The notion of “size effect” is introduced into the problem of machine stiffness by assuming that the rock specimen fails according to the slip‐weakening fracture model. Consequently, when the strain energy (stored per unit volume) in the machine and specimen is greater than the fracture energy (dissipated per unit area) of the rock, an unstable response is observed. Theory and experiments demonstrate that machine stiffness is not the sole factor in determining stability of a compression element, once the deformation has localized. In fact, the softening behavior of a rock tested under the same conditions is due to geometry and size. This means that the experimentalist may be able to design compression tests to achieve a stable post‐peak response.