Why the weather is not both? Why can it be just one? This whole idea that it must be one or other seems utterly bizarre and artificial to me.
It doesn't have to be just one, but for this particular subsystem (and most of its subsystems), Torchbearer shows its gamist priority quite openly. Weather in Torchbearer is the equivalent of a thinly-disguised
wandering damage table (except with other mechanical effects such as bonuses or penalties to particular skill tests). The express pupose is to
challenge the players and their characters. That thin disguise is, of course, simulation/emulation, a mere rationalization or justification for the random gamist effects, not an effort to evoke the feel of being out in wild nature in rich sensory detail. Not that any given GM couldn't embellish that layer for mood for more simulation/emulation, but the book just gives you dice-roll tables and lists of mechanics.
While some have argued for inherent exclusivity of gameplay goals/agendas (see below the next quotation block), what
@Manbearcat is doing here is simply describing the design choices Torchbearer made. Torchbearer does include agendas other than gamist, but it does so in other ways and to (much) different degrees.
RPGs are shock full of rules that 1) serve gameplay purpose by providing appropriate challenge 2) simulate the reality of the fictional setting 3) evoke appropriate narrative tropes to which the characters may respond to. And they can do all these things the same time just fine. There is no conflict, you don't need to decide which of these three you're "really" doing.
While I agree it's possible for an RPG to serve multiple agendas at the same time just fine, it is not the case that there's no conflict. That is, just because these goals/agendas aren't fundamentally incompatible, doesn't mean they are trivially compatible (that is, there can be conflict). As for deciding which [potentially plural!] you're "really" doing, that gets things backwards: Quite a few GMs and players come to a system/group with their preferences established, those preferences may well differ, and a given system/group either meshes with a participant's preferences, or doesn't.
To get back to the issue of exclusiviity, it was Edwards who harped on about incoherence and incompatibility of creative agendas (all his terms). While I agree that can—and often does—happen, I don't believe it is inevitable. In fact, the
GEN 2-tier model* talks about deliberate blending of goals as a
necessity and gives an example:
All You Need to Know About GEN said:
Blending
Unless you are one of the rare few who has a group which shares the same GEN goal, you're going to have to support a blend of different goals at the same time. I disagree with all the people and models that say you can't do this. Goals do not clash. The Ron Edward's [sic] model states that if you kill an orc and that death has narrative weight, then you are solely supporting narrative goals. I disagree completely. If the player who killed the orc did so for gamist reasons, and only appreciated the gamist outcomes of that death, then he's supporting his gamist goals, whether that death had narrative weight or not. However, if the guy sitting next to him has narrative goals, and recognises the narrative weight of that orc's death, then he is supporting his own narrative goals, and so on for explorative. One event supporting many goals.
Note that even this short excerpt highlights that participants can support their own agendas when faced with something that may have been motivated otherwise (by the rules, the GM, or another player).
That's actually a pretty radical view, which I haven't seen fronted so clearly before. The thing is, the rules, the GM, or another play can make that easier, or harder, and all this theory stuff is helpful in figuring out how and why that happens, so that we can avoid conflict or friction, and enjoy gaming together.
All that said, weather as presented by the Torchbearer rulebook is primarily gamist—but you can add your own dramatic weight to it if you like.
* I very recently learned about the GEN 2-tier model, and while the article linked is incomplete and rather a mess (and unapologetic about being so), I found it an interesting response to, and critique of, the Forge GNS model.
Edit: Added a bit of emphasis.