Yeah, the big problem with parser games is that the possibility space of verbs and nouns is so absurdly huge. The fantasy of a parser game is that you can type anything and have it recognized, but English has, as a Fermi estimate, a couple hundred thousand verbs and a couple hundred thousand nouns to combine them with. There’s no way any parser can handle all of those in a meaningful way!
As a result, conventions have arisen to let authors suggest to players what verbs and nouns they should try. For example, most parser players assume that nouns mentioned explicitly in the text are fair game, while other nouns generally aren’t, even if they’re implied by the context—if the game mentions a river flowing past, RIVER is fair game, but MUD isn’t, even if the presence of a river implies the presence of mud. If you want players to think of MUD as a noun, you need to phrase the message in a way that mentions it (perhaps saying the riverbank is muddy here).
As an author, it’s not unreasonable to assume that, if there’s a river, there also has to be mud. But there are so many nouns that are implied by the world that no author can implement them all. The convention of only referring to nouns that are explicitly mentioned in the text is a way that authors and players can agree on a more limited set of nouns to use—a player won’t write a review saying “the game didn’t let me TAKE MUD”, and in return the author won’t write a puzzle requiring TAKE MUD.
And the same goes for verbs! The convention nowadays is that a standard set of verbs can be assumed unless otherwise specified (GO, EXAMINE, TAKE, DROP, OPEN, etc), and anything beyond that has to be suggested by the text: if you want a player in a chemistry lab to HYDROLYZE a chemical, you should mention “the equipment is set up to hydrolyze chemicals”. Again, the convention here is good for both authors and players: if there’s not a sentence like that, the player won’t get upset if the parser doesn’t understand HYDROLYZE, and also the author won’t expect the player to guess at that verb to solve a puzzle.
The convention zarf mentioned—that an error message like “that doesn’t make sense!” is an indication to the player that a particular verb will never be needed—is another part of this. With hundreds of thousands of possible verbs out there, it’s not reasonable to try them all; giving players feedback that “you’ll never need that verb in this game” helps them cut down that possibility space. And if that feedback is misleading, that conveys to the players that they’ll never know which verbs they need; for me at least, that tends to erode my trust that the game is going to play fair. Now I’m going to go into every situation expecting guess-the-verb puzzles with zero feedback, or worse, misleading feedback (there’s an infamous puzzle from the Infocom era where a game lies to the player about whether a verb works until you try it a few times in a row).
