Anthropologists have been trying to do that for decades, beginning with chaînes opératoires approaches in the 70s, to Haidle's cognigram approach ([0]), to the

Anthropologists have been trying to do that for decades, beginning with chaînes opératoires approaches in the 70s, to Haidle's cognigram approach ([0]), to the explicitly comp-sci inspired approaches like Perrault's [1]. They tend to suffer from the same set of issues, namely that they tend to be difficult to consistently apply within reasonable epistemological frameworks and unspecific in their arguments/predictions. Even when you have those, you still need to ground-truth things with the material record of "who was using tools when", so the kind of work discussed in the linked article remains valuable. [0] https://doi.org/10.1017/S095977431200025X https://doi.org/10.1017/S095977431200025X [1] https://doi.org/10.1086/673264 https://doi.org/10.1086/673264