MO’ MOANIER MOANS

Medium’s Algoholic Stupor

How about a fit-for-purpose distribution system for the New Year?

‘Anger is an energy’, sang John Lydon. Some time between the Sex Pistols and the butter adverts (Google it and weep — and no, there are no undertones of Last Tango in Paris, before you ask).

So if Johnny Rotten’s version of psychophysics is indeed correct, I should be feeling pretty buzzy this January. Or Moanuary as I’m thinking of renaming it. Sod resolutions — spend the whole first month of the year griping about everything you hate, get it out of your system, then start doing something about it in February.

Which also has the advantage of being a couple or three days shorter. More expensive pound-for-pound monthly gym membership. But fewer actual days of sweat and Lycra. Swings and roundabouts. Or should that be treadmills and ellipticals?

But I digress.

‘This Is Not a Love Song’ — that was another one from Mr Lydon’s PiL era. And continuing the fine tradition of apophasis, This Is Not a Moan About the Awful Boost Programme.

For once.

It is instead a moan about the awful algorithm.

A ‘system’ (if that’s not overstating it) that is every bit as focused, coherent and consistent as this rambling text has been so far.

Henceforth, though, I shall dispatch my laser-guided bombs direct to their target, with arguments of military precision and impact.

What the actual fuck is going on with that shit in [checks date] erm, 2004? Oops, my bad: 2024.

‘Show me less like this,’ I say.

‘OK. Here is the exact same thing every day for the next month, and half a dozen other practically identical pieces by the same author — you’re welcome!’ it says, like ChatGPT’s infinitely more annoying country cousin.

‘Recommended for you!’ it breezily announces.

‘Really? Could you show your workings for that?’

‘Sure! Some person that you once followed 5 months ago but unfollowed a month or so later highlighted the kicker.’

‘Wow!’ I say.

Can’t argue with logic like that. All you can do, in truth, is hit your head repeatedly against a brick wall until someone arrives in a white coat with a stretcher, straitjacket, or ideally both.

Some open questions (because maybe it’s just me and I’m moaning about nothing — wouldn’t be the first time):

• Do you find the ‘For you’ tab anything other than useless?
• Do you find that the ‘Following’ tab is weirdly inconsistent with ‘For you’, and also completely clogged up with stuff from pubs/editors that you were obliged to follow months ago just to submit a piece, actually have no interest in, but can’t face the soul-destroying scroll-and-click of unfollowing?
• Are you mystified by the fact that the list under a ‘Topic’ tab in your feed is packed (in random chronological order) with pieces from the days when Medium must have been called Miniscule and was still shitting its diapers?
• And yet the other sub-feeds refuse to show anything older than last Tuesday, despite the fact that you know damned well there is a ton of content by writers you have read, highlighted, commented, clapped, followed, engaged, married and generally made the virtual beast with two backs with, that remains hidden from you? 1984 memory-holed.

If your answers to all/any of the above are ‘Yes’, would you agree that the Medium distribution algorithm is woefully unfit for purpose in the current age? By which I mean ‘since the death of the UK’s beloved Queen Elizabeth’. The First, that is.

Now, I’m not one to gripe unreasonably.

Seriously.

Maybe the Medium algo was a bleeding-edge, state-of-the-art chunk of gleaming code when the platform was first launched by that Twitter guy as ‘kind of like Twitter — more long-winded but less of a dogmeat circle-jerk’ (I believe that was the <100 character elevator pitch).

But I would tentatively suggest that it has perhaps seen better days. Decades, even. And might charitably be put out to pasture, given a divorce settlement and replaced with a younger model. And no one’s going to convince me the tech bros of Silicon Valley don’t know how to do that.

‘But how?’ the puzzled masses cry.

‘Gee,’ I reply. ‘Maybe that AILLMGPT thingy might have some ideas.’

I appreciate that refining an algorithm to reliably recommend lengthy articles, all with their own nuance and voice, is a more complex endeavour than shovelling an endless stream of cat videos down the Cookie Monster maw of mindless TikTok scrollers.

And we need only turn to the Netflix concept of ‘You liked X — you might like Y’ for an example of what would most definitely not be an improvement. Though to be fair, would it be any worse?

But the point here is that the currency of Medium is the written word. Precisely the coinage that the LLMs of this world, now available in every bespoke flavour from Rip-off Raspberry Ripple to Plagiarism Pistachio, pour into their pockets like kids in an untended sweet shop.

Are we seriously to believe that it is beyond the wit of software engineers to devise a system capable of analysing articles — from whatever year — and matching them up to the ones we have registered as being of interest to us?

We could even have a simple ‘include this article in my feed recommendation analysis’ button on each one, so as not to overload or confuse the little imps — base it only on a narrower selection of starred articles. (This could more snappily be labelled simply as ‘Analyze this!’, with de Niro doing the text-to-speech.)

Rather than basing it all on five measly tags. Which Melvil Dewey would have found embarrassingly inadequate when he was working on his library classification system a century and a half ago.

Come on, guys. This is the era of AI, isn’t it? Half the platform is an Augean stable (stall mats not included) of its reeking manure — you can’t claim you’ve never heard of it.

Get with the program. Literally.

[Rant ends — thanks for playing.]

And for more Medium moaning, this time in verse: