“…architecture and the built environment) -there is a shared interest in streamlining the process of feature extraction from visual data, and importantly, in building deeper understanding around visual communication and aesthetics in general (see for example : Lyons, 2017;Ibarra et al, 2017;Joo and Steinert-Threlkeld, 2018;Steinert-Threlkeld, 2019;Al-Halah and Grauman, 2020;Xi et al, 2020). This divides broadly into understanding individual perceptions and preferences (Stamps, 2002;Witzel et al, 2017), as well as building generalizations about desirable and compelling image composition -a theme currently explored across the social science and computer science literatures (Badea et al, 2018;Constantin et al, 2019;Matz et al, 2019). This research note aims to highlight that scalability should not be an overwhelming factor in methodology design in image mining in the social sciences, especially at the expense of detail and nuance.…”