Given a theoretical model for a self-propelled particle or micro-organism, how does one optimally determine the parameters of the model from experimental data in the form of a time-lapse recorded trajectory? For very long trajectories, one has very good statistics, and optimality may matter little. However, for biological micro-organisms, one may not control the duration of recordings, and then optimality can matter. This is especially the case if one is interested in individuality and hence cannot improve statistics by taking population averages over many trajectories. One can learn much about this problem by studying its simplest case, pure diffusion with no self-propagation. This is an interesting problem also in its own right for the very same reasons: interest in individuality and short trajectories. We summarize our recent results on this latter issue here and speculate about the extent to which similar results may be obtained also for self-propelled particles.