Nephila: software / spider / swarm

Nephila is a creative technology studio named after a species of orb-weaving spider, which also means "love of weaving". We build tools for adversarial play, shared fictions, and collective cognition.

Cultural Infrastructure

You describe as software that sits between a product and an art project. What does that mean in practice?

Nephila is a creative technology studio and we think about our OS as both a software product and an art project because the questions it asks require working code that can scale and the cultural frame around it, simultaneously. We're interested in what a product that actually gets used, distributed, and adopted can do, and equally interested in what software looks like when it's downstream of real cultural observations and conceptual stakes rather than the other way around.

Nephila started as a browser extension that worked like a second brain, then a node-based world generator, now the OS. What's the through-line? What lines of inquiry brought you to the OS?

The through-line is probably the idea of thinking together. We're quite interested in what collective cognition might look like. Our first project fe-ed, was an attempt to give a group a shared way to read the internet: a second brain in the form of a browser extension. You'd send it pages as you were browsing, and it would build a structured node-and-edge graph for you, based on editable prompts. With se-ed we extended this into a node-based world simulator. It was essentially an entity-component system that chains together granular prompts for a persistent world, so the things a group had collected could generate new outputs: game environments, websites, characters, equations, scripts.

Working on these projects, we kept circling the question of what an interface for genuinely new forms of sociality might look like and a year ago we gravitated toward the concept of an OS. The recent rise of markdown files and folder structures as the default substrate for context management has felt validating in a way. Our impulse from the beginning was to build a system where an LLM could dynamically reorganize dispersed group interests and curation. We had a hunch even then that the next era of software would be one where humans weren't the only users, and that the interesting design question was what it would take to host both. The OS is that impulse one layer deeper: not just reading and generating together, but acting together. What does it look like for a group to think with software, rather than for an individual to be thought at by it?

Technical Architecture

Walk us through the peer-to-peer setup. Where does the data actually live, and what happens to a folder when everyone sharing it goes offline?

The app uses a series of tools from the Holepunch peer-to-peer ecosystem: Hyperswarm for peer discovery, Autobase for syncing structured data between users, and Hyperdrive for data transfer. Everything is stored in a local file store on the user's own machine. If a user goes offline, they vanish without a trace until they come back again. But if someone has stumbled across what they’re sharing and decided to share it themselves, it continues to propagate. Every peer that ever joined a collection or saved a move holds a full copy of it, so contents can outlive their creator as long as a single node is still carrying them.

Why does that matter to you? What do you get from peer-to-peer that you'd lose with a normal backend?

We wanted to align the software architecture with the conceptual intent of the app as far as we could. As a peer-to-peer system, the swarm functions as infrastructure rather than something that exists on top of it. It’s truly user-powered, and thus empowering for users. There's no central authority or vantage point from which to create a linear feed. Instead, others are encountered through network proximity and a mutual desire to share and be visible on the platform.

In practical terms, it also means users can fork and extend the software however they like. They have true ownership over the stack and can redeploy it as they see fit. If we were running on a cloud backend, someone could fork the code and get the interface, but the infrastructure would still be tethered to our servers and third parties.

You cite Soulseek and LimeWire. What specifically are you taking from them?

Using these apps feels different to the rest of the internet because they are so ephemeral: there's no permanent index and media vanishes and reappears based on the actions of individuals who are given the capacity to act as custodians as well as consumers. The lack of algorithmic curation and oversight means it feels much more like crate-digging. These platforms are mostly anonymous, yet there is a deep intimacy to what a person chooses to share and how they share it.

The OS uses LLMs for auto-organizing what gets uploaded, generating moves, driving the app. Why build it around AI at all? What considerations went into that?

Something that struck us as interesting about LLMs was how they offered the ability to store and retrieve data in a fuzzy, opinionated, and nondeterministic way. If structured databases are the invisible architecture that shapes society, our tool is an experiment into what can be made using databases that are bad at being databases, that bend and contort in real time based on the needs and desires of users, and pave the way for emergent outcomes. Specifically, it means as a collection’s contents evolve, the schema which organizes its contents can adapt in real time, regrouping media into different folders based on the wider context of the collection – hopefully tempering, but potentially adding to, some of the chaos that could otherwise happen with dozens of people operating the same shared drive.

Define a "move." What's the simplest one, and what's the most complex one you've built?

A move is a packaged action that anyone in a folder can run. The simplest is something like: post a tweet to my account. A single action, parametrized so the content can vary and mechanically trivial. The most complex moves chain multiple together: pull from a shared context folder, generate something on the basis of what's there, post across multiple platforms, log the result back to the folder so the next person to run the move sees what's already been done. When many people in a folder run a chained move at the same time, it stops looking like twenty individuals posting and starts looking like a single coordinated action made by a swarm.

You can chain moves and clone them, and twenty people in a folder can run the same one at once. What's a concrete thing a group has done with that? What do you hope people try with it?

We ran an event called Sloperator in New York, which we described as a live experiment in programmable identity. Participants used Nephila to infiltrate internet fandoms, game platform metrics, spread rumors, coordinate group actions. Beyond this kind of digital LARP, we also ran an event in Shanghai where we used Nephila to seed Chinese trends into Western platforms. What these experiments revealed was the broader range of what tools like this could be used for. We've had conversations with organizations working on questions of platform legibility - people thinking about how to organize under conditions of heavy moderation, surveillance, or algorithmic suppression. Those conversations stay private for obvious reasons, but our position is that coordinated action online already happens, mostly opaquely and mostly by well-resourced actors, clipping studios, platforms themselves, political parties. What Nephila does is take that capability and put it in the open, in the hands of communities, framed as a practice people can see and play with.

Identity & Legibility

We discussed the idea that if you're not legible (or marketable to), do you exist online? Anywhere? What's your actual answer to that?

That’s a great question and we guess that depends on how you define to “exist” online. There's “existing” as in being perceived, there's “existing” as in feeling like a person, and there's “existing” as in having enough agency. If you mean the latter and the ability to take action, to make choices, to have some kind of weight, it seems that being legible does matter. But this doesn’t have to be tied to being individually addressable. You can be legible as a BTS fandom, a Remilia cult, a WallStreetBets subreddit. These are all legible as patterns that platforms and culture had to respond to and these all managed to have agency (or to “exist”) because participants stopped being individually performative and started moving as a pattern.

You said you're interested in alternative ways of being legible, not just disappearing. What's an example of being legible differently?

Probably the most interesting examples come from our research last year with JY Yan, Jenn Leung, and Yannis Siglidis on Chinese internet subcultures, for an event we titled Dirty Products for a New Era (a phrase we'd picked up from someone on Rednote using it to describe AI slop). Chinese netizens have been navigating monopoly, censorship, and unprecedented growth in a compressed amount of time. In many ways, the Chinese internet is one of the places where identity formation moves and reinvents itself faster than any other part of the world. An example for this is the XieXiu (邪修) trend, which uses intentionally absurd practices to bypass traditional expectations in life or Slow Feet, which operates through absurd references deliberately resistant to outside translation. Then there’s phenomena like Tuanbo (团播), a livestreamed group performance where viewers pay microtransactions to direct what happens in real time. What we learned is that being legible or having agency online can come from coded references, deliberate misappropriation, strategic ambiguity. People are legible as participants in a gamified internet, not as selves being marketed to.

Nephila is named after the golden orb-weaver spider. Why that name?

Nephila is indeed the name of an orb-weaving spider whose silk changes color based on light, which is also the ancient Greek term “love for weaving”. It felt right for a studio where the process often matters more than the outcome and is also a reminder that the web is not a fixed object. We have a bit of a soft spot for spiders. Like us, spiders make things, they explore environments their bodies are not adapted for, have complex methods of turning chaos into order, and court each other. Two spiders meeting is an act often interlaced with strange tenderness and play. When spiders encounter each other, they flirt, they vibrate, and they compose. Contrary to what most might believe, they even demonstrate collective behavior that has baffled scientists for years, showing a capacity for socializing, communicating, and cogitating in surprising ways.

Cognition

On our first call you talked about seeing Leonora Carrington's paintings and feeling parts of your brain light up that you didn't know were there: art as a trigger that opens a way of thinking you couldn't reach before. Do you think a tool can do that, the way a painting can?

It requires the tool to have aesthetic stakes rather than just utility. Carrington’s work reorganizes a part of your brain, it triggers new forms of thought, it opens up landscapes that you didn’t even know you had in you. Tools mostly don't do this because they're often built to remove friction and what opens new thinking is usually a particular kind of productive friction, some kind of constraint that forces you to think sideways. The abacus is a tool that rewires, while calculator apps mostly aren't. A synthesizer or a musical instrument can rewire. So can games whether they are played on a chessboard or Steam, which is probably why games have become one of the main techniques for benchmarking measures of intelligence in the age of AI. For a tool to be comparable, it needs to demand a way of seeing the world that the user has to meet halfway. We try to design this into Nephila OS where the look and feel of the app is familiar (inspired by platforms like Discord, Miro, Are.na) yet the functionality requires users to shift from operating as individuals to operating as a swarm. Whether we succeed in making it the kind of tool that rewires is something we'll only find out from the people who use it.

When anyone can generate an artifact (code, images, video, etc), the artifact stops being the scarce thing. In your world, what becomes scarce instead? Is that a better economy for artists, or just a different one to be exploited by?

There’s been a lot of discourse on taste, which we feel ambivalent about because it seems somewhat arbitrary and aristocratic. Perhaps we are just weary of established social authority getting to solely define what’s tasteful. We’d frame the scarcity around two things: context and friction. Context is the condition under which an artifact lands with a specific community at a specific moment. Friction is the process that can’t really be shortcut, proof that someone committed presence and time, fiddled with something, developed a relationship. What becomes scarce is what requires real social and temporal density because that’s one thing a machine can’t fake and status can’t buy. We’re not sure if this will be better for artists. It’s better for artists who have a practice beyond the artifact (slow inquiry, community, process) and it’s worse for those who treat the artifact itself as the product.

< First Contact >

The launch frames Nephila OS around "first contact", Bogna Konior's idea that the internet is the alien invasion, and the alien we're meeting is ourselves, scattered across platforms and performed for metrics we didn't choose. How does that frame the event, OS, or your current thinking?

Aliens feel strangely relevant at this moment. Dario Amodei opened his last essay quoting Carl Sagan, K Allado-McDowell's theory of neural media is gaining traction, alien memes flood the timeline whenever we need to be distracted from something darker in the news. What we love about Bogna's theory is that she flips the usual story and the alien encounter she describes emerges from the internet itself, from the very technologies we built to drive our desires and our communication. For us the possibility of this is quite beautiful. It means we may be the aliens we expected to meet elsewhere. And if that's the case, we have to rethink how we relate to what's unlike us, what feels foreign, and how we interface with other kinds of intelligence. We think “first contact” as both an alien encounter and a gesture of love is one of the key things that will orient us as humans in the years to come. That's the philosophy behind the event, and behind the way we will demo the OS: as a contact layer between game worlds that move each other and contaminate each other.

How do you think about performance as part of how the software gets introduced?
Software for collective action only really exists when a group is using it together, so introducing it through a performance felt fitting. We have incredible artists running real-time simulations and performing live who will be bringing their own alien and sonic worlds into the space. We don’t want the audience to feel like they are attending a product launch, we want them to feel like they are inside a working instance of the thing.

Why Paris? Beyond the timing being good, does the city have anything to do with the work?
Paris is a place where culture is undeniably upstream of technology. We’ve been thinking about this a lot lately: how can the things that constitute the essence of artistic and cultural activity directly shape the development of the technologies that increasingly define how we live? Places that have been driving tech development kind of have an opposite hierarchy where tech has mostly replaced mainstream culture. Meanwhile, Paris itself has quietly become one of the most important cities for AI outside Silicon Valley and China: Mistral, Vivatech, Softbank infrastructure. There's also a partly geopolitical shift right now toward scaling AI companies domestically rather than defaulting to a relocation, which is changing where the interesting work gets done. What makes Paris specifically interesting to us, though, is that the conversation here makes possible the inclusion of a different kind of person. Not just engineers and entrepreneurs, but storytellers, artists, philosophers, creators, people who shape cultural communities. We think the next era of tech gets defined by those people and Paris is one of the few places where that seems deeply part of the ethos.

References

What are you reading right now?

We are avid readers of anything Pope Leo publishes these days. We’re currently reading Lanark: A Life in Four Books by Alasdair Gray and rereading La Horde du Contrevent by Alain Damasio.

What are you listening to?

Amor Satyr, shortwave radio, Monolake, Seeing is Forgetting, 陳粒, BB Jacques, Paul St. Hilaire.

Who are you watching? Consider artists, writers, builders, or anyone whose work makes you want to make something.

We’re inspired by the games of Mitchell F Chan, Pippin Bar, David O’Reilly (really looking forward to his new game coming out this month). And we feel alive looking at the work of Yuma Burgess - it’s bizarre and beautiful. We’re lucky to be collaborating with him on custom USBs at our Paris event and excited to see what he makes next. We also feel drawn to S. Paul/classicselena whose work feels like it is both sprawling and unknown.

What's something you've seen recently (a show, a painting, a film, a piece of software) that opened up a way of thinking you didn't have before?

We started watching Paranoia Agent by Satoshi Kon recently, which is an incredible exploration of subjectivity. It kind of unfolds into a full diagnosis of how stories about ourselves break and re-form under collective pressure. It’s working out something we think the rest of culture is only catching up to now and doing so while experimenting with the medium.

Is there a book, text, or thinker that's load-bearing for Nephila?
We’ve referenced this short essay by Berardi titled “Malinche at the end of the World” a few times, which explores the historical figure of Malinche, who controversially served as translator and lover of Cortez when he conquered her own people. When we first started making projects together, we were thinking about the end of the world, about the process of “unworlding”, and how to create things when civilization feels like it’s collapsing. Perhaps a lot of what we do is inspired by the translations made by Malinche, by the act of forming a new language at the edge of collapsing worlds.

What's a reference people would be surprised sits underneath this?

The Star Gauge by Su Hui, which is a fourth-century Chinese poem written by a woman abandoned by her husband that can be read in any direction and generates thousands of poems from a single grid of characters.

What are you into right now that has nothing to do with the work?

Alice Sparkly Kat’s astrology readings and caterpillars.

What’s Next

You mentioned other projects in the works. What can you say about them?

We’re playing with the idea of going into hardware next and currently making a portable music device - part local file player, part file-sharing tool, part recorder, part stem player - designed around social interactions. We're trying to figure out what the MP3 player would have been if it had been built for sociality rather than solo listening. We’re also excited to work across different stacks, of returning to China, of working with collaborators in South-East Asia, and of continuing to let subcultural research drive what we build.