Saladman wrote:But I'm wondering to what degree you can choose between those: is Delphie choosing to ask the heavy questions about who's going to win when she could choose to just ask where the critical battles will be and stay blind to the outcomes, or is this all coming on her against her will?
Exactly, you want info on what's going to happen, but not info on success/failure. Knowing the enemy will attack a particular city is useful, knowing that they will take the city is not.
The best questions are ones which are almost like lookamancy. They gain info, without locking down any event which hasn't already been set.
Maybe better predictamancers have better control over what questions they ask. I wonder if just thinking about a question can trigger a prediction.
Also, "will this side fall?" sounds like a question that will be answered Yes, always. The only exception would be if there was a max time limit. The answer could be that the predictamancer can't see the side falling when they look. Effectively, the question is actually, "Will the side fall within 1000 turns?", depending on how strong the predictamancer is.
Long duration predictions presumably carry the world forward to that time, and then give a result, which is then locked in. This means that the predictamancer is predicting the fate of their side, assuming that they don't have a predictamancer. Lots of short duration predictions, focused on information gathering would be much more effective. Ironically, despite being a philosopher king, Banhammer did get the practicalities of that right.
Marle said "I so predict it" about Janis stacking with Parson. If that was a prediction, then it was a risk. Predicting it didn't really change the odds. in either direction However, if she predicted that Janis wouldn't stack, then that would be locked in.
And does Predictomancy "fix" the future once it's been seen, or is it a glimpse into what's going to be whether you look or not, so you may as well peek and try to find the best path?
That sounds to be the way it works, or at minimum how the predictamancers believe it to work.