The policy stance is directly implied by the probability estimate. This is certainly true if you accept my anecdata of ~6-8 of my relatives from the south of Ukraine and Kyiv who are, to a person, adamant about the vital necessity of winning the war to ensure their kids won't have to fight another one 20 years from now. There's absolutely no dithering in Ukraine about what the goal is: 1991 borders and one or both of EU and NATO memberships, with the NATO membership very likely being first.
These people didn't suffer for 13 months to accept some half-ass appeasement of a ceasefire to give Putin some breathing room to regroup.
> The policy stance is directly implied by the probability estimate
I think this is not the case. The policy "we won't accept peace until we achieve x" is very different to "we won't accept peace until we achieve x conditional on the achievement of x being probable"[0]. The first implies a kind of "to the last Ukrainian"/"you'll have to take it from my cold, dead hands" attitude, the latter doesn't.
[0] A conditional such as this is implied by your claim that "The policy stance is directly implied by the probability estimate": if the probability estimate leads to the stance, then it stands to reason that the stance may change if the probability estimate changes, which would (in this case) contradict the very nature of the stance (specifically, the "to the last Ukrainian" aspect discussed above).
These people didn't suffer for 13 months to accept some half-ass appeasement of a ceasefire to give Putin some breathing room to regroup.