A self-correcting quantum memory is a physical system whose quantum state can be preserved over a long period of time without the need for any external intervention. The most promising candidates are topological quantum systems which would protect information encoded in their degenerate groundspace while interacting with a thermal environment. Many models have been suggested but several approaches have been shown to fail due to no-go results of increasingly general scope. In a nutshell, 2D topological models and many 3D topological models have point-like excitations which propagate freely and change the groundstate at any non-zero temperature. A recent suggestion is to introduce effective long-range interactions between those point-like excitations. In this presentation, I will first explain the desiderata for self-correction, review the recent advances and no-go results, and describe the current endeavours to define a self-correcting system in 2D and 3D. Time permitting, I will briefly present our recent work on the thermal instability of models which aim to introduce effective long-range interactions between point-like excitations (joint work with Beni Yoshida, John Preskill and David Poulin).