For within-subject tasks we might want a counterbalancing mechanism, rather than make multiple versions of an experiment. This could be a version of a supertask, that sets the order based on which orders were already done by other participants... or at random, in case of ties.