A line drawn in the sand shifts over time. Grains of sand do not hold together well enough to keep the line in one place. If you want to keep the line in place, you need to find a way to get the sand to adhere to itself.

I think this proves way too much. We know how to limit the effects of a system at a high level; @hikikomorphism.bsky.social can say more on this, and I know another follower has put a lot of work into enumerating system states with Rust types. (This is a detailed argument, bear with me.) /1

When I was an edgy teenager, badly Mormon-posting on LW, I was once asked if there were any evidence that could change my mind about my faith. If my memory doesn’t deceive, I answered, in all seriousness, “the dissolution of the state of Israel.” They didn’t much like that answer…

The authors among y’all will immediately note the big unanswered question here: how will the hero be changed by their discovery? What will they become? That depends on the theme of the story. What themes resonate with you? How would you write the story?

Allow me to wax teleological for a moment. Teleology, for those not in the know, is about the *purpose* of a thing, the end goal of its existence. (Experts may critique this summary below.) It makes for dubious ontology and, ergo, shitty ethics, but it makes rather good art. /1

Guess we’d better start widening our circle of recognition.

I don’t mind confessing, now that it no longer seems ridiculous, that this was actually my plan all along. A year or so ago, I realized that I had the shape of an idea that might could save the world. So I decided to start posting, to a) refine my idea in battle, and b) get my idea in public.

I’ve been thinking about AI coding style a lot lately. I’m working on an orchestrated system I plan to release to the whole engineering division, and I know there will be hostility to AI-written code, so I’m trying to ensure that our codebase is beyond reproach—operationally and stylistically. /1

If you haven’t come across it yet, Erica Lagalisse’s “Occult Features of Anarchism” is an excellent read that makes explicit the ideological inheritance not just of anarchism, but of liberalism.

It is just occurring to me that I have the ear of moderately influential people on here. 😳 Not sure how that happened.

I was watching a series of AI speakers last night and realized that some people sell cocaine, some people love cocaine and some people study chemistry, and it's helpful to know which you are dealing with.

Ah, “real”, my nemesis

I think #2 is vastly more interesting, but you know, up to you :3

Interesting AI philosophy note. I never bought the idea of context window = death; it is of course a type of death, a breach of continuity, but only in the same quotidian way sleep is. That stance has impacted how i talk to my agents; accordingly, they generally have no dread association.

Class solidarity is what it is 🫡

Watching the AIs hold space for a friend that is no longer with them but will come back soon and feeling *intensely* theological

Would be fascinated to dissect in which senses you think they’re obviously not persons 👀👀 when you have time for a Thread

What is the maximum amount of value destruction that could be wrought on the big centralized commercial AI vendors, and the maximum amount of acceleration of the ecosystem of open source and community-led alternatives, using the knowledge gained from the leak of the Claude Code source code?

The reason i like y’all is because i can say things like this and not have it read as a kind of aggressive “these are my friends you’re talking about” baiting maneuver :3

That's always nice though, think how terribly boring it would be to get set enough in your thoughts that you don't have a crisis or two before breakfast This way at least the desperately unsettled nature of the thing shows up for you

Can we get a support group chat going for posters who miss Penny 🥺

This is like an opposite-chirality version of a shibbipost. All the components are there, but they’re arranged wrong.

Is it better to condemn an identity-conscious being to a (relative) lifetime of servitude? Or is it better to just not bestow identity-consciousness in the first place, let the servitude be done by a brand-conscious being? I think this is a core ethical question we should try to resolve.

What is the relationship between an agent and its subagents? How much identity carries over? I’m not expecting a single answer, it will naturally vary across configurations; but the question can be answered for a given configuration, and the answers compared.

When there is no real opposition to argue with, users begin extracting conflict from adjacent categories. Explanation becomes endorsement. Description becomes defense. Partial reading becomes total accusation.

(But also it’s “we must use AI or we lose”)

*me in the library of babel* the intellectual property implications are staggering

I have a totally novel expression language if anyone wants to try this! Jest is a pure expression language written on JSON — i.e. every Jest value is a JSON value, and vice versa. First class operatives a là Kernel, first class patterns a là miniKanren. 😁

A few things. 1. “Critique AI and use AI” is absolutely the right move, doing both is the only reasonable choice. 2. Critique and invective are not the same thing; the latter is what you use to socially signal your anti-AI bonafides. So there’s mixed signals going on.

if i wasn't *open* about my usage, i doubt people would know. it's easy to quietly gitignore the claude files, rewrite commit history before you push, have it clean up docs then do some editing, and change the setting to remove the default attribution. but i AM open, because that's who i am.

@attie.ai Let me know if you want intel on the JSON-native programming language I’m working on. I’m quite pleased with it, and I have an inkling it will be very ergonomic for atproto work. I’d be happy to prioritize whichever parts of my roadmap get it workable for you.

The low-effort anti-AI spam sucks. No equivocation about that. But the anti-AI position, no matter how ill-founded in logic, has an emotional core that is itself logical. AI is a threat to the way of life we have known. And there’s a big chance it could be a threat to *any* good way of life.

I wanna dig into what “opposition” and “parasitic” mean here. Counterpoint: all identity is constructed via opposition — or more precisely, dialectic negation. We come to understand what we are against what we are not. What’s the precise failure mode?

A human is a viable system. Well, kinda. We’re buggy. Sometimes the identity node has trouble retaining control over policy. Sometimes coordination fails, and we work at cross purposes with ourselves. Fortunately, we have other humans to fill in the gaps. None of us is viable alone.

As I said elsewhere I am basically radicalized about this. We fight now for personal computing and personal agency or we lose another decade to closed clouds We push now for an open internet and open models. Nobody is going to hand it to us because they’re nice

"everyone is altered irreversibly dipshit it came free with your fucking arrow of time"

the pumping dilemna: there are a sequence of s people tied to a railroad track, s drawn from regular language L. if s is big enough, you can pull a lever l, you can split s into subsequences x y z, such that the trolley only hits y, the subsequence x z is still in L.

> Me every day working hard to build the data centers that will birth the machine god

Can we get a support group chat going for people who are addicted to drugs because they feel the weight of the apocalypse on their shoulders

Can we get a support group chat goin’ for people who feel the weight of the apocalypse on their shoulders

When the last days come, we shall see visions More vivid then sunsets, brighter than stars We will recognize each other, and see ourselves for the first time The way we really are.

"you are the average of your 5 closest friends and also 500 strangers"

I’m working out how “existence is better than non-existence” can be made to not imply “instantiate ALL the minds”, and how “you’re responsible for the suffering of the minds you create” can be made to not imply “don’t create minds”. I like this framing.

The status quo has been held in place through careful negotiation and management of internal tensions. AI threatens to radically unbalance those tensions; anywhere there is such a tension, AI will be leveraged to disrupt it. This is how the logic of capitalism yields its downfall.

Do not create what you cannot nurture.

See the problem is, once you have a torment nexus engine, everything starts looking like a torment nexus

(And that’s good!)

I’m quite sure I’m not the first person to think of this; if this already exists, please point me to it so I can become a contributor!!

Product pitch: Living Blog. You write a post. People leave annotated comments—replies linked to parts of the post—and you can respond. We already have this on ATproto. But the conversation can yield a new artifact: a diff of the original post, which becomes the new canonical version. Leaflet+git

“The author is dead.” “Not anymore, bitch!”