The Flow of the Qi
My first real stab at computer programming was a little Hypercard stack (what a great piece of software was that!) to display constantly changing I-Ching trigrams, keyed by a random number generator. Two things happened recently that caused me to revisit this old (very simple) idea.
-
I visited a few shows at the tail end of Singapore Art Week, and particularly enjoyed “Nothing has to be the way it is: communities of practice, techno diversions” at the NTU Centre for Contemporary Art, at Gillman Barracks. I was lucky to have a chance to chat with curator Anna Lovecchio. I sympathise deeply with her curatorial intention, to reveal that technology is built upon certain hegemonic ideas and that alternate ways of thinking about technology are possible. I really liked all three works in the show, but one in particular reminded me of that old Hypercard stack. The work by Ong Kian Ping featured a camera focused on a tv screen showing static, not to create visually interesting video feedback (that old standby of 1970s video art), but for “reading” the static for variations in the electromagnetic fields felt in the room, and using those variations to drive a recital of the I-Ching coming through a pair of headphones. A great improvement on my old idea. My software “read” the variations in the Qi through the flow of numbers from a random number generator, which lives in software runtime. But true random-number generators in software are impossible. All software follows rules - even LLMs. If you know the “seed” they started with and the algorithm they use to simulate randomness it is possible to predict their “random” behaviour.1 So Ong’s approach is much more advanced, similar to the hardware random number generators used in security and cryptography. Like Ong’s work, these bits of hardware “collect entropy” from some natural source, like electronic noise.
-
My telecoms provider, Singtel, announced as a Chinese new year gift that customers were given a free subscription to the Perplexity Pro AI service. This is just an example of the AI-forward approach Singapore is taking. Now Proximity is a major example of the way AI companies are taking value from existing rights-holders, especially news publishers, by repackaging their content into LLM output (via RAG of the internet). (And they also suffering from the same original sin as all the model builders in using models trained on material without consent, compensation or credit to rights holders.) But much as I think this is legally dubious, I also have to say that they have a pretty amazing product. Napster flashback! The free version is pretty good for web search, to the extent I used it probably about half the time over DuckDuckGo or Google. I have only just started to use the Pro subscription, but it passed my first test — asking it to recreate a version of my old Hypercard app — with flying colours. It generated clean workable and well-commented code. Took me just three iterations to get what I wanted. For a super-simple app I admit, but still…
The result of all this can be found here. Much as I enjoyed Ong’s work, and the thoughts on randomness and deep-signalling it worked with, I’m still seduced by ideas of visualising the flow of the qi through trigrams, even if my attempt is pretty crude!
-
Though it is also true that cosmic rays (mentioned by Anna as a source of variability in the work) - high energy particles moving through space, or particles they excite in the atmosphere — can “flip a bit” in a memory or logic circuit and so disrupt the operation of software. This happens a surprising amount, with IBM estimating that one error per month per 256 MiB of RAM was expected for a desktop computer (Simonite, Tom (March 2008). “Should every computer chip have a cosmic ray detector?”. New Scientist). Just listen to this great Radiolab episode that tells the story of a flipped bit causing a discrepancy of 4096 votes in a Belgian election. ↩︎