@Reynard Something else to keep in mind with FitD (though it applies to PbtA, too) is that you have a lot more leeway to introduce consequences that, in a trad game, would be a logistical and gameplay nightmare. Adding four enemies to an in-progress combat encounter would be daunting enough in another system, meaning that many more statblocks ready to go, plus all the extra time added to resolving the combat. Or if you threw in an undead dragon, and now you have to worry about a TPK, since in another system that dragon has specific rules and numbers of attacks and usually one way of being dealt with (grinding down its HP or equivalent).
In FitD all of that stuff is just fictional positioning, and can be factored into position, effect, and further consequences, including Harms. So if the dragon attacks and someone says "I'll draw it away, run!" you can resolve that PC's action, without getting into combat turns and stats and everything else. Likewise, adding more enemies just means the overall threat is greater.
I'm saying all this because, to me, FitD and PbtA GMing are maybe 90 percent about coming up with consequences and complications in the moment. That usually means being willing to take the entire scene, session, and maybe campaign in a new direction at any point. But it also means making sure the fiction is really, truly dialed in—not just the immediate fiction in the scene, but what sorts of things happen in this setting, and do the players have a sense of that, so they aren't constantly being caught unawares or asking questions to help define it.
And all of that is dependent on the players having full and consistent buy-in. They have to trust the GM way more, imo, than in a trad game, because the rules will never really do the work of determining what happens. If you say that the risk of drawing that undead dragon away on your own is death—maybe there's basically no cover, nothing to duck or hide behind—and they do it anyway and roll a miss, then they are dead. Now FitD has ways for the player to mitigate that sort of outcome, such as a Resistance roll, or maybe they can burn some special armor. But if they're out of those, or they refuse to use them, or even if using them just turns a lethal injury into a serious, crippling one, they need to accept that you, the GM, have come up with an appropriate consequence. There's no damage roll to shift responsibility to, no failed death save roll, etc.
If that trust and buy-in is there, it can all flow beautifully, and insanely quickly, and the GM can feel increasingly comfortable coming up with consequences and sticking to them. But if you have the kinds of players who aggressively challenge rulings or think they can rules-lawyer their way out of situations—or who are just used to a traditionally adversarial GM-vs.-PCs approach—then it can really fall apart. So the idea of FitD being collaborative is more than a matter of individual play loops or mechanics. It's the whole thing. Swim together or sink apart.