deep acting, actually working up a feeling, related positively to customer satisfaction, while surface acting had a small negative relationship with performance.

Why judges push back rather than just deliver it

A judge who reads a script convincingly is demonstrating a performance skill, not the judgement skill they were hired to apply, and most judges who built a following doing the latter are not interested in quietly becoming the former without saying so. There is also a trust problem underneath it: a judge's value rests partly on buyers believing the response is genuine, and a habit of reading scripts, even when disclosed privately, sits oddly against that. Declining is often simply protecting what makes the rest of their work worth commissioning.

The exception

Not every judge treats this as a hard no. Some explicitly offer scripted delivery as its own separate product - closer to hired voice performance than to a review - priced and labelled differently from a judgement-based commission, so a buyer knows exactly which one they are getting. That is a legitimate distinction to offer, as long as it is not presented as the same thing. A judge who declines or reprices a scripted request outright is not being difficult - they are drawing the same line the format itself already draws.

If direction rather than dictation is what you actually want, that gets you closer to the real product: name the setting, the role, the beats you care about, and stop there, which is a request most judges are glad to take on. The line matters elsewhere too - an automated score has no lines to read in the first place, and a stated measurement does not involve a script either, since neither one is a performance to begin with. Rate Cock's request flow is built around a brief field rather than a script field for exactly this reason - it is asking what you want addressed, not what you want said. An AI system has the opposite problem in reverse: it cannot be handed a script at all, since there is no performance step for one to enter.