Matthew Garvin

I don’t think we can have ethics without feminist ethics. However, let me preface this by saying that I mean intersectional feminism. Intersectional feminism asserts that all aspects of socio-political identities overlap.

I was glad to see Alison Adam and Jacqueline Ofori-Amanfo dragging questionnaires and surveys just a little bit, while also advocating for ethnography and participant observation. And I also want to mention their criticism of the Kreie and Cronan paper because it highlights an issue that’s kind of been a recurring theme in my reading responses.

Outside of anthropology, social sciences like psychology and economics have a replicability problem. Much of psychological literature is based on studies conducted on Psych 101 undergrads at universities in N. America and Europe during a period of time when the student body was almost exclusively comprised of young white men. Many psychologists believe that they can study human behavior as something distinct from socio-cultural environmental habituation. But if you happen to have read my humans as cake, culture as eggs analogy from a few weeks ago you’ll know that in my view, you can’t distinguish between the two, the eggs (culture) are baked in and part of the cake (human).

In Emily Chang’s, “Brotopia” she talks about how women face barriers into the tech industry because of rampant misogyny and toxic masculinity. It’s somewhat similar throughout the business world. A lot of relationships and business deals are forged through social gatherings. And when these techbros grow up as introverted incels and then are suddenly billionaires, they host these extravagant gatherings, making deals in the hot tub surrounded by the kind of beautiful women that rejected them all their lives previously. Female entrepreneurs don’t usually get invited to these events and when they do, they tend to find it problematic as they often get propositioned by the men or treated as just another one of the women at the party.

There’s a software company in the UK that recently announced they developed an anti-shoplifting AI that is 99% effective at identifying shoplifters before they shoplift. They trained the ML algorithm by hiring actors to pretend like they were shoplifters. So these actors perpetuated the stereotypes that were given to them by a veteran security guard, and I think it was even like just 2 criteria. Trayvon Martin was murdered because of a stereotype that existed in the head of George Zimmerman. How do we know that these stereotypes passed to the actors from the security guard don’t hold a similar bias?

The problem with statistics is that it deals in probability, so all you can ever say to some degree, is probably. We have to remember that correlation doesn’t necessarily mean causation. P-hacking is a big issue in science, where we can use software to find correlations that achieve that p-value. In a rush to publish or perish, in a profit driven system where academic journals don’t value failure, in a research culture that seeks to reduce everything to Occam’s razor, weak behavioral correlations are rampant and often reinforce racial and gender stereotypes promoted by white supremacists. Because if you already believe the stereotype, then a p-hacked correlation can easily fit the narrative you’re trying to promote.

I found myself particularly interested in the probing method described in Boehner et al., especially the camera. As mentioned in the readings, surveys, questionnaires and even interviews can be problematic in regard to acquiring good data. Self-reporting is something, to be certain, but it’s of limited value because the person can lie or even unintentionally misrepresent their own reality. What people do, what they say, what they think they do, and what they say they do can all vary dramatically. And often times the more a person wants to be helpful, the more inaccurate the information they provide!

I think probes can be very useful in giving the subject opportunities to contribute within their own time and space. I neglected to mention as I ripped psychology in my response above, that a lot of studies are done in a lab setting. While this lends an air of credibility to the researcher, it removes the subject from the context of their own lives. I think it was the Harvard psychologist, Timothy Leary, famous for popularizing LSD in the 60’s, who noted that lab settings and sterile white coats make people nervous, and it primes them for a reaction to the environment rather than the stimulus, the people he tested had bad trips until he made his lab more like a living room.

I have a hard time opening up to an interviewer, but if you gave me a camera and told me to take a picture of the spiritual center of my house, that’s another story, and something you could really gain useful information from, especially from a design point of view. Similar to how when conducting a contextual interview, it’s important to see the workspace as the subject sees it, without them cleaning up first.

I also just want to take a moment to reference The Triumph of the Default. Although I alluded to it previously, we have these tech people deciding what the default is in many cases. And because of that choice overload, many people conform to it even though it may not be adequate for them. Kind of like in Don Norman’s DOET, where he discusses how users think it’s their fault if they aren’t good with technology. Instead Norman says, blame the designer. Just before the semester started, Kentaro published this article  about holding moratoriums on technology until it can be fully vetted for things like bias. As they say, hindsight is 20/20. We often can’t see the ramifications of our actions until we look back in reflection. A better approach is to take our time and carefully scrutinize technological innovation with diverse perspectives. Without research and training, I can’t imagine how some invention I come up with can adversely impact a disabled person living on the other side of the country, or even in my own neighborhood, because their experience is not my own.

Default states, in general can be very useful in combating choice paralysis, because believe it or not, there is a threshold to which choice is useful. But default states can also reward bias. For example, earlier I mentioned an anti-shoplifting AI trained by actors pretending to be shoplifters. If they were just acting out stereotypes from the security guard who happens to mostly catch young black men because that’s who he’s most suspicious of, then the actors just trained the AI to be suspicious of young black men by default. And if we believe that this AI is actually 99% accurate, then it would provide statistical evidence to support the racist notion that young black men are more likely to be criminals. But this doesn’t even touch on the fact that black men don’t have the same opportunities to participate in the formal economy. When you aren’t allowed to participate in the formal economy, you participate in the informal economy to survive. That’s what most theft is. Never mind that due to wealth and income inequality, the police were formed to prioritize and protect the property of the haves over the lives of the have-nots.

Featured Photo by Duncan Shaffer on Unsplash

References

Boehner, Kirsten, Bill Gaver and Andy Boucher. Probes. Chapter 14 in Inventive methods: The happening of the social. Lury, Celia, and Nina Wakeford, eds. Routledge, 2012.

Kelly, K (2009) “Triumph of the Default” http://www.kk.org/thetechnium/archives/2009/06/triumph_of_the.php

Bennett, Cynthia L., and Daniela K. Rosner. “The Promise of Empathy: Design, Disability, and Knowing the “Other.” (2019)

Shinohara, Kristen, et al. “Tenets for Social Accessibility: Towards Humanizing Disabled People in Design.” ACM Transactions on Accessible Computing (TACCESS) 11.1 (2018): 6.

Adam, Alison, and Jacqueline Ofori-Amanfo. “Does gender matter in computer ethics?.” Ethics and Information Technology 2.1 (2000): 37-47.

Winner, L. (1986). Do artifacts have politics? In The whale and the reactor, pp. 19-39. Chicago: University of Chicago Press.

Nissenbaum, H. 2001. How Computer Systems Embody Values.