Summary
Manipulations are part and parcel of empirical research. The mechanistic mutual manipulability criterion (MM) reflects the important role that interlevel manipulations play in scientific practice. Yet, it faces a complication: empirical practice does not usually offer the clean mutual manipulations that (MM) builds on. Recently, two suggestions have been made as to how manipulability criteria might be adapted to better fit empirical practice. I will argue, however, that substantial problems remain for both. Finally, I will point to a further aspect of empirical manipulability that philosophers of science should not overlook, viz. manipulability not of components and their activities but also of their organization.