A THREAD on the the divide between the "critical" and the "technological" communities, as a response to the paper by @davidjmoats and @npseaver which explores this topic. I'm posting the bits I highlighted and explaining why each of them struck me.
journals.sagepub.com/doi/full/10.11…
Ready?
journals.sagepub.com/doi/full/10.11…
Ready?
@davidjmoats @npseaver It starts with @mathbabedotorg's 2017 editorial. Apparently the critical community (not the best name, but let's go with it for now) was super surprised to hear that they were not critiquing technology. Me, I recognize Cathy's critique very naturally. The problem is...
@davidjmoats @npseaver @mathbabedotorg ...it's not that critical folks aren't looking at tech, but that the things they are saying are not often that useful for technologists.
Actually, Cathy's complaint is frequently echoed in tech. See this tweet -- which also proved very controversial.
Actually, Cathy's complaint is frequently echoed in tech. See this tweet -- which also proved very controversial.
@davidjmoats @npseaver @mathbabedotorg Here we agree -- it has been very challenging to bring the critical to tech. In fact I'm not even sure the "critical" is what needs to be imported. I'm personally a big fan of framing it as the "social", as @tristanharris also did. Sociotechnical is a wonderful word.
@davidjmoats @npseaver @mathbabedotorg @tristanharris This I found provocative. Now I'm no ethnographer but I was a journalist for a decade, so I find it easy to appreciate that ethnographic method is not merely "what data science isn't." There's a lot going on in qualitative research, and increasingly we have mixed-methods work.
@davidjmoats @npseaver @mathbabedotorg @tristanharris Seconded. Technologists talk about the social and political effects of their choices *all the time.* They have been doing so since the beginning of computers. Hell, there's a ton of CS literature on this from 1950s. See e.g. the Travistock Institute citeseerx.ist.psu.edu/viewdoc/downlo…
@davidjmoats @npseaver @mathbabedotorg @tristanharris Indeed. Technical methods first need to be understood on their own terms if they are to be effectively discussed from an outside perspective. In fact I'd say this about *any* field -- this is the old emic vs. etic. distinction. You need both!
@davidjmoats @npseaver @mathbabedotorg @tristanharris Seconded. Tech is a field of practice. Social scientists (etc.) will learn a lot from experiments -- especially if they are personally required to make the tradeoffs involved in shipping something. There are always economic, technical, and political constraints shaping tech.
@davidjmoats @npseaver @mathbabedotorg @tristanharris THIS THIS THIS. It has long seemed to me that this is the core premise of many critical scholars -- that their critiques should shape tech practice. But that may take a different sort of engagement with tech. It also depends on your theory of why tech has not already changed.
@davidjmoats @npseaver @mathbabedotorg @tristanharris Puzzling and solipsistic. Why use text analysis to engage with critical papers? Why produce an "analysis" at all, rather than a "product"? Alternatively, why not ask technologists to design something with a critical eye? That is the approach in my class compjournalism.com/?p=223
@davidjmoats @npseaver @mathbabedotorg @tristanharris Interesting to me because I would also frame the gap as, in large part, a problem of language and knowledge. I agree meaning emerges from practice... but a standard, core, essential part of tech practice is to define the meanings of terms! Akin to the logic of axiomatization.
@davidjmoats @npseaver @mathbabedotorg @tristanharris Tools shape what you find: yes, well known by technologists, thought doubtless others have gone deeper.
Different types of reflexivity: I think you're on to something. Reflexivity used to solve different problems, too.
Different types of reflexivity: I think you're on to something. Reflexivity used to solve different problems, too.
@davidjmoats @npseaver @mathbabedotorg @tristanharris Yeah. TBH I am not particularly interested in using text analysis to look at critical studies papers about data science either. No offense. I'd rather just read them -- or investigate design methodologies that incorporate these concerns. Like this: jonathanstray.com/papers/Socio-t…
@davidjmoats @npseaver @mathbabedotorg @tristanharris Was that really your aim? Seems an instance of the disciplinary hostility we are here investigating.
OTOH, such people do exist -- or at least, the stereotypical behaviors do appear from time to time, though perhaps not all in the same person.
OTOH, such people do exist -- or at least, the stereotypical behaviors do appear from time to time, though perhaps not all in the same person.
@davidjmoats @npseaver @mathbabedotorg @tristanharris Again, I don't understand the logic of using computational tools to analyze texts critical of computational tools either. It feels to me like more of a party trick.
Perhaps the authors can explain their thinking in choosing this approach?
Perhaps the authors can explain their thinking in choosing this approach?
@davidjmoats @npseaver @mathbabedotorg @tristanharris I found this fascinating, and one of the most "aha!" moments of reading for me. I mean OF COURSE critical scholars are not monolithic. No field is.
It also strikes me that ethnography of how the "gap" plays out could be very worthwhile.
It also strikes me that ethnography of how the "gap" plays out could be very worthwhile.
@davidjmoats @npseaver @mathbabedotorg @tristanharris What is the nature of the divide between the technical and the non-technical? It is a many splendored thing.
@davidjmoats @npseaver @mathbabedotorg @tristanharris Look: you can get flamed on Twitter by the "other side." Not a lot of trust either. I don't know about war metaphors but there's definitely a conflict here.
@davidjmoats @npseaver @mathbabedotorg @tristanharris Sigh. "Every time I fired a linguist, the performance of the speech recognizer goes up," as the famous quote goes.
My take is that ML/NLP reveal fundamentally new empirical aspects of meaning, the humanities and social sciences have not really integrated these insights.
My take is that ML/NLP reveal fundamentally new empirical aspects of meaning, the humanities and social sciences have not really integrated these insights.
@davidjmoats @npseaver @mathbabedotorg @tristanharris I too will add: very meta.
However, this does point back to the desire of technologists to define terms before talking, as opposed to just "letting meanings emerge." (In reality, of course, meanings DO emerge, but that doesn't free you from the need for formal definitions.)
However, this does point back to the desire of technologists to define terms before talking, as opposed to just "letting meanings emerge." (In reality, of course, meanings DO emerge, but that doesn't free you from the need for formal definitions.)
@davidjmoats @npseaver @mathbabedotorg @tristanharris I would put it this way:
Technologists are not really interested in themselves -- the method should work no matter who uses it. They are, however, extremely interested in getting reflexive about their own tools.
Contrariwise, researcher subjectivity is core to ethnography.
Technologists are not really interested in themselves -- the method should work no matter who uses it. They are, however, extremely interested in getting reflexive about their own tools.
Contrariwise, researcher subjectivity is core to ethnography.
@davidjmoats @npseaver @mathbabedotorg @tristanharris I agree that anthropologists are often "somewhat marginalized" in comparison to technologists. I don't think this particular power relation exempts anthropologists from getting emic. (And yes, I know, technologists need to understand "the other side" too.)
@davidjmoats @npseaver @mathbabedotorg @tristanharris I mean, you know, language only matters if you want to communicate.
@davidjmoats @npseaver @mathbabedotorg @tristanharris Another key insight. I might come at it this way: yes, technologists know that definitions, quantification, abstraction etc. limit their thinking. These things are also completely necessary for quantitative methods -- or if you want a machine to do anything at all.
@davidjmoats @npseaver @mathbabedotorg @tristanharris Right. The reality is that there are many people working across tech/social or tech/humanities or tech/critical or whatever. I'm one of them, and there are many others. Yet I think all hybrids would recognize the depths these divides, and experience tiny battles and stereotyping.
@davidjmoats @npseaver @mathbabedotorg @tristanharris Yes. To the extent that critical folks want to influence tech folks, they will need to be fluent in tech language (and BTW "bias" has radically different meanings in different communities.) This does not excuse tech folks from studying things outside their field.
@davidjmoats @npseaver @mathbabedotorg @tristanharris Ok that's all I got.
I hope you enjoyed my engagement with @davidjmoats and @npseaver's engagement with the data science community.
I too enjoy a good meta. But more importantly I think this sort of hybridization is desperately needed
/fin
I hope you enjoyed my engagement with @davidjmoats and @npseaver's engagement with the data science community.
I too enjoy a good meta. But more importantly I think this sort of hybridization is desperately needed
/fin
