Slave To The (Algo)Rhythm

Looking for more articles like Daniela Cammack’s ‘Plato and Athenian justice’? You might try M.B. Foster, ‘On Plato’s conception of justice in the Republic’, The Philosophy Quarterly 1.3 (1951) and Stella Lange, ‘Plato and democracy’, The Classical Journal 34.8 (1939). If you enjoyed Pantelis Michelakis’ ‘Naming the Plague in Homer, Sophocles and Thucydides’, you might also enjoy Masen J. Williamson, ‘Thucydides’ Plague, a narrative aggressor’, Brigham Young University ProQuest Dissertations (2021) or Andy Coghlan, ‘Cheap, safe drug kills most cancers’, New Scientist 193 (2007).

Yes, it’s been quiet on here recently; partly due to the fact that we are, painfully soon after the last time, living with a terminally ill cat and so lacking in energy or emotional bandwidth, but also I’ve had quite a lot of marking to do (early in the term, because it’s students from last term who’ve submitted the revised post-feedback versions of coursework). One of the things that struck me, perhaps because it coincided with an exchange on Bluesky, is that I’m still having to comment on students using deeply obscure and/or absurdly outdated publications as key sources, despite the fact that I’ve expanded the explicit guidance in my module handbooks about using random stuff found on the internet. There is a regular temptation to post in-text feedback comments like ‘SERIOUSLY?!?’ on certain bibliographies.

My sense – yes, this is a topic where more substantial research might be useful, if I had any time – is that this practice has in the past been primarily driven by students doing a general search for an essay title, or at least for very specific key words. I can think of at least three possible reasons for this: a desire for scholarship that is directly relevant to the specific topic (even when they might be far better off relying on relevant sections of more general publications, like the ones in the module bibliography); a belief that finding material beyond the reading list is ipso facto good (I’ve certainly seen assessment criteria that emphasise ‘independent research’); and the increasing uselessness of search engines, so that if you don’t have an incredibly specific query you’re going to have to wade through a deluge of irrelevant crap to find what you’re looking for. Hence the alarming habit of citing undergraduate dissertations from mid-western universities – WHY are these all over the Internet? – as authoritative sources. To be fair, final-year students do this a lot less, presumably because it dawns on them that such dissertations carry all the academic authority of the bloke sitting next to them in class – first- and second-year students clearly have an elevated respect for those slightly more senior than them.

What we are also seeing, and I’d be interested to know how long this has been the case, is publishers seeking to drive further traffic to their own publications by providing Spotify-like recommendations when you view an article online. I’d managed to blank this entirely until quite recently, but it offers a clear explanation for the association, in recent coursework, of recent stuff that I’d recommended and very old stuff that I definitely hadn’t, all proper academic publications but some very definitely marginal to the current state of the debate. My snarky first paragraph above in fact records precisely the articles that were recommended to me when I checked the pieces by Cammack and Michelakis as an experiment.

What’s suggested is relevant, and scholarly (for rather loose definitions of ‘scholarly’), but entirely useless in practical terms. I’m actually a great believer in disappearing down rabbit-holes or following chance leads. Much of my research over the years has involved finding interesting-looking things in the references of a book or article and retracing the development of an idea, often leading to whole new areas of study and unexpected new questions. So, I can appreciate the potential pleasures of looking up, say, Raaflaub on democracy, oligarchy and the concept of the free citizen, and being offered a selection of more recent ‘related articles’ in contemporary political theory, Peace Studies and anthropology – it’s like being able to follow the trail of breadcrumbs forwards as well as backwards in time. But of course I do this only if I know I have the time to follow lots of different paths and read loads of stuff, much of which might turn out to be irrelevant to my current project, and in the knowledge that I can usually judge the potential usefulness pretty quickly – sometimes just by looking at the details of the journal and the title. What I am not going to do is rely on these chance discoveries as one of the five or six key sources for what I’m working on.

My basic advice to students on the use of older scholarship (and by coincidence I wrote something like this on the Ask Historians Reddit the other day): it’s not that it’s worthless, but it’s probably worthless to you. The further back in time a thing was written, the more you need to consider it in its intellectual context, to appreciate not just how it’s out of step with contemporary scholarship but also why – and for that you need a pretty extensive knowledge of the current scholarship, in order to evaluate it – rather than being able to side-step the current stuff. It is not that older scholarship is only relevant to the history of historiography – but you need to do a certain amount of history of historiography in order to make use of it, and you probably don’t have time for that. If an older piece is actually essential, I will generally have told you about it already.

What this immediately reminded me of was the ‘if you liked that, try this’ algorithms of music platforms, and it’s interesting to reflect on this resemblance. The digital revolution has made available, at remarkably low cost (at least if you’re a university student with access to a decent library, as the equivalent of having an Apple Music subscription), a vast resource of much of what has ever been produced; whereas once, to access something old and relatively unregarded, you’d have to search through card catalogues and then look through dusty old volumes in distant stacks, or rummage hopefully through boxes of second-hand records, now it’s just there on your computer, a few clicks away.

In the olden days, of course, you’d be looking for something specific because you’d seen it referred to elsewhere (a shout-out to everyone who remembers combing through back issues of l’Année Philologique!) – and if that hadn’t already impressed on you the fact that it was Old then the process of accessing it surely did. Today, it has been suggested that Most Young People consume music in an eclectic manner, without much sense of its history or traditions or development (let alone the powerful tribal sense in my youth that if you liked X you shouldn’t like Y) – which is fine, if you have no interest in those things but just want to enjoy the music, but may be an issue if (a) you would like more of a sense of development rather than just an undifferentiated mass, (b) you start to wonder if you can really trust the algorithms, and/or (c) you fail to realise how much of contemporary pop and rock is a pale simulacrum of the genius-level creations of the early 1980s, and therefore are incapable of making proper aesthetic judgements.

This decontextualisation is obviously more of a problem for history students, where a sense of the historical development of ideas and arguments is a major part of the point* – and the operations of publisher websites do seem to be encouraging it. But the real question around music algorithms is whether they are really directly you towards things you should really like based on your listening history, or whether they are actually pointing you towards things they think they can sell you. There’s an enterprising Norwegian pianist who has clearly paid Facebook a load of money to push adverts for his pleasant but undemanding trio records at anyone who shows signs of an interest in jazz (yes, reader, I bought one); the Spotify approach is much more subtle, as of course you’ve already paid – so the suspicion is that you may be directed towards artists from whom they will make more money, not least the ones who are actually wholly-owned gAI products. Corollary: there’s a load of great music you are not being told about because it won’t make them money.

What’s going on with academic publishers? Simple: I assume that numbers of page views on Open Access articles increasingly drive the whole ecosystem, from the charges that can be levied on authors to, rather more significantly, the terms of Read & Publish contracts with universities. They’re not going to point you towards another publisher’s journals (students, I suspect, see each journal as a separate enterprise, rather than being part of a stable, and so imagine that because the suggestions are all from different journals they must be drawn from the whole world of scholarship), only stuff with relevant keywords in their own database, and they don’t actually care whether the recommended article will be any use, they just want you to click on it.

How much does this actually matter? I’d be interested to hear views… We may at some point identify significant changes in research, reflecting the younger generation’s radical eclecticism and obliviousness to traditional narratives of disciplinary development – projects inspired by long-neglected ideas and approaches rather than by the Important Male Figures who dominate conventional historiographical accounts, or revivals of unfashionable styles that may or may not be ironic (I’m looking forward to a new Finleyite primitivism wowing the crowds at Glastonbury next year). Or, for the doom-sayers among you, the spread of an increasingly random, decontextualised sense of scholarship, in which everything is tailored to individual preferences. My feeling is that at the moment it’s just a mild irritation for me as marker and a potential pitfall for students that can be addressed by some tweaks to the guidance I give them – so long as they read it, rather than trusting the algorithm.

*Oh, and don’t get me started on the fact that eBooks get given the date of electronic publication not original publication, so the priority of different interpretations gets hopelessly tangled…