I mean "error" related to the personal choices of practioners, i.e. datasets selection inside the LCA software. In fact, having the same model, the choice of different datasets from the database implemented in the software can lead to quite different results.