And then...

2) "Economic emergencies: exception, government and the management of the economy" by Egor Makarov and Simone Polillo - a useful conceptual map of Foucauldian and Schmittian approaches to emergency and exception.

www.tandfonline.com/doi/full/10....

We chat about two papers:

1) "Undoing Cultural Studies: Cultural Economy at the Open University (1979-1997)" - a part-history of "workshopping" at the OU, associated with Stuart Hall and Doreen Massey (who I refer to as Jamesonian "vanishing mediators").
www.tandfonline.com/doi/full/10....

There's something about the specificity of "cultural data" which is interesting and last.fm is for sure an interesting example. They have 20 years of quite a few users' self-tracked listening across multiple platforms - but they can't seem to make it profitable? Amazed Paramount hasn't shut it down.

...not to mention "synthetic data" and increasing layers of data-about-data abstraction. But yes that's exactly the question I was trying to formulate: the "coming crisis of empirical sociology" has evolved into epistemic permacrisis so can anything be salvaged?

Thanks, yes it looks like there's a few interesting methods. Also longitudinal changes (by now Spotify has c.17 years of some users' data). Cross-referencing as you say. Fine-tuning. I just don't know how much of this they - or other similar platforms - actually experiment with and to what end.

I think it's exactly this disappointment I'm trying to process.

Oh yes, in Spotify's case for sure. But that's my point. I find it annoying because it's a tedious marketing gimmick - but there is some kind of promise of something potentially more interesting behind it, "revealing" patterns we wouldn't normally see at a collective rather than individual level...

ChatGPT is giving me metrics such as "burstiness" which I assumed is just the sort of rubbish it would make up but apparently not.
en.wikipedia.org/wiki/Burstin...

Put differently, I listen to Celine Dion *in a different way* to how I listen to whatever is currently cool among East Londoners, or what I listened to when I was 18, or what I want to share with my kids - and I assume that the behaviouralists are interested in measuring this.

Who doesn't want that?

I am just idly wondering what techniques, in the "age of big data", are used by economists, comp-sci people, statisticians, psych- types etc for getting at a more interpretive or contextual level of depth of "preference" than simply duration, frequency or expenditure.

I was going to say I'm looking more for quote-unquote "data driven" techniques and it's hard to imagine what this looks like at scale - but with ChatGPT I guess we do indeed now have data-driven psychoanalysis.

ChatGPT is giving me metrics such as "burstiness" which I assumed is just the sort of rubbish it would make up but apparently not.
en.wikipedia.org/wiki/Burstin...

...by which I don't mean consumer ethnography. I mean proxies for moments of qualitative experience that measure for depth or intensity or exceptionality.

Hmmm. What's more confusing to me is that I've never had a "genuinely interesting" conversation with an LLM. I've found them useful and I've even learned some stuff. But for the most part it's deeply annoying. Which I guess is, at least in part, why I'm sceptical that it *makes* us do anything...