The organisations and professionals who are commissioned to design and administer Language proficiency tests assure us that their “measurement technologies” comply with certain values. The values they cite most are “equity”, “impartiality” and “distributive justice”. This paper focuses on the way these values are “declined” while designing and constructing “industry standard” certifying tests. It also looks at the effects this technology is having on language use, language users and society. While doing so we make a somewhat unsettling discovery. Namely that there is a patent contradiction between the goal of fairness as defined by LPT service providers and the consequences of the methods used to attain this goal. This is so because test engineers must have some idea of a “norm” of communicative competence all “proficient” speakers of a given language are supposed to be able to use. But sociolinguists since Bakhtine tell us no such norm exists in as much as all languages are made up of a multitude of language use repertoires” whose intelligibility to anyone not belonging to the subgroup that uses it is never better than partial. Hence, testing technologies cannot privilege a single norm of language proficiency and communicative competence without being unfair to anyone judged by that norm who does not belong to the subgroup for which the selected language use norm is a linguistic reality.