Ep. 162: How the "Data-Driven" Label Sanitizes Cruel Austerity Politics

Brilliant. I'm a little late to this one but it's even more gripping and insightful than I even expected

V K

I work with data in, as another commenter put it, my capitalist job. It's just the very ordinary sales and shipping type information any company uses to check up on itself. I've been tasked to bring order to a level of chaos that is years in the making and utterly impossible to describe - the result of a bunch of different people making their own datasets and reports without any oversight and then subsequent people taking over without really understanding what they were looking at and adding more complexity. My role is data governance and I'm supposed to be riding herd on the data analysts, and the last 8 months of this has made me really think about this a lot. The number one thing that surprises me about the data analysts I work with, and all the users really, is the lack of curiosity. People aren't looking to understand but to complete a task. And second, it is so easy to make really simple and really horrible mistakes. And it's really hard to explain to people why such a mistake matters. "The cost you are using changes every day; it's not an accurate record of the cost at the time you did cycle count." "I'm not interested in making changes to the report." - real conversation I had about a data issue. It doesn't matter that much if some VP doesn't want to listen about whether his cycle count cost reports are right, but it has given me such a different perspective on this stuff when it's used for other stuff. Sometimes it's super not obvious. I spent a week trying to figure out why a forecast number wasn't working the way it looked like it should be working, and it was because the way the calendar was written and the way the business intelligence software language worked interacted in a way that what it looked like the calculation was doing wasn't what the calculation was doing and so the forecast was bullshit, and always had been. But it took spending about 40 hours of my time tearing my hair out trying to figure out what was wrong to see it. And that only happened because I was trying to recreate the report and couldn't get it to give me the same numbers. Which is why it's so so so scary that we have incurious, over confident data analysts deciding people's lives and futures based on bad collection, modeling, analysis, reporting and understanding of data that often isn't relevant to begin with.

Deborah Bell

Excellent episode again. Irish Twitter was and is awash with 'data' supplied by economists in service of the 'get back to work' cohort in the midst of Covid. They also constantly work for the propagation of a narrative that dictates that housing/homelessness problems are only solvable by capitulating entirely to the interests of developers, meaning that standards and regulations should be done away with. They prove their points via graphs and 'studies' with provenance as dubious as that referred to in this episode.

Ciaran Colley

I remember an early data-driven austerity program that IBM was involved with Edit: Datarade is better Edit2: In an episode about laundering ideological conclusions in ostensibly data-driven trappings, I would have expected a little more self-awareness when committing the same basic offense -- and a little more humbleness with respect to the epistemological limitations which regularly make fools of all of us. Edit3: I don't think the way the guests were applying the Precautionary Principle really stands to scrutiny. Related: https://twitter.com/WesPegden/status/1506662420462321672/photo/1

greg

I am profoundly disappointed in you, Mr. Johnson. You didn't say this was a "spiritual successor" to one of your previous episodes! What even is Citations Needed episode if not a spiritual successor to other episodes!?

newdarkcloud

fixed, thank you!

Citations Needed

Good stuff, looking forward to listening. Heads up the "go Here" at the bottom link for transcript is broken. Sorry, i work QA in my capitalist job and can't turn it off. Cheers comrades.