In this article, I use a Foucauldian poststructural analysis to examine productions of progress within key discursive spaces of mathematics education. These sites of production are educational policy, mathematics education research and case studies of primary school student-teachers in England. From my analysis, I show how progress governs what is possible in the classroom, as they become constructed around a measurable, linear temporality assumed in educational policy. This encourages comparison to and pursuit of the "normal" mathematical child, which in educational policy is produced as a functional automaton, whilst for much of mathematics education research is produced as the cognitive "natural" child. These over sanitised constructions result in confusion for student-teachers who struggle to take these impossible discourses on board.