It's like the Avengers of inefficiency. Particles from different locations, having undergone vastly different journeys, all assemble, for one moment, to work as a group to jam up a hopper.
This particular experiment was done in 2011 by scientists at Duke University to study the microscopic aspects of what's known as granular flow — the collective motion of large particles working together. If you've had something grain-filled to eat today, you've probably eaten something that's been through a hopper. If you've had beef, pork, or chicken, you've probably eaten something that's eaten something that's been through a hopper. Keeping these things efficient does make a difference.
The researchers used 5000 particles running through a two-dimensional hopper made of converging aluminium walls sandwiched between two pieces of plexiglass. They concluded that the only structure that would cause the hopper to jam was an arch formation. The particles couldn't jam the thing by bundling together or forming a straight wall of particles. The researchers also found, after much frustration, that it was impossible to predict which specific particles would cause the arch. The motion of the particles was too random to be able to predict when an arch would come together.
There are some patterns, though. Apparently, "The jamming particles come from roughly the same radial initial positions, but from a range of different angular positions." This means a particle from the far left, the middle-left, the middle, the middle-right, and the far right will always form the arch, but one can drop down from just a few centimetres above the hopper while another falls from a long distance away to form the arch.
If you want a clearer picture of how all the particles are moving, check out this simulation.
[Source: How granular materials jam in a hopper.]