Firstly, stacks will no longer will remember their home locations after a single failed home routing calculation. More importantly, instead of recursively pathfinding the furthest transporter from a stack's current location, the stack will randomly select a direction to start traveling. Once it reaches the end of a line of transporters, it will attempt to recalculate a path to an inventory. If it is unsuccessful, it will simply pick another direction to start traveling and the process will repeat. This keeps the previous behavior, but does it much less CPU-intensively.