We propose a new formal definition of discovery for a Solar System object. It is based on an objective and mathematically rigorous algorithm to assess when a set of observations is enough to constitute a discovery. When this definition is satisfied, in almost all cases the orbit is defined well enough to establish the nature of the object discovered (Main Belt vs. Near Earth Asteroid, Trans-Neptunian vs. long period comet). The frequency of occurrence of exceptions is estimated by a set of numerical experiments.
The availability of a non-subjective definition of discovery allows some rules to be adopted for the assignment of discovery credit with a minimum risk of dispute. Such rules should be fair, encourage good practice by the observers and acknowledge the contribution of the orbit computers providing the identifications and the orbits, as well as the one of all the contributing observers.