https://bsky.app/profile/pfeif.garden/post/3mirenqi2ac2w Some folks on bluesky are circling around a new metaphor that I hope is in the family of Feeling of Computing. Would love for y'all to add input here. This idea is pretty exciting to me. I'll have more longform thoughts on what I'm building and where it fits soon, but figuring out this metaphor together is an important way to start. A collective mental model is the first step towards collective action. If you want to participate in the Ecological Computer, we're on our way.

This looks similar to the goals of Malleable Systems or Substrates. There's also some overlap with older ideas such as Postmodern programming. I have seen ecological metaphors in these spaces, though not as the main focus. As so often, I see the main challenge for this in growing suitable social structures that can sustain such socio-technical ecosystems. The second challenge is fighting against the "tech habits" that everyone involved in these communities grew up with.

I am interested in this. Lopecode is built on a microkernel architecture exactly so that modules can be added an removed ad hoc (even at runtime) and nothing really needs to understand the whole. There is no overarching UI and what you see is generated by userspace. This is an ecological architecture I think?! I am also interested in weaving this into atproto, I think modules would be small enough to fit in as blobs, then an app is a collection of modules. But I don;t really know where to start and I don't want to run infra. I started plating with atproto recently but very basic ATM (read only). I think I want to try an publish modules though soon.

@Konrad Hinsen Yes very much so. I spent a lot of time reading the malleable systems forum a few months ago. Their critique of I&S malleable software was super important in shaping how I started attacking the problem of taking this idea and writing it on open source code on open source hardware. The system is already malleable, if you are an incredibly talented programmer. The secret to malleability is building out in the open, on open standards, for open hardware. That’s how you encourage true diversity of thought. It has to be accessible to everyone. I&S developed their ideas on macbooks. What would it be like if I tried to make Patchwork on fully open source, friendly license software. Postmodern programming is new to me! Thanks! Excited to dig in.

interesting ideas. I think AI will do the technical stuff in the future, so future mallability is about expressing what you want easily, not implementing it.

Agreed. If you want to create a shared context between user, developer, and agent, there has to be something all three can understand in the middle. The middle is natural language. That’s the next programming language we're bootstrapping, I think. But in order for the AI to share in your full context, that plain english layer needs to be the source of truth, for both the operator and agent.

I think saying english is limiting, what about napkin diagrams

I’m close to having more to share!

I got my AI writing SVG interfaces!

Hell ya! Very much in my wheelhouse.

@Tom Larkworthy This is an ecological architecture I think?!

That's the opening question to a deep philosophical inquiry. Do ecosystems have an architecture? Can anything architectured be called an ecosystem? My answer so far is "no", but that's a weakly held opinion. What I see as the essence of an ecosystem is structure emerging from diversity, and relying on that diversity for its stability. In a natural ecosystem, organisms of very different nature cooperate, creating structure and thereby affordances for everyone. In a healthy ecosystem, no single species is critical for its existence. There is a common foundation for living ecosystems (organic chemistry, up to nucleic acids and amino acids), but that foundation is not designed to be a foundation of ecosystems. It has emerged as a foundation because ecosystems developed out of it. The digital equivalent of organic chemistry is the technological foundation that nearly all information technology builds on. That's bits, bytes, UTF-8 text, and perhaps TCP/IP. Anything more complex sitting on top of these building blocks is already in competition and cooperation with other technologies, i.e. operating at the ecosystem level. If you see Lopecode, or ATProtocol, as a new foundation for future ecosystems, you are envisaging a far-away future in which your foundational technology has displaced all of its competitors, with evolution acting at a higher level in the IT stack. That may or may not happen, but it's hard to design for it, and personally I don't see it as a worthwhile goal. If you want to do ecological computing, starting today, the main question is how to promote cooperation over competition among existing technologies that sit above today's universal foundation. It's not writing a new microkernel, but decomposing the kernels existing software systems into microkernels that can be made to talk to each other. It's not designing a new protocol for everyone, but carving a niche in communication space for your protocol that lets it interoperate with other niche protocols. The most ecologically inspired tech I have seen so far in the ATmosphere is Bridgy Fed. Going one step further, an important aspect of ecosystems and complex organisms is an immune system. Its role is defense against outside takeover of resources, but also defense against rogue actors inside the ecosystem that want increase their share to the point of endangering diversity. If you dream of an information ecosystem in which everyone migrates to Lopecode, or to ATProto, then you are a rogue actor, the equivalent of cancer in an organism. Today's information ecosystems probably don't deserve this label precisely because they have no immune system. Takeover happens all the time, and takeover attempts ("rewrite it in Rust") are dominantly seen as positive ambition, rather than as a threat. Changing such attitudes is the real challenge on the road to ecological computing.

Do ecosystems have an architecture?

yes, DNA! thats the microkernel in nature. And they steal and repurpose genes in a messy ad hoc way by a variety of mechanisms without grand design, and they use genes to build the protein machine that reads the DNA, that's usespace systems programming, lol

but decomposing the kernels existing software systems into microkernels that can be made to talk to each other.

Lopecode is very much JS/npm compatible, its using existing HTML 5 module import machinery. Its not a new language, its a container technology, which helps an existing system (JS) move around easier (single file representation).

If you dream of an information ecosystem in which everyone migrates to Lopecode, or to ATProto,

The only thing I want to displace is SaaS and centralised network dependancies. I see AtProto as a distributed store system that could be another vector of transmission, but I would not bet exclusively on it, I think drop box probably has more staying power and Lopecode is comptaible with that.

the equivalent of cancer in an organism.

Funny you should say that. I was thinking that the chance of a species something surviving is related to spread factor * longevity. So I was thinking about viral transmission with the single file format. It maximizes transmission, and by bundling all dependancies, it works on a longer time scale, so it also maximizes longevity. Literally virus were on my mind, which is very ecological, but you have phrased it in a bad way by using cancer instead of virus as your metaphor. I think viruses (lets ignore cancer) are not necissarily bad, they move DNA across species and help with diversity. Some poison you, yes, but they also help evolution on a longer timescale [1]. I think explosive growth (this is cancer) in theory is harmful in a resource constrained environment. But this is software and the web cannot do many things so its not gonna drown out anything. It will help with software diversity because its end user moldable, thats one of the major design features. [1] https://pmc.ncbi.nlm.nih.gov/articles/PMC6850104/

@Tom Larkworthy We have different definitions of "architecture", which is fine of course. Just to keep in mind. But I don't see DNA as the equivalent of a microkernel. It's more like a file system, meaning a low-level component in a multi-layer structure. I agree with your description of Lopecode as a container technology, but it's a container which depends on a non-trivial microkernel, which is a bit unusual for containers. Lots of information processing technology today does not use that microkernel, nor anything equivalent in function or purpose. So for me, that microkernel is one out of many organisms in an ecosystem.

the chance of a species something surviving is related to spread factor * longevity.

That's a question that goes beyond my superficial knowledge of ecology, but I am pretty sure that the main factor in species survival is adaptation to its environment, on which it depends for food and a lot more. Next comes competition from other species.

Literally virus were on my mind, which is very ecological, but you have phrased it in a bad way by using cancer instead of virus as your metaphor.

Viruses are indeed part of ecosystems. Unlike cancer, they normally don't kill their host as part of their reproduction cycle, and at the ecosystem scale they do have a role to play. Which is why I used "cancer" deliberately. It kills its host and thus itself, without any means of reproduction outside of the host. Which means that cancer is just destructive. On a less dramatic scale, any strong reduction of biodiversity ends up killing an ecosystem. Anything that has unlimited growth as its objective is harmful in the long run.

explosive growth (this is cancer) in theory is harmful in a resource constrained environment.

All environments are resource constrained. If you consider software as "living" only when it's used, then it requires physical resources (operating hardware) to survive. Which, on a finite planet in a finite universe, are limited as well.

It will help with software diversity because its end user moldable,

Sure, that's something I definitely agree with! I do see Lopecode users as a good citizen of ecosystems, under the condition that they don't require nor aim for universal adoption. For Lopecode specifically, I don't see this as a real risk, given the huge diversity in programming systems we have today. Unless Apple declares it the one and only foundation for future iOS apps of course 😉 But many tech communities do have a tacit attitude of aiming for unbounded growth, at the expense of competitors.

yeah ok, the https://en.wikipedia.org/wiki/Replisome is the microkernal, DNA is the digital storage, but part of that DNA encodes the Replisome, that loop is the microkernal, and then everything else hangs off that in DNA userspace e.g. an eyeball plan. Saying DNA is jsut storage is missing the forest for the trees. Every microbe on our planet is DNA based and has the same DNA replication infra running its cellular lifecycle (sans some viruses which are sometimes not considered lifeforms). I think container formats are often more complex than the thing they contain, HTML is simple, but browser layout engine, hard. Hello world C program, simple in bytes, OS/microcomputer to run it, hard, the C compiler itself also hard in comparison to the program.

survival is adaptation to its environment

thats what longevity is, but survival without replication is still extinction, but it takes longer. ... I found the actual formula I was thinking of: net reproduction rate I just flattened it over age. Real science formula!

All environments are resource constrained. If you consider software as "living" only when it's used, then it requires physical resources (operating hardware) to survive.

No I don't think digital information has to be run to survive. as long as it makes it to a library and run 50 years later, its still exists. Yes it has to be run at some point, but this is why local is different to SaaS and a domain registry entry. Those thing decay and have a shelf-life. Pure information that is self-serializing does not really need to be run on an ongoing basis. It can be dormant like a tardigrade, indefinitely, and thaw when someone double clicks. sidenote: the ecological limit is the carrying capacity, I think digital storage is essentially infinite at this point, especial documents measured in Mb.

For Lopecode specifically, I don't see this as a real risk

Nor I! Lol. But I enjoy this ecological framing, its a good model

Reading and absorbing. Good discussion :) Re: Evolution and growth and not killing the host That's why ink and switch began working on cambria. The building blocks need to be able to evolve. Cambria allows genetically different "plantings" to still be allowed to combine and reproduce. So long as the schemas do not evolve so radically that the structure is unrecognizable https://www.inkandswitch.com/cambria/

@Jacob Pfeifer Your outline sounds to me as if you are doing exactly what Unix did in the 1980s and what Web standards did in the 2000s: define a substrate that you hope to become a foundation for a software ecosystem. There's nothing wrong with that, of course, but it's not what I expect when I see the label "ecological computing". What I'd expect from that label is a serious attempt to avoid the degeneration of the ecosystem dreams of the past into corporate-controlled silos. Via some form of immune system.

the web. The one network that’s already decentralized.

It is in theory. It is not in practice. Just like ATProto as of today, which is dominated by a single company.

Do I understand correctly that your analogy is Lopecode as a species, individual documents as organisms? The ecosystem then includes everything the documents interact with.

Yeah! This is right sorta. I actually think of the real ecology as idea space - in the meme sense.

Did you think about interaction with data analysis code in Python or R, for example?

This is why I don't worry about ecosystem takeover, because the web has some massive limitations that do not apply to python, so it could never displace python. Data interoperability is needed there (i.e. files as the lowest-common-denominator, networking as well)

The ecosystem then includes everything the documents interact with. The interaction with JavaScript libraries and Web APIs looks obvious, but what else?

OK, so I like Dawkin's meme theory that the real humanity is exchange of ideas that have their own lifecycle. Thats the central ecology to humanity, the idea ecology. A document, book, webpage, they are conduits for human brain idea exchange (media). Thats the important thing. I think blog posts and books are cool, but would be better (for a subset of possible ideas), is computational blogs that include a computation model in them. So the real interop is not technological, its being able to express a broader set of ideas in a document, give that to a human (across time), and its somewhat interactive and they can play with it, look at the code, copy the algorithm, and gain more understanding than if they read prose on the same thing. I think two things don't need to have a binary interface to collaborate. Its like this forum. We chat about project, rarely does anyone actually use or contribute code, but we still collaborate, because we talk and think together. So even if an idea in a lopecode document cannot call a python function, doesn't mean the people that read something in a Lopecode can't be inspired to copy an algorithm into python. So then I design Lopecode to maximize the persistence and ease of transmission of (computational) ideas. But it would be crazy to think there is a winner-takes-all dynamics to human media choice, we have way too many sensors and individual preferences. So that probably clarifies nothing. When I say "virus" I am actually thinking "meme", thats my actual model, although I have thought about it as much as I have now.

I agree that it's the evolution in meme space that matters most. That's what human culture is all about. But embedding computational models in meme space remains an unsolved problem. It worked in the 1960s, when scientists published and reviewed their code just like papers. But those were 200-line programs. Today we have huge and messy software stacks, in which computational models are often still simple but completely drowned in implementation details. If you see some interesting work and think "nice, I'll try this on my problem", you will soon find out that it's impossible to figure out how it works in detail. There's only a fuzzy summary that makes it into meme space. I believe this is a solvable problem to some degree (see my work on Digital Scientific Notations), but I can still count the number of people who agree with me on the fingers of a hand. In the meantime, we build ever bigger computational models, share them as unreadable code, and create subcommunities of knowledge workers defined by programming languages rather than by domain expertise. Lots of tiny ecosystems that talk to each other at the fuzzy summary level only.

It is in theory. It is not in practice. Just like ATProto as of today, which is dominated by a single company.

I agree we aren’t there yet, but the Atmosphere is already thinking about hard decentralization. Bluesky is too large of a player right now. But there is lots of energy in the atmosphere for something new. I think it’s important to try.

What I see as the essence of an ecosystem is structure emerging from diversity, and relying on that diversity for its stability. In a natural ecosystem, organisms of very different nature cooperate, creating structure and thereby affordances for everyone. In a healthy ecosystem, no single species is critical for its existence. There is a common foundation for living ecosystems (organic chemistry, up to nucleic acids and amino acids), but that foundation is not designed to be a foundation of ecosystems. It has emerged as a foundation because ecosystems developed out of it.

https://atmosphereconf.org/event/rjQ96kl

hello 👋 there, guy steele also had this idea 💡 about notation : # Guy steele on Computer science metanotation (CSM) • #tags: #youtube #talk #guy_steele • • language of computer science papers, analyze scope over 30 to 50 years • try to build a interpreter/compiler for it

@Jacob Pfeifer Yes, it is important to try! And the video you pointed suggests that people in the Atmosphere are aware of the challenge, which is a good start. But this also applies to your "ecological computing" community. Your "engineers who write infrastructure in Rust" must work within a governance structure that also includes the "tool makers who write Web components". If you let the developers alone decide on infrastructure evolution, as it has happened in the past with pretty much all tech, you end up with a decision making structure that can easily be captured by corporate interests. Let's say Microsoft wants to take over your community: all it has to do is join the project and inject sufficient developers on its payroll to dominate all decisions.

@Konrad Hinsen If so far the two success stories of ecological computing are unix and web standards, we should be trying to build protocol bridge *between them that allows for data privacy. *on top of them? *alongside them?

The microkernal just deals with hot code reload and reactive plumbing and scheduling. The cool things like self-serialization and multi-notebook UI are in userspace. Mostly hanging off a user-space library "runtime-sdk" which adds various meta-programming hooks like "callback for when code was edited", "call back for a variable value changed". If you want to generate a history of code edits, you listen to that stream, which is what "local-change-history" does. local-change-history stores edits to IndexDB, and replays it if the page is opened, giving you persistent edits across page refreshes. I looked at Automerge for that part, but choose isomoprhic-git instead, because git has deduping across similar documents across branches, and it is a smaller library and I am not tackling sync collaberation yet (the docs diff nicely async in plain text for now in git). I would not want to represent Lopecode on file as a CRDT because it would not be a plain text readable file, but if we wanted realtime sync, then I think it would look similar to local change history, but instead of a discrete 2 phase "replay prior history on open, then record history going forward" it would be more interleaved. The building blocks are there though, each cell has a persistent id which is preserved across serialization.

@Jacob Pfeifer I think text chat is very low resolution, it's sloppy in many many ways. It changes how we communicate. you enter some tokens and think that i or another person responds in a certain manner. There are already too many not good assumptions here. Do you know this list of alan kays work? https://tinlizzie.org/IA/index.php/Talks_by_Alan_Kay I have seen some of them. You mentioned that you know McLuhan or some principles of his work/thinking process ? how would you say does it inform you here - right now ? I hinted at better communication. And mentioned the glass bead game. Imagine you go into a room and meet a handful of strangers. there are some posters on the wall. That is this thread. 🧵 Think about it. then maybe feel about it. How much of what came here was through responsive impulse? How do you feel reading this? Relaxed and calm or does your mind feel agitated? can you feel the difference? Here is another frame how we could try to better understand what we do here: https://docs.google.com/document/d/1B0LShb1sRxNUnzHDlYdnZWC2niYRw_s6Hg46mFagndc/mobilebasic

@Jacob Pfeifer > Human cooperation is the immune system. Collective action is. Exactly! > But if the base item is an ATProto lexicon json object wrapped in an automerge CRDT, we get... a lot of extra functionality. Also a heap of dependencies. Can you make this modular? A plain version with minimal functionality and dependencies. More to be added piecewise as required. And have several variants. CRDTs are for real-time collaboration. In many situations, you want asynchronous collaboration instead, e.g. via a shared git repository. More general advice: map out the design space before looking at specific technologies for implementation. What you want to do is more important than how you want to implement it. > 2 Minutes into Guy Steele and I’m already in love. His talks are great. While you are at it, check out "Growing a language", my favorie Guy Steele talk.

@Konrad Hinsen > Also a heap of dependencies. > What are dependencies but the real life consequences of collective action? 😉 > Can you make this modular? A plain version with minimal functionality and dependencies. More to be added piecewise as required. And have several variants. CRDTs are for real-time collaboration. In many situations, you want asynchronous collaboration instead, e.g. via a shared git repository. > I agree we want real-time and async! Patchwork does both and allows for offline editing. They have a [functioning prototype](https://patchwork.inkandswitch.com/) you can play with. P.S. I don't know if that has been publicly released yet I just got the url from one of their public demos. Their team is using it internally though so I trust that they are working on the sync problems. This is why we all need to talk to each other. I'm more focused on the kinds of interactions that tech enables and how to bring it to a wider audience. > More general advice: map out the design space before looking at specific technologies > This is where I was before ecological computing jumped out at me! I'm was in the middle of an MVP prototype. It's a mess! Not ready to share with anyone. But I know it's the next step. I did start with the design space :) Only way I know how to work. Totally agree though. We need to work backwards from behavior we want to enable. But I don't have the language to be able to talk about those problems abstractly, I can only point to existing tech and how i think it solves the problem I've been thinking about. Just pointing out my limitations. Super appreciate you engaging with this. It's very very valuable to me ❤️

@Jacob Pfeifer

What are dependencies but the real life consequences of collective action?

You mean collective action without common governance? Because what dependencies are, in practice, is the possibility for someone else to break your system, be it intentionally, by neglect, or by incompetence. Common governance could offer some protection, but I doubt it would work for the huge dependency graphs we build up today.

Patchwork does both and allows for offline editing.

Thanks for the pointer, I need to play with that!

On the topic of governance I found this clip (28 min) by Alan Watts to be very insightful https://archive.org/details/AlanWatts-LawAndOrder-Kqed-1959

@curious_reader I am finding it challenging engaging with your ideas right now because I’m still in the middle of trying to figure this one out. But I appreciate you pushing me to think about the implications and how we can improve on text based communication. I agree, writing is limited, but it also served humans well for a long time. It’s a powerful way to communicate ideas. I agree we should try to be as thoughtful as we can about future implications. We have to try not to make it worse. We are partially responsible for the tools we bring into the world. But I think there is danger in trying to solve too many problems at once. The atmosphere is trying to solve a dire problem we face as a community. The internet media landscape is so twisted by the attention economy. By surveillance capitalism. We need ideas that work right now. The decisions we all made collectively 20, 40, 60, years ago created the problems we face today. We’re going to cause problems for future generations. But I think we have to try. And the atmosphere is trying. I want to help try too.

@curious_reader Looks like a good complement to David Graeber's book "The Utopia of Rules", which also points out that any attempt to codify "good" behavior into rules is bound to fail.

very good advice Konrad! it reminds me of rich Hickey‘s Hammock Driven Development: First step of programming:

I’m doing that too 🙂 Just spent the weekend outside coaching youth sports and not thinking about this project at all.

who gets to make digital things

More precisely: who among those who can afford extended use of a coding agent. Especially if/when they will be priced at their real cost. That's still a lot more than people who can write code today. Unfortunately, among those newly coding-enabled people, many will use their new power in bad ways. So yes, we will likely see interesting innovation in UI design, as Andy expects. Is that worth the downsides? Well, we don't get to choose.

Do you think it’s possible open source AI catches up, or at least gets close enough to the frontier for people to make a choice? Because if not, then I agree, we don’t have a choice.

Yes AI is continuing the break neck pace of technological acceleration.

I’m not sure I’ve followed the thought all the way to the bottom like you have, but we share the same concerns. I still hold out hope the american people will wake up and we’ll start to fix some of this ourselves. democracy is not yet totally lost.

I watched it! I thought it was a fascinating window into how stuck we are.

There are serious implications with the problems of capitalism. Like fundamental ontological mistakes.

I can get there. I haven’t yet, but I could get there. I would want to know what you propose comes next. Or what was better that came before.

One could try to do something with laws but that doesn't work really well see the Alan Watts piece for example

No no, I’m watching it now!

Its ok the video is just a preface. So one of the biggest problems of capitalism are for example negative externalities, because those are mostly not priced things can exist in the system that are very unhealthy for earth , animals, humans. PFAS is a good example of that. Now you could try to regulate that. But the question is how such technology like PFAS came into existence, sureley the scientists must have done some checks before applying something that pollutes the global planet for multiple thousands of years right? Thy didn't, same with fertilizer and pesticides. Without those we could not have the demographics development we had. But it kills the soil. So we haven't found a sustainable way for agriculture, this is not good , even worse corporations try with patents and other negative behaviour to create food shortages etc.. So negative externalities are a problem, but why? Because I would say in capitalism the "mainstream" narrative is that the earth, soil , its just dirt, so who cares what happens to dirt? Well its not dirt, soil is the reason why we can exist on this planet. I could now engage in a long story how fungi create soil from the rocky earth then eventually plants and animals appeared on earth. But thats just an explanation. We have way too much information, and way to little affordances for experiences. We have in fact so much information that it overwhelms us completely, it paralyses us ( see the quote from kei above I shared, she is one of the tech philosophers). And that was before the recent AI stuff. So how can we treat earth like that? How can we treat other humans like that. I think it is because we forgot what is really important and indeed again what is right and what is wrong. Most of modern life or lets say culture is shaped and heavy heavy influenced by media. Chomsky showed in the 90s already how the media can be abused. Elon bought x for a very good reason. And he is creating his own wikipedia for a very good reason. At the heart of capitalism there are sitting some very very strange contradictions. Like for example how can you Kill your own biosphere? Because you think its dead , that it is not living. As a materialist What do you believe in? Do you believe that you will turn into dirt after you die? And that is why you HAVE to become immortal in this life ( hello dont die from bryan johnson, although there are some good things from him) . Materialism is false, what are the implications? Earth is sacred and so should be also human relationships. Bernado Kastroup argues for that way better then myself ( you can also see different interestings links to other people like Michal Levin there)

@Jacob Pfeifer

Do you think it’s possible open source AI catches up

That depends almost exclusively on the economic context. In the current one, as @curious_reader has laid out, the answer is no. What would sustainable, egalitarian, responsible AI look like? First of all, slow down. Put a lot more effort into studying the impact of AI than on advancing AI. Next, work on lowering energy consumption before increasing model size. Running LLMs on GPUs is very probably a very inefficient choice - that just happens to be the hardware we have. Also, put all AI development under democratic oversight, rather than under the control of venture capitalists.

I also doubt that these things can be forced. But i hope you can see why i was talking about the butlerian jihad in the other thread. The future still (ha) is. very much in flux. But science fiction may be able to give is some clues , at least to decide about things we want to avoid.

I am a wu wei fan as well, but I think we are a tiny majority of today's world population!

my position was this was all foretold by futurologists in 2000s, and it follows a curve that goes back tens of thousands of years of human progression. The curve is not gonna flatten now, it is our nature. However, that progression seems somewhat independent to how we distribute wealth, so there is a gap for hope there. I think things like energy efficiency is somewhat built in to the incentive structure. Its not like the data centers want to be generating so much heat, which contrasts a bit to crypto where pointless work was core to the mechanism. I don't like that I now have to give $100 a month to a tech giant to now do my job, on the other hand, it delivers more than $100 of value in terms of my production increase, so I take that bargain. At least there are multiple vendors ATM and an open source movement, so things could be worse.

re- wuwei - More specifically to our context what and if someone should do something about societal collapse. A a nice passage discussing lao zi and confucius can be found in the german forword to the IChing of Richard Wilhelm.

@Tom Larkworthy The obvious issue with exponential curves is that they cannot go on forever in a finite universe. To make it worse, our resource universe is just one planet plus the sunlight that hits it. I still have some hope that our species will recognize this basic physics fact before being hit by it in the form of a massive crash.

I have some fun anecdotal evidence that the butlerian jihad has already begun. The bands in a punk showcase I attended at sxsw this year all had hand drawn logos

What I was wondering about is the UX. You say "my reader". Is that an SPA? A browser extension? Something else? It's probably more complicated than "just open this HTML file" with "traditional" Lopecode. What I'd love to see is a browser extension that automagically opens at:// URLs, using some dispatcher based on lexicons. For now, access to ATProto data is always via some Web app, meaning someone else's computer.

that's hierarchical thinking! The code required to decode, assemble and render a lopecode document from an

at://

address is also a Lopecode document. So that means you can run it as a single static file from disk or from static hosting. AND you can upload the writer and reader to

at://

itself. Try https://tomlarkworthy.github.io/lopecode/notebooks/atproto.html the at:// transcoder notebook is here

<at://did>:plc:j7nm3lrd5h7fm3sfhcv3lhfv/dev.lopecode.bundle/3mkjwz5coof2w

note after rendering it you can "download" it back as a single static file again, if you prefer to run stuff locally and not with a network dependancy on the internet. Not the recursion in the screenshot, it loaded a functional copy of itself inside the "at-read" notebook. The inner one could also be used to upload a copy. I have not turned on an iframe sandbox as its harder to debug in that state so don't consider this anything beyond an idea.

Thanks @Tom Larkworthy! So the answer to my question is "An SPA, which happens to be a Lopecode notebook that you load in HTML format". My take on this is not "hierarchical thinking" but "bootstrapping". I have a browser that doesn't know about ATProto. And a notebook that's stored on a PDS. Something must connect the two. Your implementation of an AT loader is technically very elegant, but weird from a UX point of view, at least given today's UX expectations. Are there AT lexicons for code already? I see lots of possibilities there...

yeah, so the PDA store blobs WITH MIME TYPE. so you can store code files pretty easy as

application/javascript

(or css/html or text/x-emacs-lisp) rather idiomatically before you even hit the lexicon stuff. https://tomlarkworthy.github.io/lopecode/notebooks/atproto.html is the bootstrapping solution, its the same code as

<at://did>:plc:j7nm3lrd5h7fm3sfhcv3lhfv/dev.lopecode.bundle/3mkjwz5coof2w

is "https://tomlarkworthy.github.io/lopecode/notebooks/atproto.html " an SPA when it is also ``at://didplcj7nm3lrd5h7fm3sfhcv3lhfv/dev.lopecode.bundle/3mkjwz5coof2w`` ? Yes its an SPA, but its also on the atmosphere. It is two manifestations of the same thing. Thats heirarchical thinking, when you try to put something in one box or another in a mutually exclusive way. Both places contain the same code base. It is not exclusively an SPA, but it can be an SPA if presented that way. "but weird from a UX point of view" this is a prototype, its not been refined at all, its just a conversation about a technical possability, not a final thing.

Very interesting. It somehow reminded me of this Alan Kay talk, where he mentioned the following concept:

yeah, I think there are parallels to those things. We shall see I guess. Reactivity and self-serialization is a different emphasis though.

@curious_reader Peek looks interesting as a design study. But could anything like that actually become the standard interface to the Web? Would "normal" users be able and willing to manage the flexibility of peek? (originally posted in the wrong thread, sorry!)

I think actually the consumer internet , browsers where a bit of an accident, that was then exploited to push for consumerism. Additionally I also think that there could be quite different „Internet“ experiences.