Statistical Imaginaries: An ode to responsible data science

there doesn't seem to be any discussion of this article. But I feel obliged as someone who has worked with official statistics and data science to emphasise how much of a read it is worth.

one of my constant bugbears is even finding other data people who can and will take into consideration the context and uncertainty around the data itself, for in our societies and with our current levels of education and numeracy, there seems to be an bordering mystical connection between data, numbers, and authority. even beauty products, junk science, and conspiracies will often try to convince you with numbers.

and the struggle of how, as a data scientist and statistician, to use numbers in policy with largely innumerate policy makers and populations is an ongoing friction.

for an example on the inherent politics of numbers from my own country, observe the following from our official statistics office:

https://www.abs.gov.au/ausstats/abs@.nsf/mf/1370.0

note the date of last publication. this work was subsequently cancelled and will no longer be updated. I'm not here to say whether that was right or wrong, but merely to draw awareness to this tension.

meanwhile, while I was working in official stats, we were always extremely frustrated by the conspiracy theories around the likes of unemployment statistics or cpi. while all stats are inherently political, the amount of bad or invalid critiques of these statistics, as well as the implications that we were trying to generate then in some kind of conspiratorial bad faith was always disappointing.

Fascinating article. Video version of this talk is here: https://www.youtube.com/watch?v=kSdGG6wEE9k

Ugh. A lengthy rambling politically charged rant with a thinly veiled call to make science the handmaid of ideology. Not an ode by any stretch of imagination.

Signed, Samuel L. Clemens

By the way, Merry Christmas