We report on the initial attempts at evaluating "at scale" a particular technological/curricular innovation that enables more students to develop deeper knowledge. The methods, issues, and findings of the current pilot experiment speak not only to the success of SimCalc MathWorlds, the focus of our research program, but also to the evaluation at scale of a broad class of representationally innovative technologies and to the merit of long-term investment in design-based research. In particular, we present conditions and findings from a completed pilot experiment involving 21 seventh-grade mathematics teachers from Texas. Pilot outcomes suggest that (a) innovative representational technologies can have an important impact on student learning, (b) considerable impact can be found across a wide range of teachers and conditions, and (c) these gains can be detected even in the absence of other desirable conditions. In particular, detection of student gains does not, in our case, depend on having a long-term context of learning, long-term teacher professional development, or a shift to learner-centered constructivist pedagogy. The full experiment will replicate and extend our experimental design with a wider range of teachers and schools, model the factors that contribute to classroom success with such technology, and explore what happens as research support fades away. (Contains 10 figures, 2 tables, and 1 footnote.)