Webcurios 14/04/23

Reading Time: 37 minutes


Ahem. Are you well? Did you manage to avoid either crucifixion or the acquisition of chocolate-related type-2 diabetes? GREAT!

I am in something of a hurry what with having to get to lunch and it being 1155am at the time of writing and my still being in my pants and needing to get something of a move on with the washing and the dressing and suchlike – so, er, this is all you’re getting by way of an intro. Sorry.

I am still Matt, this is still Web Curios, and you and I both know that the opening few paragraphs are always the very worst of each week’s edition.

By Miloš Ilić



  • The HTML Review: This week, as a special treat for having survived without links for a whole fortnight (I refuse to believe there are other sources of links, don’t lie to me) we kick off with something GENUINELY GREAT! This is the second edition of the HTML review, and once again it’s a collection of experiments in form – the letter from the editor frankly explains the ethos better than I ever could, so, er, here: “Our 2023 issue is made up of 17 contributions that span modes of digital literature and experiment. We have poetic instruments, interactive fictions, illustrated essays, movable lyrics, linguistic gardens, and pixelated memories. I like to think each piece we publish is a specific illumination of how literature can exist in digital spaces. They are each reflective of a World Wide Web built with fingerprints visible, inviting all to participate and take ownership.” These are SO SO LOVELY – fine, ok, so I have a bias here as I am a particular sucker for anything that combines code and narrative in interesting ways, but even if you’re less of a sucker for ‘art made of words’ than I am you will, I promise, find things to love here – from click-to-advance poetry that really makes you feel each mousepress to a perfectly-crafted homage to messageboard conversations on web1.0, to the weird but very very beautiful semi-organic landscape of the html gardens, this is all just beautiful and, quite often, really affecting (again, I have a particular emotional weakness for this sort of thing, but I promise that if even someone as basically dead inside as me can find something to love in here then you can too). Practically perfect in every way, I promise you, and if you’re finding everything a bit much this week then you could do worse than just stopping at this link and just hanging out a bit.
  • Cube Fashion: What shape is the fashion industry? No, come on, it’s a serious question – WHAT FCUKING SHAPE IS IT? Know that if you answered anything other than ‘it is a cube, Matt; fashion is a cube as any fule kno’ that you were WRONG – still, you can atone for your ignorance by exploring THE UNIVERSE OF FASHION THROUGH DATA, thanks to this project by Google and Vogue (which, shamefully, is actually from October last year and which I am hideously late to and which, as I type, is causing the sort of hot/cold embarrassed sweat flushes that I associate more with childhood sporting humiliation than I do newsletter-writing) which presents the ENTIRE ECOSYSTEM OF FASHION BRANDS as, er, a series of datapoints arrayed in a vaguely cubic fashion to demonstrate each of said brands’ relative performance against a variety of different criteria (geographical footprint, green initiatives and, er, ‘corporate communications’, to name but three), letting you see a UNIQUE DATA-DRIVEN CUBIC DESIGN for each of the stars in the great fashion galaxy. And it tells you…well, it tells you the square root of fcuk-all, turns out, given there’s no scale or explanation as to how you’re meant to read any of this stuff, or any depth to the data displayed (it’s just fancy graphing, at heart), but it looks pretty and that’s the main thing. A bit of digital work in the fashion space that looks pretty but is fundamentally empty and devoid of meaning? This stuff writes itself, I tell you.
  • Bohemian RhapsodAI: Leaving aside the horror of the name – you know what they’re doing with the wordplay, but just take a moment to pause here and say the title of this website out loud;it’s genuinely unpleasant to vocalise, for reasons I can’t quite articulate – this is partly a nice piece of eyecatching digital promo work by some Dutch agency, and partly…look, I don’t want to pretend that I am some sort of massive Queen or Freddie Mercury fan, or that I have particularly strong feelings about THE SANCTITY OF AN ARTIST’S MEMORY, but at the same time, er, this sounds HORRIBLE! Basically this is at heart the sort of ‘run a bunch of different isolated vocal tracks together based on user inputs, using maths to try and make it all sound vaguely-coherent’ you’ve seen before, but with an added layer of AI-wrangling so that you can get a choir of machine-imagined Freddies warbling together as they sing all the different parts of Bohemian Rhapsody – did they have to make it sound quite so much like it’s being sung tone-deaf people who hated his music, though? If you know, live with or work with any big fans of the band and their music, this will REALLY upset them.
  • Welcome Home: THIS IS SO GOOD. Like an ARG but, as far as I can tell, without attempting to sell you anything or get you to do really, really hard maths, this is a web portal into a fan community for forgotten mid-20thC kids TV show ‘Welcome Home’, a puppet-lad show designed to educate and amuse small people in much the same way as Sesame Street. “Welcome Home  was an American children’s television program created and produced by  The Playfellow Workshop, which served as the studio’s only production. Supposedly its first episode aired on October 11th, 1969 and was broadcast onto an unknown channel until it’s last estimated air date sometime in 1974” – or at least that’s the ostensible explanation. There’s something…off about this, though, and the longer you look at the pages the more you realise that this has far more of the vibe of “Don’t Hug Me I’m Scared” than, say, Rosie & Jim, and that there’s something not quite right about some of the images, and that not all the links go where you expect them to go…I don’t want to spoil this too much, but I strongly advise you to spelunk and explore and click and keep your soul clutched tight in your chest as there is a LOT in here – also, I have just decided that ARGs are due a comeback so if you could all please start working on some that would be great thanks.
  • Inside The Great Pyramid: I feel a bit guilty saying this, but I’ve never been hugely interested in visiting the pyramids – I do wonder whether this is some sort of inbuilt Italian bias whereby my Roman ancestry means that whenever I look at any ancient architecture that isn’t the Palatine I can’t help but make the sort of dismissive hand and mouth gestures immortalised by generations worth of cinematic stereotypes – which is why I was so grateful for this new bit of Streetview-esque webwork on the Mused platform which lets you explore inside the (VERY CLAUSTROPHOBIC) corridors of the Great Pyramid at Giza and see what all those explorers were so keen to get their mitts on. This is genuinely fascinating (although also…quite aesthetically uniform, which I appreciate feels like something of a churlish complaint when one is using magical future technology to explore the thousand-year-old corridors of the immortal tomb of a child-god-king but, well, it is the future and we are entitled) without needing either to get your clothes dusty or worry about camelspit.
  • The Pear Ring: Everything old is new again! Now that everyone’s seemingly decided that dating apps are Not The One, people are once again casting about for new and exciting ways in which to monetise the fact that people, at heart, want to fcuk (yes, ok, fine, ‘find love’). So it is that we have The Pear Ring, a revolutionary product which, er, functions as a massive neon sign above your head reading “DTF!”. Obviously that’s not how the people peddling it would want to describe it – the ‘insight’ (lol) at the heart of this is that ‘76% of people are open to being chatted up in real life’ (LOL PERHAPS BUT NOT BY YOU), and that as such it makes perfect sense for all the single people in the world to all sign up to wearing the same article of jewellery so that they can identify each other in the otherwise-homogenous soup of coupled-up marrieds. WELL. There’s lots to enjoy about this idea – for a start, the thought that any woman in 2023 would want to go out in public wearing something that’s publicly-coded as ‘yes, please come and talk to me, stranger! I am open to your potentially-amorous advances!’ strikes me as…optimistic. Second, didn’t everyone try ‘traffic light’ parties in the mid-90s and decide they were a waste of time? Thirdly, as my friend Ged pointed out, this is going to get quite awkward for anyone who wears simple jade jewellery. My biggest objection to this, though, is how BORING it is – ‘wear a ring to show you’re single’ is basically the tedious, vanilla version of the old ‘wear a specifically-coloured hanky in your back pocket to show what you’re willing to do in bed’ from olden days San Francisco (and, if you replace the hankies with baguettes, Paris too, sexy history fans!), and it saddens me that the makers of this haven’t broadened the selection of rings available to encompass (for example), yellow rings for those who are open to impromptu toilet play. By the way, clicking on the website will present you with a banner suggesting that the rings are ‘93% sold out’ – the website was also saying this when I discovered it a week or so ago, suggesting either that sales have slowed somewhat or that the people behind it are lying. YOUR GUESS.
  • MoonView: This is Google Earth but for the moon. Yes, I know, it’s not quite as immediately visually-appealing as the Earth, but I for one had always been curious as to what Olympus Mons actually looked like and now I can die happy (lol) so, well, ENJOY!
  • Jasper The Doll: I understand literally nothing about this – who is behind it, and why the account’s owner, who as far as I can tell is a 20something kid in the states, is spenmding what I imagine amounts to a not-insignificant portion of their lives making TikTok videos embodying the persona of Jasper, a croaky-voiced and raggedy-looking doll which features in small skits and which really shouldn’t be compelling but which sort-of is. Look, I know that that’s a genuinely horrible non-description, but watch this particular example of the Jasper canon and tell me it’s not ART (it is, I promise you, ART).
  • To Be: I adore this. To Be is a digital poem by Alicia Guo – it’s infinite and self-generating, and I don’t quite know how it works or where it’s pulling the words from, but each time it’s different and each time it’s fragmented and magical and silly and poignant and confusing and beautiful and I would like this to be read forever by a choir of machine voices until the heat death of the universe please thankyou.
  • AutoGPT: I put it off as long as I could, I promise, but we’re quickly going to have to run through some AI stuff – I promise I will try and keep it as painless as possible, though, and we’ll be back to t he ephemera sooner than you can say “WE ARE ALL SO BORED OF AI THAT WE WOULD WELCOME DEATH AT THE HANDS OF OUR NEW MATHSY OVERLOARDS RATHER THAN READ ANOTHER FCUKING THING ABOUT GPT”. Anyway, after a frothy couple of weeks in which people have spent an awful lot of time talking about AI and What It Might Mean For Us without in any way spending time thinking about what, to my mind, is The Big Issue Here (to whit: “why are we building all these post-scarcity toys in an era in which scarcity is still very much a thing, and why are we not spending more time thinking about what this is going to do?”) here’s another piece of ammo for all the DOOMERS currently convinced that the LLMs are just a training cycle away from plotting our demise. AutoGPT is – let’s be clear – a very emergent thing right now, and it’s more theoretical than practical, but it’s…impressive in scope. The link here takes you to a Twitter thread talking about the principle and demonstrating how it functions in practice, but, simply put, “Auto-GPT is an open-source application showing the power of LLMs like GPT-4 to autonomously develop and manage different kinds of tasks like completing a code session or suggesting a business idea. You give an agent an identity, role/task, goals, and specifics about what to accomplish and it attempts to achieve it in an “autonomous” way through a framework designed to allow the model to “reason and act” to accomplish the task.”” So what this does is basically give a window into the not-too-distant future in which you will theoretically be able to ask your AI assistant to, for example, research your preferences based on a corpus of material and then, depending on its analysis of said corpus, go off and book you a holiday that you will like (checking your calendar for your availability, obvs). Is this good? It sounds convenient, of course, but then there are all the questions about the extent to which it’s a good idea to allow black boxes to behave with this degree of operational autonomy…BUT DON’T WORRY! As I said, this is (probably) MONTHS away from being reality! If you’d like a practical, hands-on way of exploring how this stuff might work, here’s a little toy that demonstrates the sorts of steps AutoGPT would go through to accomplish a broad task – oh, and of course someone has set this up with the goal to ‘Destroy Humanity’, so if you’re interested you can track the progress of ChaosGPT at this Twitter feed (at the time of writing it has decided to attempt to use Twitter to manipulate people to do its bidding, which will make you laugh darkly if you’ve ever tried to use Twitter to change anyone’s opinion about anything). Ought we be worried? NOT ABOUT THIS, BE WORRIED ABOUT THE JOBS AND THE MONEY FFS.
  • Memecam: Can an AI come up with meme gags? No, basically, is the answer, but that doesn’t mean it’s not fun to watch it try. Open up the Memecam site on your phone, snap a photo of anything you fancy and watch as The Machine attempts to come up with HILARIOUS ‘I Can Haz Cheezburger’-type copy to accompany it – these aren’t, by any standard definition, ‘funny’, but there’s something interesting about watching the software skirting around the edges of humour, and occasionally it will spit out something halfway-amusing (it correctly identified the fat, ugly cat currently staring at me through the back window as a stray, for example – yes, I have fed it this morning, and NO FATSO you are not having any more ffs). Why not spend the rest of your day wandering around your office/home/neighbourhood turning everyone you know into a meme? Yes, ok, fine, it might not sound like a fulfilling way to spend your Friday afternoon, but it will fill some of those empty hours between birth and death and that’s the main thing.
  • PoeBots: Poe is Quora’s attempt at an LLM – you all remembered this, right? You all hold this vital, crucial in your heads and hearts at all times? You don’t, do you? SO WHY MUST I BE COMPELLED TO REMEMBER THIS SH1T THEN? FFS! – and it’s now introduced quite a neat feature which elevates it a little more towards being a thing you might occasionally use rather than something you tried once and then forgot existed. Basically you can now use Poe to spin up themed chatbots with but a line or two pf prompt instruction; the gimmick here is that the bots will persist, and can be shared, and that it’s free. Which is potentially useful if you can be bothered to try and spin up a conversational assistant with a particular persona which you can persistently draw on – ok, yes, fine, you could do the same with a GPT prompt but you’d theoretically have to re-enter it every few tokens whereas this purports to last longer; it’s only playing with GPT3.5 and below-level tech, so don’t expect miracles, but this could be amusing/useful to play around with.
  • Segment Anything: On the one hand, I am aware that there is a lot of REALLY technically-impressive stuff going on with this tech, Meta’s new image-recognition AI which lets you isolate literally any element from an image with a pretty-terrifying degree of precision (seriously, play with the examples – the fact that it can clip the stick from a dog’s mouth is quite remarkable); on the other, I find it quite hard to get excited for stuff that is ‘Photoshop but easier’. Your mileage, though, may vary significantly.
  • PhotoPrompts: Ooh, this is a useful little tool – a prompt generator for images (as far as I can make out, specifically optimised for the latest version of Midjourney) which lets you select from a series of dropdowns to pick the name of the photographer whose style you’d like to rinse, the style of image you’re after, the aspect ratio and the lighting, along with the angle of shot and depth of field, and the site will create a usable prompt to start you off. Obviously this is…less good if you stop to think about the eventual endpoint for all of this, or indeed any of the thorny and still-largely-intractable questions around the extent to which a ‘style’ can be said to be ‘owned’ by an individual or otherwise, but if you don’t dwell too much and focus on the fact that it makes YOUR life marginally easier then, well, it’s all fine!
  • TextPrompts: I’m including this not because I think it’s good or I endorse it, but because I am once again slightly amazed by the grift here – would YOU pay $25 for a Chrome extension that would help you create prompts for LLMs based on a series of questions? Maybe you would, I don’t know, but this strikes me as a quite astonishingly-hopeful punt at making some cash out of the boom in interest in AI. Although, according to the site, nearly 300 people have signed up and this person’s trousered $7k+ so, well, I’m the idiot here, aren’t I? FFS Matt, you are always the idiot.
  • AI CafePress: Speaking of grift, I am genuinely agog at this platform which is LITERALLY cafepress for your terrible AI art – give it a prompt, wait a few seconds and then YOU TOO can create a generic storefront to sell your soulless, generic AI-generated imagery on poor-quality sweatshop-stitched tshirts and sweats! This is in-part funny, but also very much not funny at all when you consider the number of people currently working on selling exactly this vision to morons worldwide (there are 110% people currently selling “here’s how to create an AI-to-dropshipping pipeline for seamless passive income TODAY!” courses for $199.99 a pop), and take a moment to factor in the likely impact that this sort of stuff is probably going to have on the volumes of cheap, throwaway tat clogging up the planet in the years to come.
  • Infinite Adversaries: A fun project by Curios reader Felix Jung (THANKS FELIX!), who’s created a light ‘choose your own adventure’-style narrative game experience thing using GPT. Felix writes: “On entering the site, visitors are offered a choice of weapon (a different set of options every time). From there, ChatGPT then creates a random adversary and location, along with a set of four possible actions you can take. Some actions involve the weapon you chose. Some actions involve your physical surroundings. Depending on what you choose to do, ChatGPT then narrates the outcome. If you survive, your prize is another adversary. This was a lot of fun to work on, and I found myself pretty awed at the types of responses ChatGPT came up with (I provide some randomized guidance, but the options and outcomes are all from ChatGPT). Each encounter is also illustrated by an image from DALL-E.” This is fun-but-simple – it won’t keep you hooked for long, but it’s interesting to see what the software generates and it’s clear what the potential is from the point of view of emergent storytelling.
  • The Fantasy Internet Simulator: I LOVE THIS WHAT A WONDERFUL IDEA! Imagine browsing the internet from a half-imagined past, where the aesthetic and vibes sort-of fit but where everything’s not quite as you remember it…Nate Parrott is attempting to create exactly that with his Fantasy internet Simulator, which, in his words, works like this: “it works like any other browser: type a URL or a search query, and it loads a page! but in this case, the page isn’t coming from the internet. it’s coming from chatGPT. and i’ve asked chatGPT to pretend it’s still 1996.” You need to be able to install and run this yourself using Tesflight on iOS, but if you can then I promise you it is SO MUCH FUN and such a wonderful exercise in collaborative imagination and storytelling – wandering through a hallucinatory memory of a web that never was feels like the perfect application for this feverish generation of language manglers.

By Mark Steinmetz



  • QI Meditations:  Do you meditate? Does it…does it help? Does it stop the voices? Crucially, does it do a better job of at least keeping them a bit quiet than drugs and booze do? Maybe I should give it a go – and maybe I should use QI meditations to do so (I probably shouldn’t, though)! QI Meditations combines text-generating AI with text-to-voice synthesis to present you with a selection of guided meditations penned by machine; “Defying the norms of traditional mindfulness, Qi offers a series of randomly generated meditation exercises. These peculiar and amusing prompts urge users to embrace the absurdity of the content, promoting the idea that meditation can take many forms. As a commentary on society’s increasing dependency on technology and the prevalence of AI-generated content, Qi challenges users to scrutinize the boundaries of AI and its impact on human experiences. By presenting AI-generated art in such an eccentric manner, Qi compels users to face the unpredictable nature of AI and consider its potential consequences on art and culture.” Honestly, these are actually quite funny – there’s a featured one today in the style of Trump and whilst there’s nothing I find funny or interesting about That Fcuking Man anymore I confess to being charmed enough to listen to the whole, ridiculous thing.
  • Travel Photographer of the Year: BRAND NEW AWARD-WINNING PHOTOS! These are, as you would expect, superb – there’s a UK winner too, in the shape of the Young Travel Photographer of the Year Cal Cole from Manchester, but my personal ‘favourites’ (I use the term advisedly) are the images of the little kids boxing about halfway down because, well, just look at them.
  • CookNovels: I really like this idea, and it feels very much like something that half a dozen of you could reasonably steal for your ‘6/10-rated concepts that we can keep to lob in any presentation that’s looking a bit thin on the creative front’ folder. CookNovels riffs on the running gag about online recipe sites requiring you to read War & Peace before you get to the ingredients list – except here the joke is that you really DO get a whole (out of copyright) novel before each recipe. There are only three, but you can get your recipe for simple Chile Con Carne served with a sider order of Homer, say, or pair Moby Dick with a low-carb prawn cocktail – I refuse to believe that there’s not a million ways you can ‘borrow’ this and frankly improve it for the right brand but, equally, I am too lazy to bother to try and think of any so who the fcuk knows?
  • PrankGPT: This, though, I feel honour-bound to point out, is A BAD IDEA and Web Curios in no way endorses it. Do YOU think that it sounds like a sensible or liability-free idea to hook up an LLM, a text-to-voice engine and an web-to-phone calling protocol to spin up a system whereby you can call in an AI-generated ‘prank call’ about any subject you cabn imagine to anyone in the world whose phone number you know? IT IS NOT! It is, honestly, a very bad one! Yes, ok, fine, I know that the setup here is almost certainly made my a couple of bored highschool kids and is therefore is janky as fcuk (and in fairness I wasn’t quite able to get it to work this morning), and I am also aware that it’s probably not possible to make the bot generate anything awful because of guardrails and limited guiding inputs…but, also, I don’t think this FEELS like something that bodes well. Still, file this one away for June when you’re sitting around brainstorming Hallowe’en ideas.
  • EuroVelo: Whilst obviously I have no idea of the identities of anyone reading this, I’ve got a vague idea that the readership for Web Curios tends towards ‘middle-aged urban-dwelling men who probably work in a media-related industry’, which, from what I can tell, is a pretty much 1:1 Venn diagram match with ‘people who make cycling and the fact that they like it a core and frankly insufferable part of their identity’ (sorry, but I have lost friends to the two wheels and lycra disease and I am BITTER about it) – as such, I imagine that there are a few of you for whom this website, which lets you plan cycling trips around Europe (thankfully that’s geographical rather than political Europe, meaning even the great act of national self-harm that was Brexit hasn’t managed to spoil it for the Brits just yet), will be of not-insignificant appeal. Although on reflection ifyou like cycling that much then you probably new about it already. FFS.
  • April Fool’s On The Web: Yes, I know, there is very little in the world less funny than the lies that brands tell you on April 1st (apart, of course, from the ones they tell you on the other 364 days of the year! “Sustainability!” lol!), but at the same time you might find it helpful to have a searchable database of every single one of the (to repeat, DESPERATELY UNFUNNY) corporate gags committed to HTML each year since 1994, if only so you can use it to resurrect every single one of these with the addition of “…but with AI!”.
  • Colouring The Past: A photography series by Eri Erlick which serves not only as an colourised archive of great shots but as useful reminder that trans people have been a thing throughout recorded history, all over the world, and recent attempts to portray the concept of transgenderism or gender fluidity as some sort of uniquely-modern aberration born of post-web decadence is, frankly, stupid and wrong and ahistorical. “Using recent breakthroughs in photo editing techniques, Eli colorizes, restores, and digitizes photos from queer and trans history. The following images are originally from 1897-1973.After noticing how much more responsive audiences are to color photos, Eli decided to share these amazing moments from queer and trans history. During a time when politicians can openly argue trans people did not exist until 2015, it is important to use reminders like these that we have always been here.” It’s honestly astonishing and not a little miserable to note the degree to which the tenor of the conversation around gender has in many respects worsened rather than improved over the course of the past 50 years.
  • High Rises: Gorgeous photos of very tall American buildings. “The prosperity of early 20th century America resulted in a boom of skyscrapers that still tower over cities across the country today. Focusing on the character and craftsmanship on display at the top of these landmark buildings in a way that can’t be seen from street level, the Highrises Collection reveals fascinating details and stories of these distinctly American icons.A drone-mounted camera takes multiple high-res photos of the top of each Highrise. Images are stitched together manually to create a elevation scan with flattened perspective and enhanced lighting effects to accentuate depth and form.” The way in which these images are created lends them a slightly-odd flattened quality (I am fluctuating between describing them as ‘videogamey’ and ‘Wes Anderson-y’, but given I hate Wes Anderson let’s go with the former), but that’s part of the appeal here imho.
  • Reliable Robotics: To be clear, this links to the website of a robotics company – it may be a super-exciting robotics company, but it is very much a corporate website rather than a FUN DISTRACTION FROM MEATY HORROR. EXCEPT! The scroll animation features a genuinely-delightful graphical render of a plane, and features the near-perfect legend ‘scroll to activate take-off’, and I can’t tell you quite how much this apparently-unnecessary piece of graphical overengineering pleases me (and I hope it pleases you too). FLY, POINTLESS LITTLE CG PLANE!
  • Shed Of The Year: Many years ago, when this first launched and I was working for a company who did the PR for Amazon (yes, but those were simpler times when they were less evil, I promise) we launched an ill-feted attempt to get the ecommerce behemoth to sponsor this (look, they had just launched their gardenware vertical, it made sense at the time) – the fact that they basically laughed and rejected the concept out of hand is up there in the pantheon of my greatest career disappointments (other examples: not getting the job at The Sun when I was 20, the fact that I ever worked in PR in the first place), but I like to think that the company’s current reputational travails might have been somewhat eased had it been they rather than Cuprinol who badged the UK’s annual Shed of the Year contest. Anyway, click the link, marvel at the sheds and, if you’re a middle-aged man, feel quietly jealous of ‘The Fortress of Solitude’ (and all the rest tbf).
  • Soren’s Clever Design Gags: You’ve probably seen some bits of design sature doing the rounds online over the past few weeks – apps whose interfaces have been alterered slightly to make a point about how TERRIBLE they are, like LinkedIn adding an ‘opentowork’ button to acknowledge the fact that the only reason anyone engages with that horrible, soulless colosseum of misery is to climb another rung up the slick, viscera-greased pole of corporate success. Almost all of them will have been by this person, Soren (no surname supplied), who collects examples of his work on this website (you can sign up to get a new gag via email each day should you be so inclined).
  • The Shapes of Stories: Ok, now this is an interesting use of AI. One of the most-helpful ways of thinking about LLMs (and indeed the current crop of ‘AI’ tools) is less in terms of ‘this thing generates words’ or ‘this thing generates images’ and more in terms of ‘these tools let us develop links between different disparate things in ways that have never been attempted before’ – which is exactly what this experiment is doing. Taking a bunch of classic stories, the person behind the project (Superb Owl, whoever they may be) asked GPT to ascribe a numerical value to the emotional state of each character at various points in the story, which can then be graphed to create a visualisation of the emotional journey experienced by protagonists throughout the novel in question – which is both interesting from the point of view of narrative construction and the Vonneguttian idea of ‘story shapes’, and also from the perspective of ‘how The Machine can effectively do some really interesting quant/qual-mixing stuff’. The idea of taking this sort of approach and then adding an instruction to ‘and now translate this graphical representation of a story arc into a symphonic composition for bassoon’ really rather appeals to me – I am basically waiting for someone to create an entire, machine-imagined operetta based on the emotional beats of the entirety of Crossroads, so if any of you fancy knocking that out that would be great, thanks.
  • Tape Cassette Inserts: Who wouldn’t want a Flickr page which exists solely to catalogue all the different graphic design styles that were applied to the bits of paper and cardboard that you used to inexplicably get inside blank cassette tapes when you bought them? NO FCUKER, etc! Leaving aside anything else, I reckon you could probably make a reasonable line in passive income (sorry! sorry!) from setting up a company selling 80s-esque bedsheets in these sorts of colourways to the recently-divorced.
  • MouseTok: This is the TikTok account of one Martin Critchlow, whose feed consists exclusively of videos of his collection of fieldmice just sort of hanging about. In particular, Martin seems to focus on one particular mouse he has named ‘Mr Jingles’ – tragically, Mr Jingles recently lost his partner (“Mrs Jingles”, should you be curious), but both Martin and the mouse appear to be managing ok in her absence (although I challenge you to watch the tribute to Mrs Jingles on the feed and not tear up a bit). I am going to assume that this is entirely pure and sincere, and as such I wish nothing but the best to Martin, Mr Jingles and the rest of his nameless troupe of rodents.
  • Frog App: OH GOD NO MORE SOCIAL APPS EVER PLEASE GOD. Still, if YOU are in the market for a BRAND NEW video-first social app, but this time one that’s focused on FRIENDS and CREATIVITY (now where have you heard that before? Oh, yes, that’s right, everyfcukingwhere!) then you might be in the market to check out Frog, made by former LSE students and which aimed SO far outside my demographic that I can barely make sense of the website. You post ‘Sups’ apparently (yes, that sound you can hear is my jaw clicking in barely-concealed irritation), and there’s a nice gimmick in that anyone can continue any of their friends’ stories by creating ‘sequels’ which effectively follow on from the original content and which feels like a nice tool for all sorts of creative endeavours but, honestly, everyone featured on this website looks so young and fresh and full of life that it mostly made me want to just lie down and bleed into the floor until it all fades to white and mercifully just stops.
  • Permanent Beta: DON’T THINK JUST CLICK! Oh, ok, maybe think a *bit*, but generally I found this website was at its best when just approached with a sense of curious and open-minded wonder. If you have any professional interest in / intersection with the arts and digital then I think you will very much like this – or, at the very least, find it a though-0provoking exercise in digital practice/praxis (sorry). “[permanent beta] is an online platform that visualises the ongoing research of Fotomuseum Winterthur in the field of algorithmic and networked image cultures. As a platform meant to remain in a state of perpetual change, it tracks the research process, turning it into a creative and performative act in which knowledge around a specific thematic focus is collectively collected, (re-)arranged, edited and streamlined over time.” This will, I guarantee, only make marginally more sense when you click through, but I am fairly-certain you will like it.
  • PrisonGames: This is so interesting – Marcos Paz runs programmes in Argentine jails helping to teach inmates useful skills to help them gain employment in the games-and-games-adjacent industries after release, but also as a means of helping them express themselves through creative storytelling and ludic mechanics. This site collects some of the work that the people he’s taught have produced, seemingly all pieces of interactive fiction using Twine or similar – it’s in Spanish, as you’d expect, so you’ll need a bit of language, but these are poignant and, honestly, in parts heartbreaking short works of semi-autobiography.
  • Word Solitaire: This is a good game, but it is SO HARD that it has made me feel really quite thick and I am, to be honest, mainly sharing it with you all so that you can all feel stupid too.
  • Legend of the Red Dragon: The final miscellaneous link of the week is a proper throwback – remember BBSs? Oh, fcuk, you may not – erm, bulletin boards! Does that help? Erm, no, ok…just think of them as old, slow forums, basically, or, alternatively, just go away and Google all this and then come back when you’ve done your homework. Done? GREAT! Anyway, this is a wonderful and painstaking recreation of oldschool multiplayer RPG Red Dragon, whose Wikipedia entry describes it as follows: “The premise of LORD is that a red dragon is wreaking havoc in a town where the player has recently arrived. Multiple players compete over a period of weeks to advance their skills and to kill the dragon. In order to achieve this goal, players must face combat to gain experience. Once they have gained enough experience, they must face their master at Turgon’s Warrior Training and advance in skill level. Advancement increases the players fighting stats and gives an additional skill point in the current skill (up to 40). Advancement also presents stronger enemies and masters; a player must challenge and defeat master Turgon himself to reach level 12, the final level, before attempting to search for and slay the dragon.” Enter text commands! Meet other players! ENTER THE METAVERSE AS IT WAS IN 1984!!!!! Lol that is funny satire because aside from the bells and whistles that is basically what this is. A time capsule, a bit of history and a weird illustration of the whole general ‘plus ca change, plus reste la meme chose’ vibe of modernity right now.

By Claire Morgan



  • Fanzine Haemmorhage: Ok, fine, not *technically* a Tumblr but it’s a fabulously-named website and it feels like a Tumblr and, as per, MY SITE MY TAXONOMY FFS. Anyway, if you want a place where you can find a truly marvellous selection of old scans of fanzines alongside writings about the bands and scenes they describe, this will be PERFECT for you: “Fanzine Hemorrhage explores sub-underground music fanzines from the 1960s to the present, and attempts to explore the worlds they defined and in some cases created.” This is WONDERFUL (and reminded me that a few weeks ago I did a guided walk by Paul Talling (who runs this website and whose stuff I recommend unreservedly) which basically descended into him sharing increasingly snakebite-tinged reminiscences about “that time I put on Big Tomato Dave at Filthy McNasty’s” (that is basically a verbatim quote) and which is pretty much the same vibe as this site, all told).
  • False Knees: An ACTUAL TUMBLR! Also, I have for years seen images from here across the web and never known the source – if you’ve ever seen cartoons featuring beautifully-drawn and hyper-literate birds chatting about the travails of modern life then you may well recognise the work here.


  • Siegfried Und Joy: Yes, ok, fine, they ARE magicians, but they are FUN magicians! The subway exit magic will honestly NEVER get old to me.
  • Rob Strati: Smashed plates and surprisingly-good artworks.
  • Rupublicans: Republican grandees, reimagined by Midjourney as drag queen versions of themselves for SATIRICAL PURPOSES. On the one hand, a pretty lazy, one-note gag – on the other, some of these images are GREAT.
  • Project JDM: Combining maths, animation and music, this is so so so smart and so well made – I am slightly in awe. Thanks to whoever it was who sent this to me this week, whose identity I have totally failed to recall.
  • The Tube Map: Do you love the London Underground map? Would you like more of it on your Insta? GOOD WELL HERE IT IS THEN. Beautifully, this doesn’t appear to be affiliated with TFL in any way, which means that this is just someone who really loves the tube map a LOT. Thankyou, anonymous tube map fan!


  • Ian Hogarth on AI: Per an earlier comment, I KNOW YOU ARE ALL BORED OF AI I KNOW YOU ARE. Still, for the few remaining of you who can stomach another few thousand words on WHAT THIS ALL MEANS FOR US and WHERE IT IS GOING and ALL THE REASONS TO BE SCARED, this is a very good read by Ian Hogarth in the FT, which does a decent job of straddling the divide between ‘WE NEED TO BE FRIGHTENED OF THE ALL-KNOWING AUTONOMOUS MACHINE OVERLORDS WE MAY ONE DAY BIRTH!” and “WE NEED TO BE FRIGHTENED OF ALL THE WAYS IN WHICH THE HALF-FORMED VERSIONS OF AI WE HAVE ALREADY BIRTHED ARE MESSING WITH SYSTEMS WE ARE ALREADY ONLY BARELY-CAPABLE OF UNDERSTANDING DUE TO THEIR MASSIVE COMPLEXITY!” – the upshot here is that Hogarth is…a bit frit, but in a reasonably-coherent and non-hysterical way, and after reading this maybe you will be too.
  • The Data Delusion: I thought this piece in the New Yorker was an excellent companion to the latest ‘wow, a lot of you really enjoyed this, didn’t you?’ article to do the rounds of the strategyandplanningsphere – the ‘Age of Average’ piece by Alex Murrell was shared by seemingly everyone in advermarketingpr circles with ‘strategy’ in their jobspec as a way of proving…well, whatever they wanted to prove, from what I can tell, but mainly that we need to be BOLDER and RISKIER because focusing on the middle of the bellcurve leads to work that’s homogenous and dull and uninspiring… Anyway, the New Yorker piece to which this is ACTUALLY a link to looks at a potential problem of the emergence of predictive AI models based on large corpuses of data – to whit, the likelihood that they will result in a clumping of output towards averages and predicted centres, exactly the sort of thing that Murrell was railing against. You know what the problem is? The problem is fcuking DATA and our fetishisation of it – but you’re not allowed to say that, seemingly, despite the fact that (and I firmly believe this) that (for example) making decisions based on a bunch of data scraped from Twitter is nearly-always going to produce work that is worse and less interesting and more stupid than just sitting and thinking for a bit.
  • GPT Is Getting Worse: Or, more accurately, OpenAI is getting better at locking it behind bars. This piece by Frank Lantz explores how in his experiments with the OpenAI stable of LLMs the software has become progressively less ‘imaginative’ with each update, and more fixed into a slightly-bland positivity in its responses. I can only echo this – Shardcore has kindly spent a bit of time seeing about introducing GPT to the Curios archive to see what happens, and it’s interesting to see that it is near-impossible to break the base model from its emptily-bromidic ‘Eddie from the Heart of Gold’ tone of voice and writing style, however much of my appallingly-convoluted prose you force down the machine’s digital gullet. The upshot, basically, is that I’m sadly stuck writing this for the foreseeable as I don’t seem to be able to outsource it to The Machine yet.
  • Making One Of Those AI Balenciaga Vids: Vox gives you a step-by-step guide on how to create you very own variant on the recent spate of ‘what if [INSERT FILM FRANCHISE X] directed by [INSERT VISUALLY STRIKING DIRECTOR Y] as a Balenciaga ad?’ vids – it’s both easier and slightly-harder than you might think, and the results look significantly less polished than the hugely-viral one’s you’ll have already seen on Insta/Twitter/TikTok…but they don’t look that bad, and given that teenagers have for the past two decades proved beyond any shadow of a doubt that they will consume any old crap under the guise of fanfic I think we can probably all agree that someone’s going to create a movie-length version of one of these by the end of the year (my money’s on a Potter-inspired story with a pro-trans twist, just for maximum culture war relevance) and it will comfortably do 10million+ views.
  • The AI Village: In many respects this is the most remarkable story of the week, I think. The link here takes you to an actual academic paper, but I promise you that it is properly readable and interesting, and if you have any interest in games or theatre or TV or general storytelling then you owe it to yourself to at least give this a go. The precis is basically that a bunch of researchers created a selection of ‘autonomous’ AI agents existing with a videogamelike closed sandbox (think a Stardew Valley-type gameworld), and then just…watched them. Each of the ‘agents’ in the simulation has its own wants and impeti, and these motivate them to act in specific ways, and the intersections of these interactions create emergent narratives…”Generative agents draw a wide variety of inferences about themselves, other agents, and their environment; they create daily plans that reflect their characteristics and experiences, act out those plans, react, and re-plan when appropriate; they respond when the end user changes their environment or commands them in natural language. For instance, generative agents turn off the stove when they see that their breakfast is burning, wait outside the bathroom if it is occupied, and stop to chat when they meet another agent they want to talk to. A society full of generative agents is marked by emergent social dynamics where new relationships are formed, information diffuses, and coordination arises across agents.” Anyone who’s been reading this for a while (and, er, who’s also been paying any attention to what I say, which I appreciate is probably not many of you) knows that I have long been obsessed with the idea of creating a modern version of Little Computer People using this sort of kit – this feels like an incredible well of potential just waiting to be tapped, and if you’ve spent any time marvelling at the sort of mad emergent storytelling gems that games like Dwarf Fortress or Skyrim throw up then this will tickle your interest significantly (oh, and if you’re curious, you can watch the how the experiment played out via this playback system).
  • Substack Hasn’t Thought About Content Moderation: Do you remember a couple of years ago when Substack was getting big and splashing the cash, and people started asking questions about the company’s policy on what it would and wouldn’t allow people to write about, and the company basically revealed itself to have not really given any proper consideration to issues of content moderation and responsibility? Do you imagine that in the intervening time between then and their announcement of the Twitter-like ‘Notes’ functionality on the platform this week someone senior at the company might have spent some time thinking about what they would do if they were in charge of a realtime short messaging network? Do you think they might also have briefed the CEO on lines to take on this thorny-but-incredibly-predictable line of questioning in their first media interview post the Notes announcement? OF COURSE NOT! This is a transcript of the interview between The Verge and Substack CEO Chris Best, and it really is worth reading the whole thing – if you’ve ever worked in comms, there will come a point about ⅓ of the way through where you start to see where this is going, and then a point about ⅔ of the way down the page where you will, if you’re anything like me, be literally be reading through your fingers while feeling genuinely sorry for the person whose job it is going to be managing the fallout of all this. Note to all CEOs – if someone asks you straight out whether it is ok for someone to use the thing you have made to be massively racist, THE RIGHT ANSWER IS ‘NO’.
  • Posters: I tend not to include much stuff on THE CRAFT OF ADVERMARKETINGPR in here, mainly because it bores me to tears and I tend to find that 99% of it is pointless, overwritten w4nk designed to add an air of profundity to a series of disciplines which really don’t warrant any such thing, but I will make an exception for this piece which extols the virtues of good copy on posters and, more importantly, gives you not only some excellent examples of said GOOD COPY but also some gentle, non-patronising explanations as to why said copy is good. This is a really good post by someone called Dave Dye – thanks, Dave Dye.
  • That’s My Tank On Fire: A typica;ly-superb piece by James Meek in the LRB, detailing his experience of watching the social media war play out on Telegram and elsewhere, and what it feels like when your experience and knowledge of a conflict is experienced entirely through visual memetic propaganda. So so so interesting, and will be particularly so to anyone who remembers the first Gulf War welll enough to also remember the Baudrillardian discussion around its status as ‘hyperreality’ and the role of the media in the constructed reality of the conflict. It’s also a great piece because it gives an impression of the increasing impossibility of achieving anything more than an intensely-impressionistic and altered view of events like this – any events? – from hereon in, which is only going to become more pronounced with every day that passes.
  • Interview With The Mercenary: This is quite a remarkable interview with quite a remarkable man (I use the word ‘remarkable’ in its literal, value-neautral sense here) – Simon Mann is…what is he? A relic of an England that doesn’t exist any more? A ‘soldier of fortune’? An old colonialist? Whatever you might call him, don’t call him a ‘mercenary’ – he doesn’t seem to like the term, and it struck me throughout this piece that Simon Mann is not someone you probably want to upset very much. If you’ve ever wondered what the life of an Eton-and-Sandhurst-educated former SAS officer who has attempted a coup in Equatorial Guinea, who worked with Mark Thatcher, who’s spent time in some…tasty-sounding jails and who appears to be entirely at peace with all this then, well, this will be right up your street. Equally, if you want to be rendered speechless with rage about the blithe attitude towards the lives and countries of others evidenced by this modern-day Haggard (this is how I imagine Mann would like to see himself, at least) then, well, this will also do the trick! Something for everyone!
  • Cruise Ship Invasion: I can’t help noticing that the latest ORGANISED FUN CONCEPT that Big Business is currently trying to flog to millennials and GenZs is cruises – you too will have seen the ads for Virgin’s nascent cruise ship business, promising SEXY YOUNG PEOPLE and NO CHILDREN – which made this piece, looking at the environmental impact of the cruise ship industry which tours Alaska from Seattle every year, feel partciularly timely. The short precis of this is ‘lol, you thought that this was OK for the planet? You rube! You absolute groundling! You know nothing!’, but the details are quite startling and the scrollytelling (sorry, but until someone comes up with a better word then that’s what we’re using) is pleasingly-done.
  • Being Colourblind Online: Friend of Curios Andy Baio (HI ANDY!) writes for The Verge about what it’s like for him using the web while colourblind – this is not only a really interesting piece, with some helpful graphics that show you exactly what a crappy experience navigating the vast majority of the web is for people with even minor visual impairments, but a useful reminder that accessible design is important and more necessary than we often imagine. It’s part of a wider series of articles all about the business of UI and the impact it has (and has had) on the way we use the web – “we explore the small design decisions that have had an outsize impact on our lives. From simple card game browser UIs to deliberately complicated video game setups, all-too-forgotten accessibility options for colorblindness to the curious incentive-driven histories of the shuffle and log out buttons, these stories delve into the ways that user interfaces have driven us forward, or failed, or found an entirely new way of living” – which is full of interesting pieces, whether or not you’re involved in the business of design (although, honestly, I think everyone is basically a BIT of a UI person these days, or ought to be).
  • Don’t Monetise Your Kids: Or, “hang on – given the fact that we’ve all decided, rightly, that sending kids to work in factories is not ok, why are we seemingly totally fine with their parents exploiting their image and occasionally their labour for the big big influencer bucks?” To which the answer is ‘we’re not are we?’, or at least I’d sort of hoped it was – still, if you’ve ever wanted to read an exhaustive piece about why it’s perhaps no ok to force your three year old to do sixteen takes of her eating breakfast until it looks appropriately ‘cute’ then FILL YOUR BOOTS!
  • Nonsensical Advertising Futures: This is genuinely a bit miserable and dystopian and chilling, but, also, depressingly smart. Commuters in New York City have reported seeing confusingly-nonsensical video ads on their subway digital billboards, featuring the sort of content you’d expect more from chef/magicians on Facebook making some sort of artery-destroying ‘here’s a block of cheese wrapped in 19lbs of ground steak!’ monstrosity. Curious as to why, this article investigates and discovers that it’s a basic attention hack to grab passers-by’s attention for a fraction of a second – turns out that this sort of ‘WTF’ish video really does make you look, and therefore makes people marginally more receptive to whatever REAL ad comes on next. This is 100% going to become a thing, isn’t it?
  • Clitoral Enlargement Surgery: Because I’ve featured a few posts here about weird trends in masculine body enhancement over the past few months and felt the need to even things up a bit. Is this a trend? I…I doubt it. Still, if you’ve ever wanted to read a deep-dive into the (largely dubious-sounding, if I’m honest) benefits of having one’s clitoris enlarged then this will, er, hit the spot – even if you’re not interested in the full article, can I please ask that you click through and scroll down til you find the photos of the person doing this surgery who looks EXACTLY as you would imagine a West Coast plastic surgeon specialising in clitoral enhancement treatments would look.
  • Real Life Dates With Imaginary Boys: I generally try and stay away from stories that are basically ‘lol Japan you so funny!’, not least because it feels a bit lazy and twodimensional (Web Curios? Lazy? Two-dimensional? HOW DARE YOU, etc), but on this occasion I will make an exception because, well, SO MUCH OF THIS IS MAD. I have no idea if this is an actual ‘trend’ or just something that a couple of people have done but which is going to be dressed up as a ‘thing’ regardles of its niche status, but let’s hope it’s the latter. The article describes how young Japanese are paying cosplayers to go on ‘dates’ with them in character as their favourite heroes from ganes or anime – so, rather than going to the trouble of interacting with a real person you can instead have an afternoon with someone who you are paying to assume the persona of a fictitious individual from a game or cartoon. Is…is that good? It doesn’t *feel* good. Still, the whole thing is agreeably mad-sounding, rendered all the moreso by the photos that pepper the piece in which everyone looks like a terrifying anime character because that is what phones make LITERALLY EVERYONE look like in Japan as far as I can tell.
  • Meet Dril: His first out-of-character interview and my overriding impression was…sadness, honesty. I can very much relate to the idea of being stuck inside a persona/system/process that is simultaneously a comforting snug and a terrifying prison, and of the horror of ‘being a webmong at 50+’ (for ‘webmong’ substitute ‘poster’, or other content-appropriate epithet of choice), but I think the thing that touched me most was the description of him attempting to make his ‘thing’ the right shape to fit other media and realising that actually it probably doesn’t really work anywhere else other than Twitter – there’s a proper, real outsider artist vibe to this which I find poignant in the extreme (oh, and it’s also worth taking note of the Patreon figure he’s drawing down – tHe CrEaToR eCoNoMy lol!).
  • Long Sentences Rule: This is, I think, the second essay in the space of a year to plough this particular furrow, and it;s proof positive (to my mind, at least) that the tyranny of Carver is FINALLY COMING TO AN END (I like Raymond Carver a lot, don’t get me wrong – or at least the version of him created by his editor – but I also find the idea that a sentence must not be longer than 10 words or so on punishment of death to be, well, fcuking stupid), and it’s a wonderful celebration of prose that is sinuous, serpentine, rich and lustrous and which is unashamed of its clauses and subclauses and parenthetical diversions. It should come as no surprise to anyone who’s read Web Curios that I am a firm believer in the long sentence – I am, however, self aware enough to admit that that is because I am a lazy and flabby writer of prose rather than because I have a unique and masterful style that deserves room to breathe (SO PLEASE DON’T FEEL THE NEED TO TELL ME).
  • Blurred Lines: Because I am not a music radio person, and because I don’t drive and so don’t often get exposed to commercial stations, I was perhaps the last person in the UK to hear Blurred Lines when it was big 10 years ago – I can’t have been the only person, though, who on first hearing it thought ‘man, this is going to get old REALLY QUICKLY’ and also ‘wow, this really is…very, very r4pey, isn’t it?’ (an aside – am I the only person who thought the lyrics to ‘Get Lucky’ were basically all about trying to stay awake long enough that the girl you are with gets drunk or tired enough to just give in and sleep with you?, and that Daft Punk rather got away with one there? No? Ok, fair enough). This is a great Pitchfork piece looking back at the song, the Summer that it characterised, the Robin Thicke ‘phenomenon’, and that odd period when Miley Cyrus was everywhere and American pop culture got really weird about race (again).
  • Suicidal Moments: NB – THIS PIECE IS ALL ABOUT WANTING TO KILL YOURSELF AND OCCASIONALLY TRYING BUT NOT QUITE MANAGING IT. So, er, caveat lector and all that jazz. Still, if that doesn’t put you off then I think this is a very good, very clear-eyed piece of writing about what it feels like when you want to kill yourself and how to maybe deal with said feeling when it occurs.
  • Out To Lunch: An immensely-enjoyable piece of writing, again from the LRB, by Paul Theroux, which simultaneously tells the general story of The Literary Lunch in swinging/seedy (delete per your perception) London of the 1970s and the very specific story of how he tried and failed to have an affair with an artist called Molly. Superbly-observed, unsentimental and perfectly-written, this is all the better for being pleasingly-honest about the author’s intentions and what a sh1t he was being – and, of course, for being about lunch and the eating thereof.
  • John Wick is So Tired: Finally this week, a poem about John Wick – I have never seen any of the films, but I don’t really think that matters. This is called John Wick Is So Tired and, honestly, I think this is beautiful and near-perfect, and I think you might too.

By Brecht Van Den Broucke