So, that’s one way to do things, but there’s another perspective that looks more like the multiplier approach:
Suppose that in order for B to happen, A needs to happen first. Furthermore, suppose that you’ve never observed A or B. Then, instead of using Laplace to forecast B, you should probably use it to forecast A, and then multiply by p(B|A). I think this isn’t such an uncommon scenario, and a US-Russia nuclear exchange probably fits (altho I wouldn’t have my probability factor be as low as 0.5).
So, that’s one way to do things, but there’s another perspective that looks more like the multiplier approach:
Suppose that in order for B to happen, A needs to happen first. Furthermore, suppose that you’ve never observed A or B. Then, instead of using Laplace to forecast B, you should probably use it to forecast A, and then multiply by p(B|A). I think this isn’t such an uncommon scenario, and a US-Russia nuclear exchange probably fits (altho I wouldn’t have my probability factor be as low as 0.5).