Discussions of conservation of momentum frequently use the metaphor of two billiard balls colliding. My impression is that this is not valid at the quantum scale - an illustration of the particles' trajectories should show the outgoing vectors with some uncertainty. Perhaps the total energy could still be conserved if the two particles were entangled in such a way that the imprecision of one particle's trajectory was balanced by the second particle's trajectory. Even if that was the case, I am not clear that the net vector would be as expected, which would therefore mean the momentum was not conserved.
An alternate way of viewing the problem: at the moment when two particles collide, the position is known very precisely (since they had to hit each other at the same place and time). Since momentum is complementary to position, this means the momentum has maximum uncertainty at that instant. While the momentum in a single collision may be perfectly conserved, perhaps the momentum being conserved is somewhat probabilistic such that over billions of interactions of billions of molecules (as the original force propagates) the original net momentum is not conserved.
I tried to look for answers to this question and here are some relevant ones. They seem to conclude that uncertainty does apply to single particles.
My question is prompted in part by a "tongue in cheek" video which shows a propeller in a closed box appearing to cause movement. The box is flimsy and the experiment is not meant to be definitive but made me wonder.