An allocation bias criterion is formulated from a necessary self-consistency requirement. • Error-and bias-free algorithms are required to remove mathematical and algorithmic inconsistencies. • All tested pattern-based LUCC models are formally incorrect and biased to varying levels, but Dinamica EGO is close to be error-and bias-free. • A comprehensive Land Use [and cover] Model (CLUMPY) based on error-and bias-free allocation algorithms has been developed and is freely available. • For algorithmic testing and validation-instead of spatial location validation-comparison and validation must be performed in explanatory variable space, not on maps.