1. typically, it's Go Aggro: you're shooting some dude and that dude doesn't want to be shot, but you're insisting.
2. if the PC doesn't really care about the harm they're taking and/or is trying to seize a particular objective, regardless of the cost, that's Seize By Force (Seizing means the PC takes harm as well), but that doesn't happen very often,
I see this (or something like it) a lot and I have to say I am profoundly skeptical of this breakdown when it comes to dealing with armed people who are fighting back. (And you know, as a rule if someone is shooting you you want to be either fighting back or getting away.) Mostly what concerns me is the idea that the Go Aggro option is universally available, simply based on the PCs intentions regarding whether or not they personally want to take harm. At what point do you decide that to shoot the person they must be doing it "regardless of the cost", and otherwise they simply cannot shoot them at all? Isn't the actual deciding factor not the PCs attitude towards the confrontation but whether or not the NPC in question is in a position to defend themselves?
It's true that mechanically, it is easier to kill people using Seize by Force, since a 7-9 result will often allow you to do enough harm to kill them, whereas a similar result on Go Aggro leaves them unharmed. But it also seems a little kludgy to me to have a 10+ Go Aggro result that in practical terms removes the NPC's choice -- since 'die or take 3 harm' is basically 'die or die'.
In any case, my own take is not practically that different, I guess, but I think Seize by Force has more of a place than common wisdom seems to be slanting, assuming your game involves actual straight-up fighting. Lots of games don't, I guess, but I have never really encountered a situation where I felt like Seize by Force was used mistakenly, or produced poor results when it was used -- but I hear a lot of people suggesting that it has done that for them. I would be curious to hear some specific examples of this problem, from people who have run into trouble with the move.
My group has used 'seize X's life' by force and it worked perfectly -- it was used during a skirmish-level firefight to give the PC control over that NPC's life, ie. it was essentially a set-up move that then allowed them to decide whether or not they wanted to kill the NPC or not (which they did, on their next turn.) They could just as easily have used it to then wield the NPC as a hostage, or paint a silly moustache on them, or anything else you can do when you have basically established that you can kill somebody whenever you want.
We have also used it to account for simple firefights or melees between two armed, violent, mutually aware individuals, and it worked just fine. In these situations it barely seems necessary to define what is being Seized by Force, because the results of the move are so clearly appropriate to the fiction -- I think this idea that there must be
some thing that is being seized is a kind of overthinking it. If 'what you want is for the NPC to die' is appropriate for Going Aggro, then 'seizing this guy's head with my fire axe' should do just as well.
Like Vincent said, the things that the move allows to happen seem like the best guideline in the final analysis. Go Aggro is perfect for violent situations where what you want to know is whether the target gets away unhurt or not; Seize by Force is for violent situations where what you want to know is how much hurt it costs the PC to achieve their objective.
And I guess Acting Under Fire is an all-purpose filter that can be applied overtop either situation, either replacing the other moves or enhancing them.