A glossary problem

The same register has different names on different sites because nobody ever standardised them: the labels grew up in separate communities and overlap without mapping cleanly. One judge's "roast" is another judge's "mean," so treating a label as fixed is how a brief and a delivered clip end up mismatched.

Why there is no shared dictionary

Registers grew up on different platforms, among different communities of judges, with no central body defining what any of the terms mean. SPH, humiliation, degradation, roast, tease and mean overlap heavily and were never coordinated to do otherwise. The problem is old and well studied: in their 2005 analysis of collaborative tagging, Golder and Huberman found that synonymy - several words for closely related meanings - "presents a greater problem for tagging systems" than one word with several senses, because inconsistent terms hide relevant items. A judge who came up doing "mean" content on one platform and a judge who came up doing "degradation" on another may be offering close to the same clip under two different names, or two genuinely different clips under names that sound interchangeable.

There is also drift within a single term over time, as newer judges adopt a word slightly differently from how it was first used.

What this means for a brief

A label is a shortcut, and shortcuts fail when the two parties do not share the same map. Describing the tone - flat and blunt, or sharp and demeaning, or light teasing that never quite bites - survives the mismatch that a single word does not. Stating the register in one line works better as a description than as a category name, for exactly this reason.

This is not a case for a full dictionary of terms in every brief. One clear sentence describing the tone wanted, with an example if one comes to mind, does the job a label alone cannot. For the fuller set of terms this site uses and what each one means here specifically, the glossary defines them plainly in one place.

Elsewhere, the same problem in different shapes

Automated systems avoid this entirely by not naming a register at all - an AI score has one voice regardless of what anyone calls it, which is a different trade-off, not a solution to this one. A numeric score has the same property: Penis Rater's scale means the same thing on every result because there is no persona vocabulary to drift. Measurement has it worst of all solved: a centimetre means the same thing everywhere, which is most of why a stated figure is worth taking properly rather than guessing. None of that consistency is available from a person working in named registers, and it is not meant to be - Rate Cock's comparison of the two approaches is a reasonable place to weigh which trade-off you actually want.