The problem with statistics
[Podcast Reaction] Banana Data : Ethical Implications of Humanizing Your Data
When I first read "humanizing data", I definitely thought it was some idea of personifying or respecting data—which it both is, and isn't.
The humanization of data refers to understanding the human relationship to the data, both in implication (what is this data saying about those creating the research) and in interaction (how is this data going to affect us?). Basically, data science isn't as objective as we think it is. And I think anyone who has taken courses in critical thinking or any research field shares this understanding.
The podcast makes a few claims:
1. It's important for those researching and publishing the data to be transparent and honest about the limitations of the data: This I largely agree on. And I can't help but feel some sort of internal monkey-rage when I see people citing studies that have no business being an accurate representation of what is being discussed.
I'm not here to stand on a soap box and pretend like I'm some "woke" better-than-thou person. But I also know that, prior to studying research related fields, I literally took any data-based chart as bible. And because I've been on both sides, I understand that for anyone who hasn't dipped their toes into academia, they will see the words "research shows" and start spreading those claims like a COVID infection in a maskless movie theatre (too soon? I've had COVID so I feel like I can make this joke).
2. Researchers/Companies have a moral responsibility to the distribution of their data and its effects: Part of this claim is about data security and potential misuse (either by the actual company or by third parties), but I think the more interesting part of this claim lies here; if the result of the data will actually have a negative social impact, is it morally appropriate to withhold the data? Or do we publish it in service of "the truth" or "technological advancement"? And if so, what reparative measures are we conducting to reduce social harm?
This starts to go deep into the territory of ethical AI/technology. When do we stop? How do we measure harm vs. gain? I mean, we've already crossed the line in so many different fields—from deep-fakes spreading misinformation to addiction/attention-mining—when is it enough? And that's not a rhetorical question; I mean it earnestly. And I think it opens up to a new question: What should the distribution of technological responsibility look like? So basically; whose fucking fault is it, really?
And like everything, I don't think there's a complete black or white answer to this. It's probably a balance of both: companies/researchers need to be responsible for how their data is used and presented, AND consumers need to be educated enough to make accurate inferences. So, that's my one gripe with this podcast episode; it put all of the weight of data ethics on the distributors while not holding the population accountable.
And I'll admit, it sounds like I'm victim blaming. To clarify, I'm not saying we are solely responsible for misinformation; obviously, sources that present themselves to be trusted need to deliver off that promise (we treat news from TMZ vs. news on Scientific American very differently). But I do think it's part of the solution; we need to make education more accessible, and figure out how to create cultural value for critical thinking.
Because honestly, being an intelligent consumer (mentally and physically) does not help capitalism. And if it doesn't sell, then it's off the market.
