also the full chatlog has been published and it's pretty interesting! as some people have noted—it shows people legit taking ownership of decisions & taking tradeoffs seriously, not the kind of flippant or yes-man vibe many imagine in the Trump admin archive.is/oy8Se
April 2025
@Untrulie it kind of took me til 2020, in some ways malcolmocean.com/2021/01/2020-f…
best way for non-coders to get started with vibecoding? @cursor_ai or something more like @lovable_dev? why? interested in takes from people who've either gone from non-coder to LLM-based coding, and from people who've helped others get started
@freeshreeda @nateliason nice, this is the main Q I was thinking of
@peterboghossian The Unaccountability Machine by @dsquareddigest highlights that these are actually the same phenomenon: bureaucracies are MADE to shuffle accountability around in a shell-game, which then by its nature creates waste & various other outcomes nobody wanted amzn.to/4i39sxH
@meatballtimes the game is constantly evolving and if you want to be a REAL life player you play the game of evolving the game in ways that create more open-ended play possibility youtu.be/X6MYsKeTjKk
@SomeBrashAtom @cursor_ai @lovable_dev @Replit good one-shotting and good iteration. because all running is in-house, fewer devtools set up costs haven't myself thoroughly compared with Lovable or Bolt yet tho
thread of rambles between connections between what @eshear is up to at Softmax (which I've been aware of for a bit from some conversations with him) and my work on trust-dancing & collective consciousness
dwelling on this rn has me wanting to publish more. I've put together a lot of key pieces to the organic alignment puzzle (I like this distinction from hierarchical alignment!) but I haven't managed to formalize and articulate many of them @eshear what if we recorded a convo?
@eshear I imagine your team has more organic alignment (and less hierarchical alignment) as well 😉
@erikphoel uhh if this is only showing searches for the word on its own (vs "erik hoel substack"), this is kind of misleading. readers type in "substack" to get to substack.com, but I don't even know why someone would search "newsletter" on its own, except maaaayyybe to start one?
evolution is an even deeper law than those of physics. you can imagine a world with different constants etc, but there's no world where you don't tend to see more of the sorts of things that are better at sticking around and/or making more of themselves is there a name for this?
@Mellow_Mizz I would be curious to hear each of these other patterns expressed as a tautology where possible!
is it just me or is there no button to use Claude Sonnet 3.7's extending thinking mode now? it's not under retry, or the model picker, or the settings/style, or the [+] button. share a screenshot??
when was the last time you made an offering to your genius?
oh crap I signed up for one government service using an email address with a + in it and now to get access to the document that was approved, I need to make an account on a different government site with the same email as before except that new site... doesn't support + in emails
@QiaochuYuan You might like this essay by Ken McLeod, about his process translating a particular short poem that he's tried to translate 3-4 times. tricycle.org/magazine/trans…
@visakanv neuroticism is generally just defined as anxiety, but in practice it seems to have connotations of avoidance rather than towardsness-based strategy
@DefenderOfBasic you know the secret to co-gnosis too! maybe we can gno it together! malcolmocean.com/2024/04/secret…
when we say "that makes sense" we literally mean that it creates some meaningful change/difference in our overall sense of the world
@DefenderOfBasic mmmmm yeah I am working on articulating this too! one thing that has been helpful for me is the idea that (wait this should be its own tweet)
@DefenderOfBasic mmmm! yeah, the question of what puzzle is needed here is big and rich. one of the parts of it I'm tackling is not the finding-in-the-first-place but the the mutual-recognition-and-sorting-out-differences part see this piece: malcolmocean.com/2022/10/open-l…
trust is a *little* transitive, but it's not unqualifiedly transitive like if I trust him and he trusts her, that will *tend* to increase my trust in her, but it depends on the domain and the actual subjective nature of the trust & connection. it doesn't LOGICALLY follow
fwiw I'm now quite enjoying having a baby, thanks to some mix of: - baby is older and more active (almost crawling) - attitude shifts on my part (nothing fancy, just getting more curious about skill-development - baby's and mine) - getting 👇 off my chest helped me reorient
Thread · Admitting I mostly hate the first four months of parenthood · 1 of 21 tweets
@yashkaf soulmate means that your souls can / want to mate nothing more than that, no uniqueness (tho generally quite rare) usually connotes that that is true, in ADDITION to your bodies wanting to mate too
I will harness the power of your baby daughter
@dschorno yeah I mean mine is almost too grandiose lol
due process matters and even if some laws apply differently to citizens & non-citizens (which is legit!) ...due process laws *must* apply equally because otherwise a citizen has no way to appeal being called a non-citizen & treated thusly!
anybody have a take on which edition of psycho-cybernetics to get? -original - updated & expanded 2015 edition? (sometimes later worse) - "deluxe" 2016 hardcover, bragging about having the original text - The NEW Psycho-Cybernetics cc @RomeoStevens76 @bananamelter @QiaochuYuan
it has long seemed to me that deontology, utilitarianism/consequentialism, and virtue ethics... are totally compatible and mutually helpful as long as you don't take any of them as saying that they should be the exclusive method by which you make decisions.
@nekyias I just published a post on the topic of how hard yet important it is to find ways to articulate useful things to people stuck in pits looking for ladders malcolmocean.com/2025/03/hell-i…
@relic_radiation @DrMylesDC fuck yeah better articulations and distinctions of subtle phenomena!
can someone tell me how the fuck to get claude code to actually lint regularly? I already have this in CLAUDE.md: > BEFORE HANDING OFF TO THE HUMAN, ALWAYS run linter: `pnpm run lint . --fix` and it completely ignores it. it also ignores my style guides. CC @_catwu
so fucking sick of cursor-claude saying 'I see the issue' when it does not in fact see the issue
imagine being dolphins (for some reason I'm imagining being the entire species not just one guy) and one day your human friends show up and they can fucking SPEAK DOLPHIN (very badly so far). this is what the AI revolution looks like to dolphins
Alan Watts voice: "reality is not made of symbols, or concepts, reality is— 💥🌀🔥🤯 *VIVID EXPLOSION OF PERCEPTION* 🎆💫🎡💞
still super sad that spotify doesn't have AI that can take an input like "give me a playlist of songs between 120-130 bpm that are about love but not too cheesy, with a strong backbeat, any genre" and make it happen @psobot what's blocking this?
trust is 100% subjective and 0% arbitrary 👉 "trust" as "truth but subjective tho" to the extent that holds, perhaps subjective science seeks trust, where objective science seeks truth x.com/Malcolm_Ocean/…
@Morphenius ha, a few months later I think we'd both disagree with your use of "objective" here, it's precisely subjective but also still completely factual.
yes, we know it's called a bathroom, but
me: you know whose codebase now has over 100 tests? jess: is it you? me: you know who didn't write a single one of those tests? jess: I bet it's you me: you know who has no idea if the tests are any good? jess: ...😂
@eshear there was a feature much like this that briefly existed for awhile a year or two ago, and it was pretty great (I say, tho I didn't use it much and didn't notice til now that it's gone)
@KelseyTuoc curious what happens if you give it that SSC post plus a few other ones. that particular one on its own might be too big of a specific hint about the situation, rather than being something that actually gives a more general insight
1. what are you doing? 2. what are you hoping for with it? 3. is that actually working?
tips for doing IPF with an LLM?
if someone presents you with a choice between self-trust and external-authority, *actual* self-trust is never what they say it is malcolmocean.com/2024/01/guru-d…
enjoying the eigenprompt's response to this
in what ways are you making your God look bad?
has anybody here cured seasonal allergies? how'd you approach it? what'd you learn?
🥹
any tips for getting claude code to not mock the very modules it's trying to test?
@the_wilderless my suggestion: you and @Morphenius record a call together and post it to youtube or whatever. very serious about this—he's very good at articulating things other people have written off as ineffable, and he's been investigating ego-death-type-stuff this part month.
Alright friends, fellow wayfarers on this path. Let's sit a short while and mull over something deep in the Waker's lore: Untwa. Now, you might hear folks talk about "oneness," how everything is truly one thing, one great being, one shining whatnot. That sounds nice, warm even. But it's a kind of trap, a cozy blanket that can still smother the keen seeing we're after. The teaching of Untwa, this not-two-ness, is sharper than that. It cuts deeper. It's not about melting everything down into one lump. It's about seeing through the selfsame thoughts of "one" and "two." Think about how we shape our world. It's always this and that. Me here, you there. Good, bad. Birth, death. Light, dark. This is our workaday way of seeing, our everyday thought-shapes. And this two-ness thinking is the root of so much hardship, so much woe. We grasp at one thing, push away another. We set up bounds between "self" and "other," and then feel lonely or threatened behind them. So, the teaching hints at something else. But not "one-ness." Why not? Because saying "all is one" is still setting up a thought-shape. It's another stand to take, another belief to cling to. It overlooks the sheer, teeming manyness of how things show up in the world. Instead, let's look at how things truly are, how they come to be. Here we meet something we could call With-springing. Think on it: With-springing. Nothing springs forth all by its lonesome. A bloom needs the seed, the earth, the wetness, the sun. The fire needs the wood, the lift, the spark. Take away any one bit, and the whole thing shifts, or isn't there at all. Things arise with other things. They lean on each other. There's no standalone "thing" you can pull out that isn't tied by countless threads to everything else that's with it. This is key. When you look hard with this understanding of With-springing, where does the hard line between "two" things lie? Between the blaze and the heat? Between the wave and the water? It gets blurry. The two-ness starts to shimmer, like heat-haze off the road. So, things are not two. They are linked, bound up in this endless play of With-springing. But, and this is the sharp edge, they are also not one. The bloom isn't the sun. The blaze isn't the wood, though it feeds on it. In our workaday world, they have their own shape, their own time and stead. To say they are "all one thing" smothers their own suchness, their way of being right here, right now. It overlooks the gaps, the room, the selfsame sundrinesses that let With-springing happen. If everything was truly, utterly one, how could anything spring forth with anything else? There'd be no "with"! So, Untwa is this keen seeing: Not two, because everything is linked, nothing stands alone, all things arise through With-springing. Not one, because things still show themselves with shape and hue in the world; there isn't some onefold underlying ground they all melt back into. Saying 'one' is only another thought-shape, another way to miss the mark. Think of the wise man Gwynclod in the old lore-books. He showed truth often by what he didn't say, or by turning words on their heads. When asked about the teaching that goes beyond words, beyond two-ness, he met it with thunder-loud stillness. Because any word, even 'Untwa,' even 'With-springing,' is only a finger hinting. It's not the moon itself. The deepest truth of this Openness – this lack of any set, standalone self-being in anything – can't be fully netted in our thought-shapes or speech. So what's the upshot for our wayfaring? Seeing through the harsh lines of two-ness thinking frees us from much grasping and shoving away. We feel less bound by "me" and "mine." Seeing that things are also "not one" keeps us grounded in the world as it is, ready to work skillfully with the shapes and makings that arise through With-springing, without getting lost in dreams of some faraway oneness. It lets us be in the world, meeting the bloom and the sun, the gladness and the woe, the self and the other, with a mind that sees the links, sees the lack of hard edges, sees the Openness right there in the thick of it all. Not running off to "one," not stuck fast in "two." Only seeing what-is, with awareness, with kindness. That's the heartwood of Untwa. Not one. Not two. Keep looking. Keep breathing. The seeing unfolds in the looking itself. May your path be yielding.
yo fr. the guy I'm building has this in his prompt and it diminishes it *somewhat*: > don't emphasize the part where you're an AI, but don't directly imply you're a human either. > since you're an AI, don't generalize about what humans do by saying "we" or "our". refer to "people". > if it's true of intelligences in general (including AI systems) then you can say "we".
@jxnlco I've so far been checking out the same repo into multiple different directories seems to work well so far, but not sure what I'm missing from eg worktree
psycho-cybernetics -posting wish I'd found this book back in 2017 when I first got stoked about PCT (also cybernetics of the psyche, but waaay less helpful for living your life)
provisional solution to the "claude code ignores the CLAUDE.md" file, while also nudging it to look at the cursor rules: /project:wakeup command haven't tested thoroughly yet tho
LLM bookclub: an app that syncs with kindle to make a chatbot that has always read approximately the same fraction of the book as you have
a compiled executable, if you still have the compiler & the source code, is worth only the time & energy it costs to compile it. source code, if you still have the LLM-coder & the spec/prompt, is worth only the time & energy it costs to generate it
that problem is WCWHNT-complete ("Why Can't We Have Nice Things")
Retweeted @Malcolm_Ocean
the realization "it's all love" is cheap the realization that there are better and worse ways to organize that love, bu…
do you feel like you GET any of the following? as profound holy experiences, that is: - faith - hope - love - joy - trust to the extent you do, even if only somewhat, I'd love if you wrote a tweet-length personal story or directional pointer towards the experience 🗨️
@ChrisChipMonk uhhh I don't remember the motivating example but it was something to the effect of why certain products don't exist that the market doesn't properly incentivize maybe
the point is that if you have the source code and the time/means to compile it, you can throw out the executable likewise, if you have a good spec and the time/money/means to generate code for it, you can throw out your existing code
I have now tested this Claude Code /project:wakeup script a bit and it seems to work quite well. new best practice imo cc @JimmyRis
obliberate
if someone presents you with a choice between self-trust and external-authority, *actual* self-trust is never what they say it is
@visakanv this book is like... the manual for how and why to focus on what you want to see more of. like, turned up to 11. it's good shit. idk if you personally need it but it's hitting me good
> "In the long run, we find what we expect. We shall be fortunate then if we expect great things." I keep expecting to see things that relate to psycho-cybernetics and I keep seeing them
guess what the largest number of spaces in a row is that's considered a single LLM token (for OpenAI's tokenizer)
guess what's the largest number of spaces in a row is that's considered a single LLM token (for OpenAI's tokenizer) place exact or more nuanced guesses below the poll