God I don't want to get involved in this conversation (and
@hawkeyefan 's point was to prevent exactly what is happening here!), but even when you do have actual quantitative measurements (eg a mapped 10 * 15 * 10 dungeon room where the exits have suddenly shut), the GM isn't performing the actual calculations necessary to determine how quickly the cubic volume of the room will fill up with the water that is pouring into it after a trap has been triggered. They're eyeballing things (or following the adventure modules eyeballing of things) and using the exploration rounds as units of time and currency for action declaration/resolution as a proxy for resolving "can you get out in time or solve the trap problem"?
So even when a mental model or thought experiment actually could be accurately parameterized and run to accurately answer the question inherent to the model, no GM is doing that! They're qualitatively abstracting and using the game's units/resolution mechanics (or they're using a Scientific Wild Ass Guess if the resolution mechanics/intersection with the game's units are wonky/lean/incoherent and/or the game is "loosey goosey") as a proxy for deciding the interactions between the players, the system, and the gamestate!