The fragmentation of self-gravitating protostellar discs into bound objects has long been considered an alternative planet formation mechanism, and may be required to explain massive planets orbiting at large distances from their star. In its most recent incarnation, Tidal Downsizing, the disc fragments may obtain solid cores through sedimentation, migrate inwards and be tidally stripped, forming objects with a range of masses and properties.

I will present several investigations of disc fragmentation and tidal downsizing, which include high resolution radiation hydrodynamic simulations and semi-analytic population synthesis models. The combined evidence is clear that disc fragments have a high mortality rate. The survivors (bound and free floating) remain at large distances, which we show are easily visible with current direct imaging surveys, placing strong constraints on how frequently discs fragment.

However, even if fragments don't survive, the fragmentation process irrevocably alters the disc chemistry. I will show preliminary work demonstrating that while disc instability might not form planets, it sets the initial dynamical and compositional conditions for planet formation by core accretion.