I was watching this Veritassum about snowflakes the other evening. The episode features ‘snowflake scientist’ Ken Libbrecht about his work and it is all really interesting. Of course, they get to the inevitable question: ‘Is it true that no two snowflakes are the same?’ At this point, Libbrecht laughs and points out that the question is somewhat ludicrous; no two anything are the same: No two trees, animals, rocks, grains of sand, salt crystals are the same so why would it be any different for snowflakes? I imagine the spirit of the thought that ‘no two snowflakes are the same’ is based on the seeming contradiction between their visual geometric-ness yet lack of any uniformity. Something in us imagines there must be some mass production at work for such forms but I find this reminder that everything is unique quite beautiful.
Simulate this
‘To make our future real, simulation is the answer.’ The lede is buried a few paragraphs down in this Nvidia blog on Earth-2. The audaciously named new hardware project Nvidia are building in their metaverse – ‘Omnivrese.’ Nvidia will invest more in faster computation to try and achieve this with the aim of modelling climate change, and in doing so, contribute significantly to climate change. Matt Webb pointed out another thing this week; a chip with 1.2 trillion transistors that can apparently simulate the things ‘faster than the laws of physics.’
As Matt points out, this is a rather spurious claim. We’ve been able to simulate things faster than physics like the movement of celestial bodies for ages, long before computers, but there may be an application at the atomic or micro scale. In fact the test run on it was fluid dynamics, which as anyone who’s ever got me on this in the pub will know, is almost impossible to do in real time partly because a) it’s really hard and b) mathematicians haven’t actually solved fluid dynamics.
Nonetheless, the provocatively named ‘Wafer Scale Engine’ manages to run water physics faster than real time – no small feat. There’s an interesting reverse hardware tendency here as well; the designers of the engine suggest that the prevailing logic of daisy-chaining a bunch of CPUs and GPUs together is not as effective as simply building bigger chips – a reversal of decades of miniaturisation.
This tracks with something that’s been floating around since I read that Hardware Lottery paper, the crux of which is; we only got this type of computation because these were the bits and bobs that were lying around in labs but they’re probably not the best bits for the job. See also the well documented subversion of the GPU for AI despite originally being designed to send pixels to a monitor in Image Objects and as Murray Shanahan describes in this week’s Exponential View: ‘[I]t’s a sort of hack. It’s repurposing of a technology that was meant for something completely different.’
All this tracks back to this metaverse thing as well – the big Internet rebrand. The socials are trying to construct captive economies but Nvidia are more interested in simulating the Earth, again; weird flex, but ok. An interesting thought experiment to do is to do a classic design school crit review on the metaverse ask what demand the metaverse is responding to? What audience or user need has been identified that the metaverse is a response to? What is the problem that it is trying to solve? I know this seems a fickle thing to ask of a lot of so-called innovations but it quickly becomes obvious that the problem is the hard ceiling on attention, time and property that the big socials have hit and the need to artificially raise that ceiling by crafting whole new economies. And the audience for these metaverses are really the companies themselves, not users. So the metaverse is basically a guarded open-pit mine built with coincidental, repurposed hardware that’s being used as the hastily assembled barricades. It will probably work too, simply because of all the things I go on about; the language of inevitability, the foreclosure of alternatives, the co-opting of everyday interaction inside these walled gardens.
Maybe it’s just the people I hang out with but a common complaint I have never heard is ‘I wish I just had more Facebook in my life and it could store all my stuff.’
Is there a possibility for us to construct our own metaverses? That might be nice, I suppose something like Minecraft or even the popularity of Animal Crossing hint at that – a world of your own you can build that you can bring people into that isn’t quite as hostile and extractive? (Ironic given the title of Minecraft). Even then, as J-Paul Neeley pointed out in a recent chat, it would still be built on the back of Amazon Web Services. This is perhaps (again, coming up in the same chat) a space to revisit the Dewey-an ‘public’ – start from scratch. Rather than a focus on platforms, what are the objects, issues and loci around which a public assembles and then what is the appropriate metaverse for that public that they can build? What tools might they use?
Is there something better than the. open pit mines of the socials, the deep sea mining of NFTs or the hubristic folly of whole-Earth simulation that you can do with fast computers?
Short Stuff
- A marketing company has set up ‘Earth’s Black Box‘ – a plan for a megalith in Tennessee that will record 30-50 years worth of data documenting the demise of human civilisation. Sort of like a fancier decline.online with a physical presence. Some notes of skepticism: The website appears to show a stream of tweets with no indication as to methodology from the marketing company behind the project. It may be a somewhat speculative proposal to drum up activism on climate change, in which case fair enough really.
- Spotify founder Daniel Ek has invested a bunch of money in a startup doing military AI. Another type of shareholder rebellion is the artists and users of the platform calling for a boycott.
- Quantum spin magnets finally observed.
- The forgotten advertising of climate change denialism. There’s a great graphic design project in there somewhere.
- https://www.independent.co.uk/arts-entertainment/music/features/long-blondes-band-landfill-indie-b1972792.html
You probably saw that I shared the news. I’m leaving London College of Communication after an amazing long journey to go to Arup Foresight as Design Futures Lead. I’ll be starting around the end of February but leaving LCC imminently to take some time off. I’m very sad to go, I love LCC and my friends and colleagues, but it was the right time and the right opportunity. My old job is up and advertised and if you want to ask any questions, get in touch.
Anyway, I also love you, but you know that. Speak later.