@anique.yael and I are just back from Oxford, where we enjoyed the luxury of two full days of deep reflections about networks science as applied to ethnography and anthropology at OICD. Sadly @amelia, who had set the whole thing up, could not be with us. Director Bruce White and researcher Emilia Groupp presented a theoretical model meant to detect and contrast what they call the "weaponization of identities" to fan the flame of conflict. When these guys say "conflict" they mean it: they are accustomed to operating in war zones. Places like Rwanda and Kosovo came up a lot.
I barely understand the model, but I am going to summarize it anyway here, for the benefit of discussion. Bruce and Emilia can always correct me. Here we go:
- In a "default" (non-weaponized) state, people have access to concepts, which are arranged into a network. From what I understand, this is a sort of associations pattern. For example, for Bruce "grandfather" might be connected to "Oxford", because his grandfather went to Oxford.
- This network is nonrandom. Starting from a sort of "root node", people access "versions of themselves" that correspond to certain subgraphs of the main concept network. For example, I might have an economist version of myself, with concepts like "consumer theory", "Hirschman", and "demand function", quite tightly associated to each other. I might also have a musician version of myself, with concepts like "accordion", "rhythm" and "touring", also quite tightly associated to each other. Getting from "demand function" to "touring" is harder than getting to "consumer theory". In network language, my associations graph is modular. Nodes communities in it correspond to versions of myself.
- Here it gets tricky. Part of my association patterns describe my identities, which will normally be multiple. I can identify as European, Italian, runner, economist etc. Weaponization of identity consists of external intervention that tries to limit my access to one or more versions of my identity. For example, I might be discouraged by a nationalist attacker to see myself as a European, and only see myself as Italian. Weaponization is recognizable because it denies people access to some versions of themselves.
The nice thing about this model is that it can immediately tell weaponization (evil) from other non-evil cultural influences: the former restricts access to someone's identities, whereas the latter extends it. It is never wrong to argue that you can see yourself as something else as well. For example, I could identify as a heir to the Roman empire who occupied Northern Italy since republican times; someone telling me that I might as well claim the legacy of the (later romanized) Celtic tribes that inhabited the area earlier does not constitute a weaponization of my identity, unless he tries to deny my ties with the Romans.
But this model is super difficult to operationalize for someone like me. We spent half the seminar asking pretty low-level questions: what constitutes data? How do we collect them? How do we interpret them? How do we define "links" in this context?
I enjoyed the time in Oxford. Bruce, Emilia and I were standing just at the edge of being able to understand each other, shouting to cover the distance across our disciplines and research styles. It is clearly a generative place to be, if an exhausting one. We are trying to schedule a followup involving @melancon.