Logical piano: 2024-W25

Solstitial reflections on Boolean logic, divination, and ritual machines.

Undertakings:

• Ritual machines: Another week, another art application deadline. Teamed up with two remote collaborators for a sprint to meet a commission exploring the intersection of AI and ritual practices. Coordinating across time zones and platforms (Miro boards, web docs, WhatsApp, etc.) posed its own challenges, and we ended up working in a kind of threaded relay, breaking only for dinner and a walk around the block to pet some neighbourhood cats. Don’t know what our chances are, but I’m content that we submitted a convincing proposal, and enjoyed the process.

• Static site: Spent most of Tuesday on the ongoing website redesign. I’m doubling down on Inter as a primary typeface, but have tweaked some of its OpenType features to make it more distinctive. Also added some drop shadows to block quotes, to separate them from the surrounding text, and fiddled around with my use of tables. It’s been a relief to make a clean break with WordPress, even as grappling with CSS and typeface weights has felt, at times, like squinting my way through an optometry appointment (“Option A, or … Option B?”).

• Re/imagining technology: Set aside a couple of days to work on FoAM’s “Anarchive”, getting stuck into our latest multi-author reader, “Re/imagining technology”. I’ve been trying to recontextualise some work from the early 2000s for a contemporary audience, tacking between past insights and future possibilities. From midweek, I shifted from editing to starting work on my own contributions; figuring out the parameters for a planned text sketching one possible scenario for the (near-)future of remote collaboration.

• Querent.py: Completed an initial prototype of this tarot-reading programme I’ve been working on, a significant early milestone in my Python journey. Inspired by, and striving to reconcile, the works of Ian Bogost and Alejandro Jodorowsky¹, this undertaking devoured a not-insubstantial chunk of the past fortnight. A central early challenge came in roughing out a data structure flexible enough to capture the tarot’s associative symbolism, with the EAV model emerging as an unexpectedly promising approach, despite its mixed reputation.²

Side-missions:

• The log-compost continuum: Published a log entry on barefoot coders and malleable software, bridging my recent Python misadventures and longer-standing interests in improvisation and appropriate technology. I’m trying to get back into the habit of writing and publishing on the web, something that seems to involve figuring out what qualifies as a sufficiently “loggable” line of thought. At a structural level, I’m also trying to get greater clarity on the relationship between my website’s log, as a chronological record of discrete observations, and this new “Compost” digital garden, a mulch of interconnected hunches and notes-to-self; slow, rhizomatic, and iteratively refined.³ The next step is to improve these notes’ navigability, unpicking the assumptions and operational logics of backlinks and the network graph format. I’m aiming to maintain this as a living, breathing knowledge base, running alongside (and underneath) the rest of my site.

• Excavating ELIZA: As part of “Enexata,” mine and Tim’s “sprawling mutual aid skill exchange” (his words), I’ve forked a Python emulation of ELIZA, the pioneering chatbot developed by Joseph Weizenbaum at MIT in the 1960s. To better understand ELIZA’s inner workings, I’ve started annotating the Python codebase. I’m treating this exercise as a form of experimental media archaeology, hunting for traces of the programme’s original social and cultural contexts, while considering how its architecture might be repurposed for contemporary conversational interfaces. Alongside this ELIZA side-mission, I also made some headway on (though didn’t finish) this article on “ugly code” at Russian search engine firm Yandex. Sitting at the juncture of valuation studies and critical technical practice, the article examines how aesthetic judgments and values shape software development. Aiming to discuss these themes with Tim next week, exploring how they might shape the next stages of our work together.

• Taming complexity: This week, I’ve been grappling with the challenge of instilling some kind of order on a firehose of ideas, particularly for the multi-author reader I’m working on with FoAM. Enter “Lineage”, an Obsidian plugin approximating much of the functionality of Gingko Writer.⁴ As I understand it, the plugin adds a layer of invisible HTML comment “breaks” to markdown files, visually interpreting these as “cards” that can then be hierarchically organised with parent/child dependencies. I’m still experimenting, but the multi-column, drag-and-drop interface could be a game-changer; dramatically reducing the friction in outlining complex texts.

• Huff-duff: A standing call with Andrew L-B yielded some reflection on professional legibility in an increasingly fragmented digital landscape. In a blog post this week, he draws an analogy between WWII-era high-frequency direction finding (HF/DF or “huff-duff”) and the information asymmetries inherent to the challenge of building (assembling?) a career. Like a submarine, he argues, we “transmit” our skills and contributions, hoping to be “detected” by allies and would-be collaborators. But this analogy raises some tricky questions: constant transmission can be exhausting, and risks reducing us to a narrow, distorted slice of our capabilities. This challenge is being compounded by the erosion of mass-userbase social media platforms, with our “transmissions” fragmented across digital spaces. Perhaps the goal isn’t to be a perfectly detectable submarine, but to cultivate a network of trusted huff-duff stations: allies, mentors, and collaborators who can appreciate our evolving mix of skills; helping us navigate, even as we change course.⁵ (※)

• Boolean loops: Found myself stuck in an unexpected flow state late Friday night, plumbing a speculative rabbit hole with the newly-launched Claude language model. Sparked by a passing reference in a Benjamin Labateut essay, this, combined with a familiarity with “bools” from my recent Python noodlings, led me to enlist Claude in working through the implications of George Boole, father of Boolean logic, surviving an 1864 bout of pneumonia and living into the 1880s. The model’s speculations ranged from the plausible (earlier applications of probability theory in actuarial contexts, more reliable telephone networks) to the offbeat (“True/False cutlery sets”, Boolean gardening, a proliferation of logical pianos). As someone who’s been drawn to counterfactuals since my teenage years, I’m now starting to think about more structured approaches to these machine-assisted thought experiments. And while I’m not yet sure what form this might take, there’s definitely something interesting here.

Next week:

• Preparing the ground to pivot to phase two of this Python-ethnography collaboration with Tim, having just nerd-sniped him into bootstrapping Humus, a permacomputing-aligned “composting database.”⁶
• Then to London Monday afternoon, where I’m catsitting while D’s sister and her fiancé brave predicted rain (?) at Glastonbury. I’ll be volunteering at the RAI Anthropology & Education conference, lured in by a run of sessions on how generative AI is starting to sluice into anthropology education and ethnographic methods.
• Catching up with friends and colleagues while in London.
• Likely capping the week with a day at the Digital Folklore conference, before heading back up to York.

← Earlier