Hello everyone! HELLO!
One on the one hand it was fun this week watching Google’s PR fcukup with the whole ‘your AI search engine’s a liar LOL’ story; on the other, I can’t help but think that none of the serious coverage of this I have read in the media has spent quite enough time thinking through the potential consequences of what AI-led search might mean and that, given our previous experiences over the past 15 years with ‘letting technologies we don’t fully understand and haven’t fully evaluated the potential impact of loose on our society without doing anything even halfway-resembling proper, meaningful user testing because that’s what investors want and caution be fcuked!’, we might want to perhaps employ a bit more caution rather than ploughing straight ahead with the whole ‘let the black box tell me what it thinks is true, despite the fact that neither I nor indeed anyone else currently alive has any idea by which mechanisms it arrives at those evaluations’ thing.
Christ, sorry, we’re only at the beginning of this fcuking thing and I’ve already done one of those horrible nested paragraphs that seems to go on forever like some sort of hideous mobius strip of badly-crafted prose.
ANYWAY. I hope you’re all well and having fun, and that you enjoy this week’s Curios; I am still Matt, this is still Web Curios, and you should probably start looking up old editions of the Encyclopaedia Brittanica to use as some sort of gold standard of truth in the coming veracitypocalypse.
WE BEGIN THIS WEEK’S EDITION WITH THIS PLEASINGLY-ATMOSPHERIC TECHNO MIX BY JOACHIM SPIETH!
THE SECTION WHICH OBVIOUSLY CAN’T TELL YOU WHAT TO DO BUT WHICH IS JUST GOING TO LEAVE THIS LINK TO DONATE TO RELIEF EFFORTS IN SYRIA/TURKEY HERE, PT.1:
- Here Comes The Video AI: This isn’t quite ready to take over your feed in quite the same way as video and text AI have done, but it feels like it’s only a matter of time as the examples here really do look quite magic. The link here takes you to Runway’s announcement of their ‘swap new elements or apply new styles to an existing video’ tech (which you can apply for access to should you be so minded) – this effectively means that you can give the software any moving image file you fancy and (for example) tell it to ‘swap the ugly people for beautiful ones’ and VWALLAH! Ok, fine, it’s not quite that simple, and it obviously looks a *bit* potato-ey at the moment, but, equally, it doesn’t look that potato-ey, and this is another of those increasingly-common moments where I feel like looking over at everyone standing in large groups outside Covent Garden pubs as they take a break from the hard work of post-production editing and maybe suggesting they enjoy the good times while they can. It’s not only Runway doing this – there’s another paper here about similar-but-different software which does broadly similar stuff, and you can try a different version of this tech at this HuggingFace link should you be so minded. Having tried it out, it seems to work best for light style transfer-type modifications, so why not spend some productive hours this afternoon asking it to reimagine all of your favourite sporting moments as though drawn by Hieronymous Bosch or something?
- The Prompt Engineering Handbook: Despite having previously said that I don’t think ‘prompt engineering’ is going to be a thing for very long, I am conscious of the fact that right now it can be quite useful to know how to better-wrangle The Machines. So, er, here! This is a link to a whole bunch of resources on both the theory of ‘talking to the AIs’ and also the practice of how best to construct your increasingly-byzantine instructions so as to produce the exact shape and style of sexy fursona you’ve been imagining (that is what you’re going to use it for, right?). BONUS PROMPT ASSISTANCE: This is a rather useful prompt-constructor for Midjourney, which offers you a variety of prompts and sliders to help you communicate with the machine and which is particularly useful for all the subtle ‘emphasise this, deemphasise that’-type tweaking which marks the best AI image generation (and this link is also part of the reason why I don’t think ‘prompt engineering’ is going to last long as a thing). ADDITIONAL BONUS PROMPT THING!: Should you be particularly interested in this sort of stuff, and should you be practically involved in it (or, probably, should you want to be practically involved in it), and you can get to London on Monday evening, you might be interested in attending the London Prompt Jam on 13th Feb at Newspeak House (although, to be clear, Web Curios accepts no responsibility for the amount of creepy AI-generated cheesecake manga imagery that you’ll be subjected to should you attend).
- AI Accompaniments: Muchlike the somewhat-magical music composing AI previewed by Google the other week, this is another link to a work-in-progress piece of research which features some quite impressive results; once again it’s Google, once again this links to a rather academic blogpost which, if you’re stupid like me, will mostly mean the square root of fcuk all to you but which does feature a range of different examples of the tech in action so that you can be suitably astonished by the speed at which this stuff is evolving at. The working name for this kit is SingSong, and the paper explains that it’s “a system which generates instrumental music to accompany input vocals, potentially offering musicians and non-musicians alike an intuitive new way to create music featuring their own voice” – so imagine a not-too-distant future in which you can croon a few bars into your phone as inspiration strikes, and The Machines (I don’t know why but I’m feeling an increasing desire to refer to themselves with the capitalisation and definite article – an affectation which, I promise, I will do by best to resist) turn it into a TikTok-ready clip complete with backing track. Again, this isn’t currently very good – you wouldn’t, I don’t think, actively choose to listen to any of this unless you were possibly on a lot of very experimental drugs or on the brink of some sort of nihilistic episode – but it’s more about the potential and what this will almost-inevitably enable.
- Stable Attribution: This made me feel stupid earlier this week, and I am slightly grumpy about it – but, equally, it provided me with an excellent practical demonstration of why you should greet anything AI-related with a reasonable degree of raised-eyebrow scepticism (lol Google Bard!). Stable Attribution purports to be kit which will identify the source imagery used by Stable Diffusion to generate its own compositions – so effectively lifting the lid on the materials that have been ‘used as creative inspiration’/plagiarised (delete as applicable). Except, as pointed out to me by various people after I shared it, it doesn’t ACTUALLY do that – what it appears to do is run a relatively simple reverse image search on the image and pull out stuff that shares rough visual DNA with the endpoint image (which, based on the latest understanding of this sort of kit, may well indeed be possible), meaning it would also pull out ‘source imagery’ for stuff which was very much created by an actual human. HOWEVER, despite literally not doing anything that it says it does, Stable Attribution is actually weirdly useful as a means of finding other semi-related visual styles to a specific image, so should that be something you feel might be useful to you then YOU ARE WELCOME.
- GlossAI: Look, I PROMISE you that all this AI stuff is going to die down a bit soon. Promise. Til it does, though, I can’t help but be fascinated by the various layers that are being developed to make use of it; I’m reasonably confident that most of the services and products popping up in the current frothy ‘OMG AI!’ explosion won’t exist in ~12m time, but I think it’s interesting and instructive to keep an eye on what people are playing with and the ways in which this stuff could potentially be employed. Here, for example, is a service which I am reasonably-confident won’t actually produce anything particular stellar but which offers a vision of a future in which creating multiple, platform-appropriate assets from a single source really is as easy as clicking a button. The advertised gimmick here is, basically, ‘give us a video and we will turn it into about 300 different content types for all the different social channels, as well as probably some written content too and maybe a presentation because why not?!’, and you and I both know that the outputs will be shonky as hell and largely nonsensical, but, well, that won’t always be the case, and I would advise the low-end video editors out there to maybe join the post-production people outside the pub because, well, you might as well, eh?
- The Braggoscope: Absolutely one of my favourite projects of the week, this, partly because of the fact that it involves BBC Radio 4 and therefore is very much up my middle-aged street, but also because it’s such a wonderful example of how you can use all this AI stuff to do interesting, useful and creative things RIGHT NOW. Matt Webb has used GPT-3 in conjunction with a few other development tools to analyse the transcripts of every single episode of the long-running historico-cultural radio show In Our Time and arrange them taxonomically by theme on this site. So, for example, you can now explore all the episodes of the show which deal with economics, say, or literature and rhetoric – which doesn’t sound like much, but the person hours required to do this would have been VAST and it’s such a nice, smart way of demonstrating how astonishingly useful for brute force work this sort of stuff can be. The obvious extension to this, with a bit of additional hackery, would presumably to then offer per-topic summaries based on the programmes’ contents (although I get the impression that the whole thing would probably start falling apart a bit at that point, quality-wise), but hopefully you can get a vaguely-exciting flavour of what the possibilities are (and maybe about how you might use this sort of thing yourself – I keep having (admittedly quite boring, fine) thoughts about what you might be able to do with GPT if you create a bespoke version and feed it with all of a company’s information, for example (like this sort of thing, or this sort of thing, but, frankly, more terrifyingly comprehensive – from the website to EVERY document, to file structures, to (perhaps more controversially) employee emails and Teams conversations and effectively treat it as a living knowledge base that can be interrogated using natural language, and that sounds…sort-of cool, in a very tedious, corporate-y sort of way, but also really fun in terms of how you might use this. Anyone else? Eh? Oh).
- Poe: It feels rather like we’re (rapidly) coming to the fag-end of the period in which OpenAI is going to let us have infinite playtime with its AI toys for free – everything that’s come out this week about the Bing integration and Microsoft’s plans for monetising the tech at enterprise level makes it feel a bit like you’re going to start having to cough up for anything but the very bare minimum quite soon – not least because ChatGPT’s been rendered basically unusable by the fact that seemingly every awful right-wing fcuker in the world is constantly trying to get it to agree to using hatespeech. Anyway, Other Large Language Models Are Available! And this is Poem which has been developed by Quora and which is now unrestrictedly available (to iOS users only, chiz chiz) and which basically does the same tricks as ChatGPT but in an app and on your phone and, for the moment at least, free. Given it’s aforementioned Apple-only status at present I’ve not tried it out myself, but the feedback I’ve seen so far has been broadly positive – it did make me think, though, of the weird potential future in which there are a selection of these sorts of tools, all trained slightly differently on different corpuses of information and with different tweaks and weightings given to their balances, and what it might be like should we all have on-device AI-assistants whose knowledgebases are all subtly different and whose perceptions of ‘truth’ and ‘right and wrong’ vary from instance to instance. Interesting, no? Oh suit yourselves ffs.
- Claralell: Type in your name and listen in rapt joyous wonder as the machine spits out a tune based on some maths involving spelling (there is a proper explanation of the rules behind how it works on the ‘About’ page, for any of you who are less tin-eared than I am and who understand things like ‘staves’ and musical notation and ‘not being tone deaf’) – honestly, this is SO CHARMING and I was pleased to find out that ‘Matt Muir’ is a delightful bit of light piano; see what YOU sound like! This made me halfway wish that music generating AIs were slightly further along, as I would love to feed the Matt Muir snippet to another machine to create a longer, more involved composition (I have seemingly stumbled upon a subconscious desire to have a symphony written for me).
- Howl vs Loewe: Are YOU a fan of the novel Howl’s Moving Castle by Diana Wynne Jones? Or, perhaps more probably, the lavish Studio Ghibli animated film inspired by said novel? GREAT! You might like this charming little (mobile-only) site which lets you answer a few questions and thereby determine which of the book’s characters best represents you – and THEN ,because this is a collaboration between Studio Ghibli and high-end fashion house Loewe, you get shunted into a storefront and asked whether you fancy shelling out several grand on a handbag that looks like Calcifer. I am not judging – some of the kit looks pretty cool! – but I personally found the transition from ‘obviously kid-friendly mobile game for fans of the book or film’ and ‘BUY A VISCOSE TSHIRT FOR £1200’ slightly-jarring. Still, I will never not enjoy anything Howl-related, and I must say the mobile site is rather nicely made.
- Europeana: This is an absolutely brilliant resource and I love it and it made me briefly annoyed about Brexit again before I remembered that it was several years ago now, and that continuing to whinge about lost political battles many years after the fact is the sole preserve of the w4nker. “Europeana provides cultural heritage enthusiasts, professionals, teachers, and researchers with digital access to European cultural heritage material. Why? To inspire and inform fresh perspectives and open conversations about our history and culture. To share and enjoy our rich cultural heritage. To use it to create new things. We give you access to millions of cultural heritage items from institutions across Europe. Discover artworks, books, music, and videos on art, newspapers, archaeology, fashion, science, sport, and much more.” Honestly, this is a STAGGERING collection of art and photography and history and academic work and odd ephemera, and if it were a museum it would be the size of Birmingham and I would likely never leave.
- MammalWeb: This is a GREAT project, tracking wild animals around the UK and continental Europe – it’s a volunteer effort where anyone can participate, either by manually identifying animals in submitted imagery or by setting up their own camera trap to see what critters wander into shot during the days and nights, and then sharing that imagery and information with the initiative. Lovely, and the sort of thing that might be fun to get involved with should you live somewhere where you see animals other than pigeons.
- The Doomsday Alarm Clock: This is cute – an alarm clock app which will gently wake you each morning with terrifying news of apocalyptic horror. Want to be roused at 7am by a voice gently telling you that the machine uprising is finally here? Would greeting the day with the news that nuclear war has broken out put a spring in your step? Maybe it would. This is, as far as I can tell, a hobby project by a couple of devs – there’s a link to environmental charity Earth Justice at the bottom of the page, but tbh this feels like something that could be licensed and repurposed by a climate charity and jazzed up a bit as a fun awareness/fundraising driver (sorry, I really must stop thinking about things like a fcuking PR, it’s a sickness I tell you).
- 253: I have DEFINITELY mentioned this in Curios before, but I don’t think I have ever been able to link to the original site which has just relaunched after…some years offline. 253 was a novel written for the VERY EARLY WEB, published between 1996-8, with a genuinely interesting and novel structural and thematic premise – the book is a collection of short vignettes, with the following specific technical guiderules: “There are seven carriages on a Bakerloo Line train, each with 36 seats. A train in which every passenger has a seat will carry 252 people. With the driver, that makes 253. This novel describes an epic journey from Embankment station, to the Elephant and Castle, named after the Infanta de Castile who stayed there, once. This is an example of the verbal imprecision that costs British industry millions of pounds a year. Numbers, however, are reliable. So that the illusion of an orderly universe can be maintained, all text in this novel, less headings, will number 253 words. Each passenger is described in three ways: Outward appearance: does this seem to be someone you would like to read about? Inside information : sadly, people are not always what they seem. What they are doing or thinking : many passengers are doing or thinking interesting things. Many are not.)” I didn’t stumble across this until I found the print version in a charity shop in the early-00s, but it was for years one of my favourite books and I must have read it a few dozen times over the years, mostly sitting on the tube) – the fact that the original site has been resurrected was both a wonderful reminder of the book’s existence, and a lovely look back at how it began, and at the early days of the web, and, honestly, I am getting weirdly misty-eyed as I type this so should probably stop and have some more tea. Before I do, though, PLEASE DO CLICK ON THIS ONE AND READ SOME OF THE STORIES THEY ARE GENUINELY WONDERFUL.
THE SECTION WHICH OBVIOUSLY CAN’T TELL YOU WHAT TO DO BUT WHICH IS JUST GOING TO LEAVE THIS LINK TO DONATE TO RELIEF EFFORTS IN SYRIA/TURKEY HERE, PT.2:
- Photobooth Dot Net: Do YOU want access to the web’s premiere resource for all things related to the little kiosks you use to get passport photos? OH GOOD! This is a genuinely-charming and very oldschool little site, run by a proper enthusiast who, as far as I can tell, seems to spend a reasonable amount of time wandering the world and checking out different examples of photobooth tech (and, frankly, why the fcuk not?). This is partly just an interesting overview of the history of the things and the various baroque differences in style and tech that they have employed, but it’s also got a search feature which means that you can, should you desire, find EVERY SINGLE PHOTOBOOTH IN THE WORLD (ok, fine, this probably isn’t strictly true, but) and plot your future travel plans accordingly based on whether or not you’ll able to enjoy some booth-based analogue photo fun. There are worse ways to plan an itinerary tbh.
- Cloudwatching: A meditative little webart project where any user who lands on the page can participate, either by designing their own ‘cloud’ using a rudimentary Paint-style interface or by sitting and watching as the clouds created by other anonymous webmongs glide past over a background of cerulean blue. Admittedly the slightly contemplative and gentle nature of the experience can be jarred somewhat by the occasional cloud in the shape of phrases like, er, “I Eat Kids”, but generally this is a supremely gentle way of spending 5 minutes staring at a screen.
- Shortest Lives: This is a really interesting project, but it’s probably worth pointing out at the top that it’s about miscarriage and as such, depending on your personal experience and circumstance, you might want to skip reading this one. Shortlist lives is an initiative being developed in the Netherlands – specifically at the UZ Gent hospital, at least initially – which is designed to give parents who suffer miscarriage a tangible, personal memory of their child via the medium of a 3d render of their heartbeat. “Shortest Lives originated from An-Sofie’s personal experience. In 2017 she became the godmother of Lily, the stillborn daughter of her cousin. By experiencing the loss up close, she saw how parents are left behind in a great void. What started as a project for Lily’s parents has, thanks to the support of UZ Gent and Berrefonds, grown into an opportunity to support other parents as well. Shortest Lives is an initiative that wants to give parents a powerful reminder of their child. Based on an ultrasound, it is possible to make the heartbeat of a stillborn child audible again. This heartbeat is then visualized in a unique way.” I think this is rather beautiful, personally, and I like idea that it’s something which could in theory be extended to all parents who have access to ultrasound data, wherever they are – I equally appreciate that everyone’s experiences with miscarriage are personal and different and that you might not agree.
- Calculating Devices: This is a YouTube channel whose sole raison d’etre is to explain and ‘review’ various obscure or antique counting devices. Have you ever wondered how the ancient Hittite method of enumerating things differed from that popular with Mesopotamians c.Hammurabi? No, probably not, but that is why we are here – TO LEARN (and to fill all these empty hours between birth and death with CONTENT). This is, honestly, significantly more interesting than it has any right to be, not least because if you’re moderately innumerate like I am it actually provides some useful insights into how one my helpfully go about conceptualising the concept of ‘numbers’ and thereby better deal with them (I’m…I’m not selling this, am I? FFS).
- The Aircade: I am always amazed and slightly-appalled at recruitment-focused PR efforts for the military and the various different ways that different nations’ various branches of the armed forces vie to attract the best and brightest cannon fodder into their ranks. This website is one part of the US Airforce’s efforts to entice the youth into the whole ‘death by plane’ business (I presume they spunked the rest of the budget on the recent Top Gun film which from what I could tell was effectively a two-hour recruitment ad for the airforce which people astonishingly paid to watch) via the medium of a series of genuinely really rather good games – honestly, I wanted to hate this but unfortunately all the stuff you can do here (from the in-browser elements to the downloadable AR-enabled ‘Command The Stack’ game) are annoyingly really quite fun and super-polished, as I suppose I should have expected from an institution whose annual budget is probably something comparable to the GDP of the UK. Whilst I find the very existence of this website and the associated content somewhat…icky, I have to grudgingly applaud the fact that it’s all very nicely made indeed.
- Crewmap: I can’t quite work out whether this is a genuinely smart idea for the new age of remote/hybrid working, or some sort of dreadful parody of ‘modern working life’ – YOU DECIDE! Crewmap is a platform which exists to help create bonds between colleagues and coworkers – the idea being that in a working world in which many of us see our colleagues on screens for 90% of the time, and where we don’t necessarily enjoy the same opportunities for smalltalk and socialising and general, fluffy ‘getting to know you’ chat as we might once have done whilst sharing corporate meatspace, you might want to find other ways of getting to know the other humans who are doing the same largely-pointless, soul-destroying job as you are. Basically the platform lets you set up quizzes and questionnaires about your likes, your interests and your hobbies, share them with your team members, and then get told who has what in common based on everyone’s answers – which…sounds like it might be fine in theory, but I can’t quite help get the feeling that this could quite quickly get awkward – in part because of the fact it feels, I don’t know, just a bit *too* close to the sort of thing that people associate with dating apps, and partly because I can envisage a scenario in which people end up feeling isolated rather than connected when they realise that none of their teammates like the same things that they do (“Noone else ticked the self-fisting box? I QUIT!”, etc). Still, maybe I am wrong and this is a genius idea, in which case YOU ARE WELCOME!
- Read Jpeg: I LOVE THIS. Honestly, I can’t explain how much this pleased me when I came across it (I think via Kristoffer again, but I couldn’t swear by it) – I think it’s because it feels zine-y in a way that always appeals to me. Read Jpeg is a newsletter that is just a stream of images that link out to places: “Curated streams on contemporary culture. Image essays. Subject matter varies. No text. Just images. Click–throughs for surprise. Mailed irregularly by the zip.” Really, this is practically-perfect – odd, esoteric, aesthetically (in)coherent and consistently surprising – you can go back through past editions to get a feel of how this works, but given how much of newsletterland is SERIOUS and BUSINESS and about KEEPING AHEAD it’s nice to see one that is pure vibes.
- Petswitch: I have a special place in my heart for websites which do one thing and one thing only, and which do that thing practically-perfectly; Petswitch is not such a website. It does, admittedly, do only one thing – to whit, taking your face and ‘swapping’ it with the face of any animal you choose to offer it an image of – but it does it in a way that is frankly quite shonky and a bit ‘phoned it and I promise you that when I found it this morning and had a quick play with turning myself into a sad-faced dog I lost it for a full two minutes (it was just after 6am and I was tired, in my defence). This is, to be clear, not very good at all, but it is VERY FUNNY and I confidently predict that you will all know at least one person who will be driven absolutely mental if you start sending them one of these a day featuring them combined with a series of different critters.
- WikiShootMe: Simultaneously interesting and really not interesting at all (sometimes I like to think this is the exact liminal space which Curios occupies and that’s perhaps why no cnut reads it), WikiShootMe is “a tool to show Wikidata items, Wikipedia articles, and Commons images with coordinates, all on the same map” – so basically you can look up anywhere in the world and see ALL THE STUFF that Wikipedia has on the area, on a map! From places of interest to images of the local area, it’s a surprisingly-detailed look at a particular geographical space which leads you (or at least it did me) to look slightly differently at where you are and what surrounds you and what is and isn’t significant about it.
- The List of Stews: An actual Wikipedia entry this time, both tasty AND seasonally-appropriate! Given it continues to be so chilly here that I seriously contemplate buying a pair of fingerless gloves to type today’s Curios in, it feels like this encyclopaedic list of international stewed dishes might come in handy – from Peruvian chicken to the (apologies to any Flemish readers) frankly-repellent sounding horsemeat stew with gingerbread(!!!!) of the Northern European lowlands, there is everything you could want here (presuming your ambitions tend towards ‘the consumption of largely-brown, largely-meat-based meals).
- OpnSouls: Do you ever feel like your friendships are stuck in a rut, that you don’t really talk anymore? Is your groupchat moribund? FEAR NOT! OpnSouls is an app which promises to reinvigorate your friendships by, er, sending your group of mates a question each day which you all have to answer. “Every day, you will receive a new question that is designed to spark meaningful conversations and help you learn more about your friends. You can share your own answer and see how your friends have responded, leaving comments and replies along the way.” Depending on what you and your friends are like this will either be HILARIOUS or the sort of thing that will see you all blocking each other on every single platform going within a month.
- Poline: I have to confess upfront here that I really don’t understand how this works or what it really is – that said, my total confusion as to its exact utility and purpose is part of the reason it appeals so much. As far as I can tell, Poline is a palette-creation tool – beyond that, though, well…Let me give the the website’s own descriptor text and see if you can make head or tail of it: “”poline” is an enigmatic color palette generator, that harnesses the mystical witchcraft of polar coordinates. Its methodology, defying conventional color science, is steeped in the esoteric knowledge of the early 20th century. This magical technology defies explanation, drawing lines between anchors to produce visually striking and otherworldly palettes. It is an indispensable tool for the modern generative sorcerer, and a delight for the eye.” GREAT! So, er, one for all you generative sorcerors, then – erm, could one of you possibly explain to me what the fcuk is going on here, should you have a second?
- WTFDoesThisCompanyDo: Balls, this is a GPT-ish thing and should be uptop, but, well, I can’t be bothered to scroll back and add it in. Sorry. ANYWAY, this is a genuinely fun application of the tech – plug in any website url you like and the tech attempts to give you a description of what the company actually does based on the copy on the homepage. What this is useful for, I’ve found, is for gently pointing out to people that the copy on their website is meaningless dogsh1t, as evidenced by the fact that not even a machine trained on the entire corpus of the web can make sense of it – honestly, this is particularly good for agencies and I strongly advise you to pick your own favourite ‘disruiptive creative studio’ and see whether or not the machine can work out that what they do is ‘make videos to sell petfood’.
- CandyCoated: I feel I ought to explain that I stumbled across this site via this article rather than through any sort of personal interest (not, of course, that there’s anything wrong with being into this stuff!) – that out of the way, let’s all enjoy browsing through the esoteric world of hand-made, bespoke, inflatable rubber fetishwear! Have you ever wondered where people whose idea of a perfect sexy time is, say, being inside a space hopper while someone attempts to tug them off from the outside, find an appropriately-modified space hopper? WELL WONDER NO MORE! Personally speaking it was the combination whale toy-slash-dildo that gave me greatest pause, but you will all find something to love here.
- Manifesturbation: It seems that every few years someone reinvents the basic lie at the heart of new age classic ‘The Secret’ – to whit, that you can basically get the universe to give you what your heart desires simply by following a few simple instructions. There was the original, there was Cosmic Ordering, there has been the modern reframing of it as ‘manifesting’, and now…now we have this. I feel compelled to point out that this link was a reader submission (I am sworn to secrecy), and that there’s a very slight possibility that your employer might not massively enjoy you clicking on the link whilst on their dime…and now let’s dive in! What would you say if I were to tell you that there was a way of securing EXACT SUMS OF MONEY through no activity or effort whatsoever other than having a VERY MINDFUL W4NK? I imagine what you say is “fcuk off Matt, I am not a moron and I was not born yesterday, this is obviously rubbish, what are you trying to sell me?” – but, seemingly, there are others for whom the answer would instead be “SIGN ME UP FOR CASHW4NKS!” And it is for these people who Manifesturbation exists – people who, in exchange for the secret of how to “co-create your life through juicy, lit-up pleasure instead of to-do lists, overwhelm, stress and struggle” are willing to fork out a (NON-REFUNDABLE) $200. Yes, that’s right, for a mere two hundred quid you too can have someone tell you ‘facts’ about how if you just finger yourself properly you’ll be granted your heart’s financial desires (except, er, the small print is keen to point out that there are no guarantees here). This is, let’s be clear, a quite astonishing grift and the sort of thing that I feel no compunction whatsoever being mean about because whoever is doing this (my girlfriend pointed out that they also do a podcast about their sex lives which is quite famous, which I guess figures) is, I feel comfortable saying, making this all up and peddling snake oil (also, please watch the testimonial videos – those people do not look like the most discerning of consumers, is all I’m saying here). Still, MANIFESTURBATION! What a fcuking world we live in.
- Order Automatica: Finally this week, a palette-cleansing card-battling game which will hopefully clear your mind from the last link and which is a really pleasing pixel-y looking timewaster which you can happily while away a few hours with.
By Deon S’Souza
LAST OF THE MIXTAPES THIS WEEK IS THIS LOUNGEY CRACKER FROM LOUIS LARGO!
THE CIRCUS OF TUMBLRS IS ONCE AGAIN EMPTY! POOR TUMBLRS!
THE TROUGH OF (INSTA) FEEDS!
- Typewriter Artist: Anatol Knotek makes art from an old typewriter, typography and, occasionally, coloured ribbons. This is minimal and beautiful and feels like the art style of a campaign that’s just begging to be commissioned.
- Models Architecture: This Insta account really should be named ‘Architectural Models’, but I will forgive them the confusing reversal because there is something SO SO LOVELY about architects’ models and dioramas, and there’s some superb work here.
LONG THINGS WHICH ARE LONG!
- Governance vs Moderation: We kick off the longreads section this week with a piece by Ethan Zuckerman that looks at how communities work online, and how one might go about optimally organising them in a way that minimises all the awfulness that tends to happen when people congregate. In a week when Twitter’s looked even more parlous than usual under the Glorious Reign of That Fcuking Man (honestly, the site’s just…fcuked at the moment, creaking and seemingly held together by string – if you’ve not read it, Casey Newton’s account of What Is Going On is by turns funny and depressing) it feels appropriate to share something that takes a more considered view of the problems inherent in managing the twin poles of rights and responsibilities – the conclusions here strike me as broadly sensible, with the overall inference being that it is better to allow communities to develop their own rules and regulations and standards and codes than it is to attempt to impose them from above, but that this requires a degree of engagement in the management and maintenance of said communities that we have previously been (mostly) willing to commit.
- What Is CoreCore?: I’m including this mainly as I am sick of seeing variants on this piece floating around the web and this happens to be one of the better ones I’ve read about this ‘new aesthetic’ – look, it’s a new twist on Dada-ism, can we move on now please? I did quite like this closing para, not least because it gave me momentary hope that I never have to hear the ‘core’ suffix ever again: “The discourse around corecore feels like a work of art in itself, indicative of the fluid and fast, often convoluted way art genres emerge and calcify on the internet. The scene is being historicized in comment sections in real-time as viewers argue it’s being ruined or claim the subgenre was always meant to be Dadaist-absurd. New offshoots are spawning by the week—there’s corecorecore, nichecore, coretok. Publications are writing it up. Podcasts are puzzling it out. Everyone is erratically trying to clutch at what it means. True to its name, corecore is like a project we’re all working on, the most nebulous “aesthetic” of all time.”
- Turning GPT Into DAN: Or, ‘the latest series of attempts to mess with ChatGPT in order to see what sort of mad or terrible stuff you can get it to say in complete contravention of the OpenAI guidelines’ – this is a link to a post in a subReddit dedicated to attempting to ‘jailbreak’ the software, in which the OP explains how to use DAN, a ‘roleplay model’ to effectively cut out all the guardrails ChatGPT has built in to stop it explaining in detail how to dispose of your lover’s corpse with quicklime and a shovel. This is all sorts of fascinating for all sorts of reasons – partly the general arms race that we’re seeing between the people who make this stuff and the users trying to subvert it for the lols, but also the fact that, again, NOONE SEEMS TO UNDERSTAND WHY THESE TRICKS WORK (see also this, and this), and there is something halfway-sinister, even accepting the unnecessary anthropomorphism inherent in the idea, of ‘scaring’ the software into obeying you.
- How One VC Is Automating Themselves: I am increasingly interested in seeing practical examples of the ways in which people are seeking to integrate AI stuff into their lives and work – the theory’s fine, but there’s something particularly fascinating about seeing the extent to which this stuff can be useful RIGHT NOW. This is a post about how a VC called Yohei Nakajima is using the suite off tools currently knocking about to make themselves more efficient – the piece is surprisingly detailed on the ‘how’, but here’s the ‘what’: “Nakajima is a GP at a small venture firm, Untapped Capital, so he has a GPT-3 bot that answers common questions from founders. It emails responses to them and includes a way for him to refine them over time—so it continually improves. He has another bot that automatically summarizes every email interaction he has. He reviews these before meetings so that he’s always up to date on founders and LPs. He has another one that, given a startup’s website, can draft an investment memo that includes data like the company’s value proposition, a description of its product, its likely competitors, and even the sentiment from its Product Hunt launch. It goes beyond his work life, too. He has a bot that summarizes and makes searchable emails from his kids’ schools—so he never has to spend time looking up deadlines or important events again.” Small aside – reading all the millions of words about GPT and its ability to write for us, and seeing all the people attempting to take ‘written communication’ out of their lives altogether via the medium of AI-penned emails and notes…does noone actually enjoy writing?
- The UK Media Nepo Map; One of the main drivers of The Discourse this week, at least amongst a specific class of the UK’s media-literate Twitter users, was this map compiled by Mic Wright which neatly delineates some (but very much not all) of the familial relationships which are the bedrock on which the modern media landscape in this country works. I don’t mean to speak for Mic here – he very much doesn’t need me to, for one – but I don’t think the point of this is to suggest that this is the only problematic aspect of the structure of the fourth estate, more that this is just one aspect of a very tangled web that encompasses private and public schools and Oxbridge and which if you attempted to map it fully would require the sort of 4-d model that makes datavisualisationmongs weep hot tears of eroticised fear. Anyway, regardless of your opinions on whether this is a) THE MOST DISGUSTING EXAMPLE OF ENTRENCHED PRIVILEGE THE WORLD HAS EVER SEEN; or b) MAN DISCOVERS HOW WORLD WORKS, FAINTS, it’s an interesting overview and I don’t think anything like this has been visualised in this way before, and it’s worth remembering next time you see a new senior appointment at one of the big media houses.
- The History of the Web: The History of the Web has been a newsletter and website for a while, but Jay Hoffman has turned it into a book which is all available online, and if ever wanted to read a comprehensive historty of how the infrastructure that enables you to read this fcuking newsletterblogtypething in the first place actually came to be then you will very much enjoy this.
- AI Twitch Streamers Are Doing Reaction Streams Now: You know how there have been various flavours of apocalyptic predictions about what the coming AI tsunami is going to do to the content landscape, and how we’re going to end up in a situation where the majority of stuff produced online is made The Machine and watched by The Machine with no need for unpleasantly-messy meaty involvement at any stage? Yes, well, it’s already happening, sort-of. This piece looks at AI VTuber NeuroSama (cobbled together from a combination of LLM and text-to-speech voice-synths) whose creator has started setting it up to do reaction streams – so, to be clear, this is an AI simulating its reactions to other content on Twitch. At some point last week, before it got banned for transphobia, NeuroSama started ‘watching’ and ‘reacting’ to the infinite AI-generated Seinfeldish stream ‘Nothing, Forever’ – so, there you have it, in early 2023 we have what I think is the first instance in recorded human history of an artificial intelligence watching and reacting to content produced by another artificial intelligence. However you might want to quibble about definitions and meanings and ‘AI’, it’s hard not to feel a little bit like this is a small Rubicon we’ve just left behind us.
- They’re Working On AI Identikits Now (This Is Why It’s A Bad Idea): It seems redundant to have to point out that ‘using nascent AI image-generation which we don’t 100% understand and which we are already aware is a hodgepodge of biases based on the training data used to create it as a means of creating mugshots for the identification of criminals’ is not necessarily the smartest idea in the world, and yet here we are.
- The VFX Industry is Broken: Is there any industry anywhere in the world which doesn’t feel a bit like it’s going to sh1t right now? AI hypetrain aside, of course. Apparently the CGI FX world is all sorts of screwed as a result of the insane demands being placed on digital artists by studios and directors who – and this will shock you – don’t necessarily always perfectly understand the technical requirements and limitations inherent in CG tech. This is, I think, a growing problem that is only going to get worse while we still rely on (lazy, inefficient, uselessly-meaty) PEOPLE to do this stuff – there is SO MUCH involved in much of modern business and commerce nowadays that involves stuff that lots of people simply have no idea about I mean, look, it’s been 20+ years ffs and 90% of people have no fcuking idea how to build a website, how long it might take, how much it should cost and what needs to go on it, so I can only imagine how frustrating it must be to work on stuff that is actually quite technical and quite hard and which is constantly being interrupted by people saying things like ‘yeah, but you can just do that with AI, right? We need it in an hour’.
- The MegaPorkFarms of Modern China: I’ll be honest with you, this is included mainly because it’s the most nakedly ‘dystopian future’ thing I have seen all week. Forget the robots, forget the AI, it’s the concept of a 25-story MegaPigFarm in a fifth-tier Chinese city that really gave me the futurefantods this week. The photos in here are just sort of dizzying.
- VR Sex Parties: I have to admit that this link felt ever so slightly like coming home – not, to be clear, because of any personal experience, but more because ever since Web Curios first began on the H+K blog all those years ago I have been sneaking in links to teledildonics and marveling at the endeavour of people who really, really want to bone in VR. It seems that every new generation of tech gets its own version of this stuff – from Second Life to Google Glass and now to VRChat (for all I know there was a burgeoning teledildonics hacking subculture in Runescape, but I sort-of hope not). There’s nothing particularly new here, which I find interesting in itself – over two decades or so, the community of people who are into this stuff has remained niche at best, and the tech, whilst increasingly evolved, is still mostly homespun and jury-rigged, and I wonder whether there’s just simply a hard cap on the number of humans living at any one time who want to strap a motorised piece of latex onto their groin and enjoy some long-distance robofrottage.
- The British Rail-designed Flying Saucer: This,,,this appears to be a real document, and I am AGOG. From 1970 comes this old application to the patent office, in which British Railways Limited sought to get official recognition for their design for a propulsion unit for a flying space vehicle. I AM NOT SH1TTING YOU THIS IS LITERALLY A BLUEPRINT FOR A FLYING SAUCER (scroll down to Fig.1 if you don’t believe me). I have so, so many questions – is this really real? What happened to this project? Was a prototype ever built? WHY AM I STILL FORCED TO SPEND HUNDREDS OF POUNDS ON SUBSTANDARD RAIL TRAVEL WHEN I COULD HAVE HAD A FCUKING SPACESHIP BY NOW? I would love to know more about this should any of you happen to be in possession of some facts.
- Chicken Stock: I really enjoyed this article, which begins with a bit of a disquisition about the magical properties of chicken stock but then broadens into something far more interesting – Noah Galuten writes about their efforts to try and find out what’s in shop-bought stock cubes and quickly discovers some fascinating (and not hugely appetising) facts about the way in which the industrial food production machine works. There’s something perennially intriguing about getting a glimpse into worlds you know nothing about – so it is with this depiction of the b2b side of the food production market, where foods become chemical formulas and the quality of a particular type of chicken stock is measured not in terms of flavour but in terms of the fact that “Liquid is flowable at ambient temperature.” If you have slight qualms about Big Food then this is unlikely to make you feel better about the whole ‘industrialised food production at scale’ thing.
- I Hired 5 People: I don’t, as s rule, include longreads unless I think they are good or interesting, and I certainly don’t tend to feature stuff (in this section at least) just to point and laugh at it, but, well, this post is simultaneously really interesting and ALSO something that just begs to be pointed at whilst laughing. So. In this edition of their newsletter, Simon Berens writes about how they experimented with hiring a bunch of assistants to (and I am not making this up, this is literally what they were hired to do) sit behind him each day as he worked to stop him from becoming distracted and wasting his time. Take for a moment your initial gut reaction to that statement – I am sure you have one – and now go and read the piece and enjoy all the many, many other feelings that you will experience as you read Berens’ assessments of his assistants, their value, and his expectations as to what they might do for him. This is one of the most impressively tone-deaf things I think you will read all year.
- I Turned Off My Appetite: Paul Ford writes about taking Mounjaro, a drug designed for the treatment of diabetes which, in common with others related to that pathology, also has seemingly miraculous properties when it comes to managing weight loss. This is interesting not because of the detail about the drug or the weight loss, but about what Ford seems to realise as he’s taking the treatment – as someone whose life had, per their description, was previously ruled by their appetite, removal of that appetite fundamentally changes something about who they are and how they experience the world, and there’s something both hopeful and…weirdly sad about what that might mean for their future.
- Meet The Archive Moles: This is on the one hand what can practically be described as a very, very long puff-piece for a particular publishing house, but I will forgive the author their self-promotional excesses because it’s such a charming read – the piece looks at how publishers of forgotten books go about scouring the world for great works that have either languished forever in unwarranted obscurity or which have fallen oddly out of fashion, and the picture it paints (which is basically of a bunch of tweedy literary detectives scouring the stacks with magnifying glass in hand) is basically catnip for anyone who considers themselves even vaguely bookish.
- Do You Ever Feel Like A Plastic Bag?: YOU MUST READ THIS. Honestly, I literally cried laughing whilst doing so and I think you will all absolutely adore this – Michael M writes about being in a not-very-successful band and the singular experience of performing the worst ‘acoustic cover version of a pop song for a live radio show’ ever committed to broadcast. I cannot stress enough quite how spectacularly good this is – were it not for the fact I still had to write the rest of this fcuking thing I would go and read it again RIGHT NOW.
- The Crack Up: Finally this week, a piece from the past which you can either enjoy as a great piece of writing about the personal experience of a mental breakdown, or as a wise assessment of how we’re all of us broken and irrevocably messed up in our own particular ways – F Scott Fitzgerald, writing in Esquire in 1936, about the time he ‘cracked’. What’s startling about this, other than the candour, is how modern so much of it feels, and how many of the assessments Fitzgerald make about The Lot of Modern Man are very much still applicable nearly a century hence.
AND NOW, MOVING PICTURES AND SOUNDS!: