A ballistic missile submarine's position isn't completely random, but they do move around somewhat randomly to reduce the risk of detection. They intentionally avoid repeating movement patterns or lingering in the same spot. And this also reduces the risk from spies on land. If a spy sees the patrol orders he'll only know a large general area but won't precisely know the submarine's position at any time.
So you're saying the collision _didn't_ occur at a time when the sub was trying to stay hidden, you're saying that the operators could have / should have used active sonar? Is that it?
Look, either the sub was trying to stay hidden at the time it accidentally collided with another submarine, or it wasn't trying to stay hidden. In either case it didn't do a good job. If it wasn't trying to stay hidden, then it shouldn't have collided. If it was trying to stay hidden, then its location was not as unpredictable as it should have been.
It's misleading to describe collisions as "theoretically possible" when we have evidence that these collisions have occurred. The fact that collisions have occurred makes me think that these submarines are not randomly located in a large area where it would be difficult to locate them.