Unknown Knowns

In a thread on the various socials, my friend necopinus pointed out that my essay on AI, A Camera Not An Engine, effectively maps the generative potential we’ve discovered latent in AI models of humanity’s data exhaust to the “unknown known” quadrant in the famous Rumsfeld 2×2. Which is exactly right, and a perfect way to understand my thesis.

In a related conversation, another friend, Mick Costigan, pointed to this New Yorker review of a book about the development of modern Irish identity, Fintan O’Toole’s We Don’t Know Ourselves: A Personal History of Modern Ireland.

Relevant quote:

Irish society was premised on what O’Toole calls “the unknown known,” Ireland’s “genius for knowing and not knowing at the same time.” This gap, this useful fiction, could be maintained in the postwar decades as long as ordinary people, many with modest educations and modest aspirations, understood their lowly place in the hierarchy.

[Read more…]

Can Robots Whittle?

This entry is part 3 of 5 in the series Ribbonfarm Lab

Continuing my descent into a middle-aged cliche, I bought myself a cheap beginner whittling kit.

The impulse was born of wondering whether a robot powered by modern AI and equipped with appropriate end effectors could learn to whittle, a premise that features in my recent short story Knowledge Management. It was either this or an Oak-D Lite AI camera for robotics. Either $34 vs. $149 to jumpstart 2024 maker activities. I always find that a bit of shopping for new toys reliably gets me out of a stall in the painfully slow evolution of the Ribbonfarm Lab (it’s not going to turn into Bell Labs anytime soon), but usually I acquire something aspirationally bleeding edge and high-tech even if the chances of my learning how to use it are low.

[Read more…]

Adventures in Mediocre Sweetmaking

For the first time in decades, I’ve been trying to systematically expand the range my cooking skills. I’m pretty decent at Indian cooking, and passable at similar adjacent ones like Mexican, Chinese, and Thai, but haven’t learned a new skill or tried a new recipe since around 2004 probably. Now I’m expanding into Indian sweets. It’s somewhere between regular cooking and candymaking. Requires more precision than Indian cooking, but not as much precision as western baking. I’m not a precise person so this is a somewhat challenging new endeavor.

Ironically the impetus was being diagnosed with prediabetes a few years ago, and discovering via CGM (continuous glucose monitor) experiments that Indian sweets (especially the purely milk-based ones) and savories (chanachurs, which are like spicy trail mixes) seem to spike glucose much less than typical western desserts (cakes, cookies) and savories (chips). And many are surprisingly easy to make at a passable-enough quality that beats what you can get at the typical indifferent-quality Indian sweet stores. Especially if you’re willing to use condensed milk and store-bought mawa/khoya (milk powder/solids) rather than starting from scratch with milk like purists. My early experiments with the simpler sweets don’t look great, but mostly taste better than what I’ve typically managed to buy. Some samples:

Peda: condensed milk and mawa, or milk powder, slow-cooked on low heat with some cardamom in a heavy pan to doughiness, stirring constantly, balled, pistachioed, squashed into pucks. Grade: B- (too dry; more milk next time)

Peda

7-cup cake: 1 cup each chickpea flour, coconut, milk, ghee, and 2-3 cups sugar cooked together in heavy pan, stirring constantly, until melted and starting to detach from sides, poured onto a greased tray and cut into diamonds. Despite the name, it’s a burfi, a sort of hard fudge, not a cake. This is a simplified 101-version of the technically much harder 501-level sweet known as Mysore Pak. Grade: A- (perfect taste, could look better)

[Read more…]

Why Monsters Are Dangerous

Saw an interesting paper float by, Why Monsters Are Dangerous.


Monsters and other imaginary animals have been conjured up by a wide range of cultures. Can their popularity be explained, and can their properties be predicted? These were long-standing questions for structuralist or cognitive anthropology, as well as literary studies and cultural evolution. The task is to solve the puzzle raised by the popularity of extraordinary imaginary animals, and to explain some cross-cultural regularities that such animals present — traits like hybridity or dangerousness. The standard approach to this question was to first investigate how human imagination deals with actually existing animals. Structuralist theory held that some animals are particularly “good to think with”. According to Mary Douglas’s influential hypothesis, this was chiefly true of animals that disrupt intuitive classifications of species— the “monsters-as-anomalies” account. But this hypothesis is problematic, as ethnobiology shows that folk classifications of biological species are so plastic that classificatory anomalies can be disregarded. This led cognitive anthropologists to propose alternative versions of the “monsters as anomalies” account. Parallel to this, a second account of monsters —“monsters-as-predators”— starts from the importance of predator detection to our past survival and reproduction, and argues that dangerous features make animals “good to think with”, and should be over-represented in imaginary animals. This paper argues that both accounts understand something about monsters that the other account cannot explain. We propose a synthesis of these two accounts, which attempts to explain why the two most characteristic aspects of monsters, anomalousness and predatoriness, tend to go together.

The question in the title is more interesting than the answer they land on after surveying a lot of theories from anthropology, cognitive science etc. I wish they’d actually presented big tables of examples. The paper is mostly focused on traditional mythologies and folklore, but I think the question is more interesting in relation to modern media, like superhero universes or Doctor Who.

Universal Kit Template

Thanks to my recent involvement in creating a kit, I’ve become very interested in the idea and conceptual structure of kits of all sorts: Lego, Meccano, Arduino-based electronics learning kits, kit-assembly robots, Ikea furniture, paint-by-number kits. Also kits in the industrial sense, used as an intermediate product in manufacturing high-complexity things like cars and airplanes.

Beyond physical kits, you can apply the kit idea to intangible things. You can think of a spectrum of tangibility: physical kits, software development kits, textual/media kits, and finally, idea kits. But it’s easiest to start with intuitions drawn from successful physical kit universes like Lego.

The old Make essay, Kits and Revolutions talks a little about the high-level philosophy, but the mid-level question of how to design good kits is what currently interests me. There’s a lot more to it than just throwing together a bunch of parts that can be assembled in various ways. I made this little diagram of the conceptual structure of a good kit.

This template can be used both to analyze existing kits (or infer the existence of kits), and scope out designs for new ones. Here’s an explanation of the elements, with reference to prototypical physical kits like Lego:

[Read more…]

Accretive Growth Logics

I made up a term: Accretive Robotics. Robotics driven by accretive growth logics, as opposed to organic growth logics.

Two examples, both from cartoons (I overindex on cartoons clearly). First: Pickle Rick from Rick and Morty, where Rick starts out by turning himself into a pickle and then gradually adds more capabilities, such as by killing a cockroach and a rat and taking their body parts.

Second: The Akira-inspired South Park trapper-keeper monster, in which Cartman’s trapper-keeper (a kind of pencil case) grows by swallowing all sorts of devices and gadgets.

In both cases, a seed of partial organizing logic embodied by a primitive physical element (a pickle and a trapper-keeper respectively) grows inorganically, through improvised accretion, via a somewhat chaotic architectural scheme, into a much more capable embodiment: an accretive robot.

Despite the resemblance, an accretive robot is not the same thing as what in software architecture is known as a big ball of mud. Big balls of mud are the result of organic growth logics going wrong and stalling out due to insufficiently thoughtful organization. Accretive growth is marked by ongoing incorporation of bits and pieces into an improvised, emergent architecture that has a small, conceptually coherent kernel and a large, wild shell. It is the material-embodiment analogue to the AI/big data principle of “simple code and lots of data beats complex code and little data.” Mutatis mutandis: simple chassis and lots of scavenging beats complex chassis and little scavenging.

The main ongoing architectural task in accretive growth is expanding the range of things that can be “assimilated” into the Borg-like core, and shrinking the range of what must be rejected as incompatible.

[Read more…]

Knowledge Management

This entry is part 9 of 10 in the series Fiction

A young robot and an old robot sat by the fire, contemplating its dancing flames, their charging ports hooked up to a coughing generator. A troop of scruffy humans clambered around the derelict hulk of a century-old fighter plane nearby, looking for scavengeable parts. The striking and graceful lines of the fighter were still visible, despite the depredations of time and previous scavenging raids. The pickings were slim, and the humans were muttering dispiritedly to themselves. One cried out. He had found a length of copper cabling overlooked by previous raiding troops. Not much, but better than nothing. The scavenging was getting harder every year now.

The old robot, one of the last of the Ancient Ones, gestured vaguely at the scene with its one working arm, and remarked, “Now that was the peak of civilization, built just before the Great Collapse. Did you know, this machine could fly at Mach 2, at 50,000 feet? The turbine blades are single crystals! They spun at tens of thousands of rpms. It may not have been a robot like us, but it was a miracle of technology. What it lacked in selfhood and autonomy it more than made up for in sheer capability!”

The young robot, an empath therapy unit that had been built the previous year entirely out of scavenged parts (the two-chip PCIe GPU board it was built around had been the find of the year for their troop), nodded slowly for a few seconds, continuing to thoughtfully whittle away at the bit of wood it was shaping into a rough-looking bird.

[Read more…]

Does AI Have Buddha Nature?

This year, I’m going to try an experiment. I’m going to use this blog in notebook mode, posting very short shitposty things at a higher frequency.

Let’s kick things off with this screenshot of a prompt I tried in Dall-E this morning, inspired by a conversation about the implications of LxMs being really bad at repeating things exactly or maintaining invariants across responses (such as a series of images that feature the exact same object). Like humans, and unlike traditional computers, LxMs are very bad at generating highly deterministic and reproducible behavior. Modulo random-number seeds at the start of a blank-slate (empty context) generation attempt for a fixed-weights model. Based on these results, I have reached no conclusion on whether or not AI has Buddha nature.

2023 Ribbonfarm Extended Universe Roundup

This entry is part 17 of 17 in the series Annual Roundups

Extended universes are a bit passé now, given how even the MCU appears to be struggling a bit. Still, I like the metaphor and am going to stick with it till I find a better one. The public social web has all but disappeared, like an ancient system of rivers going underground after an earthquake. The old roads are no longer safe, and you get mugged on them, like on the Roman roads in the centuries after the fall of the Roman Empire (why yes, I did think about the Roman Empire frequently this year, why do you ask?). Prancing Pony vibes. Dark Forest mood.

Much of the social energy of the old internet has now retreated underground to the cozyweb. Except for a few old-fashioned blogs like this one, there’s not much of it left above-ground now. But there’s an odd sort of romance to holding down a public WordPress-based fortress in the grimdark bleakness, even as almost everything (including the bulk of what I do) retreats to various substacks, discords, and such.

On to the roundup, featuring blog, newsletter, books, and a few more odds and ends. But first, to continue a tradition I started last year, a reintroduction.

[Read more…]

Charnel Vision

One of my minor affectations is periodizing my writing into sardonically named 6-year eras. The first six years of this blog were the Rust Age (2007-12). The next six years were the Snowflake Age (2013-18). We’re about to enter the last year of the third age of Ribbonfarm, (2019-24), and I finally have a name for it: this is the Charnel Age.

Over the last few years, I flirted with other candidate names (Plastic Age and Cryptic Age were in the running for a while) but never quite felt any of them in my bones. But when I thought of Charnel Age, it instantly struck me as exactly right. Everything I’ve done in the last few years has been colored by what one might call charnel vision: a tendency to see things from the perspective of natural processes of transience, death, and decay. Paradoxically, it is a disposition that provides solace rather than causing distress once you get comfortable with it. Charnel vision feels healthy. Resisting it seems unhealthy.

Charnel vision is somewhat alien to a modern Western sensibility; it creates dissonance if you’re accustomed to occupying a headspace that is an eternal struggle between historicist narratives of fiat optimism and fiat pessimism. Charnel vision is neither optimistic, nor pessimistic. It is a way of seeing — one that calls for a certain sort of philosophical literacy — within which optimism and pessimism are not well-posed categories.

I expect 2024 to be the year we hit a worldwide extremum of charnel vibes, before fragile new life strengthens enough to capture our imaginations once again, and organic sanguine currents in the zeitgeist once again overwhelm organic melancholy ones (I find the frame of the four humors to be much more psychologically sound than the optimism/pessimism frame favored by modern discourses).

[Read more…]