A key conceptual matter in language evaluation in general, and in Alan Davies' work in particular, is never fully settled, viz. how is responsible language test design linked to ethicality? This goes back to the unsolved discussion about validity and validation, that has ended in a loss of conceptual clarity about solid language evaluation. The inconsistencies inherent in first stating validity to be the overarching concept, and then trying to extend it by fostering other ideas to be the prime considerations, further mix up the matter. This paper asserts that such expansion is unhelpful. A way out is to consider language test design as being responsive to certain typical and general conditions. In that relationship, between the (subjective) making of the test as an artefact that presents us with a reply to certain design principles, and the designed object, the actual test itself, lies a possible way out of the deadlock.