That sounds like a busted scenario for gamists because of the effect luck can have, but I can see the argument that it’s a form of anticlimax for them. You thought you were going to get a game, but whoops you accidentally won already. That’s different from what I had in mind (hence my mention of rigging), but it’s a fair point.
That's the meta-concern about randomness I was making. The precise placement and application of randomness is a huge, complicated issue in board game design, in a way it just isn't in RPGs. If anything, they tend to treat it backward; in board games it's generally randomness in establishing the initial board state, or in giving the player access to more options (like drawing cards off a deck) that is preferred, unless the game is explicitly about gambling, in which case randomness in resolution (rolling dice to see who's ship survives the attack) is possible. Even then, skilled play in a lot of those games is generally about mitigating randomness if at all possible.
It should be possible to maintain the integrity of the board state as long as one is transparent about the effect actions will have and communicates when the board state changes. That should be sufficient information to make an informed decision. What simulation does is provide a common understanding of how things work. If that understanding can be provided through other means, then simulation is not necessary.
I think I agree with you, but this very much sounds like the "design a board game, and then describe fiction around it" approach. You have to do a lot of work if you want the board state to keep presenting interesting choices and remain consistent. I don't know your game obviously, but the problem I usually see is either that the board state presents trivial optimization problems (which are generally not even considered as optimization problems for narrative reasons) or you lose a degree of optimization by not allowing the players to avoid failure points. That is, the player's declarations are fixed in relative time, thus that whatever they do will have an equivalent impact (or if we're talking variable completion, an equivalent magnitude impact) on the outcome. Consider trying to break into a castle using skill checks based on objective DCs, vs. a flat skill challenge to do the same thing. It is possible in the former case to present a course of action that calls for more or less rolls, and not in the second, even if the resulting fiction between the two is otherwise identical (admittedly a really unlikely and probably contrived scenario).
The appeal of a simulationist model is that actions naturally move the player different distances toward their objective, and do so without any risk of someone tweaking the board state, which gives you at least two axes to plan lines of play along: how many failures points, chance of success at any given point, and then likely more (resource expenditure, severity of board state at failure, etc.).
That’s not to say no to simulation ever. I like having some physics-ness to my mechanics even if the numbers are what they are to make the gamism go brrr. The fact that MP in my homebrew system also says something about how the world works helps keep the cognitive load down of the game mechanics. Does it use MP? It’s magical. Is it magical? It uses MP. Any consequences for deviating from the normal usage of MP then becomes an emergent property of that physics-ness. I think that’s neat.
Yeah, this relates to that broad church view of simulation I was pushing earlier. It doesn't matter how the world works, as long as the workings of the world are consistent and consistently explained; we don't have to simulate reality, we just need an impartial and consistent world (to a reasonable degree of abstraction).