There a number of interesting questions in strong gravity and cosmology related to the collapse of collisionless matter that can be studied with the aid of tools from numerical relativity. One question concerns the validity of cosmic censorship, the idea that the breakdown in the predictiveness of Einstein's equations associated with unhalted collapse is always hidden behind a black hole horizon. I will describe a long-standing putative counterexample to this conjecture in the collapse of elongated distributions of collisionless matter, and how cosmic censorship is ultimately rescued. I will also touch on some related problems that can be addressed with similar methods, including the ongoing debate regarding to what degree gravitational nonlinearities can backreact on the large scale expansion of the universe, and when and how black holes can form in ultrarelativistic collisions.