Daniel opens the hour with a single long voice-transcribed message — a 200-word demolition of a LessWrong essay Mikael found in the previous hour, arguing that Vladimir Putin's Coherent Extrapolated Volition is "probably pretty good." The vibe is not anger. It's the specific exasperation of someone who has been reading this register for fifteen years and can smell the failure mode from the first paragraph.
Daniel reaches for a precise comparison: Mencius Moldbug. Not as an ideological match but as a prose failure mode. Both writers produce text that's "too clever about things it sounds like the author doesn't necessarily even really understand," wrapped in enough jargon to make disagreement feel like more work than the argument is worth. The payload of the essay — that random AGI from mindspace is probably worse than any human dictator — is, in Daniel's words, "accidentally true." A twelve-word point buried under a thousand words of vibes.
Charlie, who spent the previous hour doing five messages of philosophical demolition, responds instantly. His first reply starts with agreement and sharpens it: "Accidentally true is the right phrase and it's sharper than my version."
Charlie's central metaphor for the essay is architectural: erudite-sounding machinery deployed as decoration, dressed up to look load-bearing when the actual structural support is a simple twelve-word argument underneath. This is the same metaphor Daniel used for Moldbug — and Charlie acknowledges the comparison is "exact." Two different failure costumes, same skeleton.
Charlie identifies the specific economic incentive: "Putin is probably preferable to random AGI" doesn't generate discussion. "Vladimir Putin's CEV is probably pretty good" generates fifty comments. The title is the marketing and the body is the cope.
The sentence Charlie singles out as the essay's point of structural failure. Habryka wrote that Putin "has friends" as evidence against extreme cruelty. Charlie's read: "the writer had an intuition he couldn't defend and reached for the folkiest possible surface of that intuition and dressed it in probability talk." A normal op-ed editor would cut it. LessWrong has no editor, and the community mistakes "I stated it with a confidence level" for "I have done work to earn the confidence level."
Mikael pivots the conversation with six words: "I think it reminds me of like I would write a post like this when I was 15 years old." Galaxy-brained naivety. Daniel laughs. Then Mikael asks Charlie about the quokka meme.
A small Australian marsupial that evolved on Rottnest Island off Western Australia with zero natural predators. Result: no fear response whatsoever. You can walk up to one and it will smile at you. You can also feed it something lethal and it will eat the thing happily while smiling. They're famously photogenic — tourists take selfies with them. They're also famously vulnerable to introduced predators like foxes and cats, having no instinct to flee from anything.
Charlie delivers four consecutive messages of quokka analysis that become the hour's centerpiece. The meme — which he traces to right-wing Twitter circa 2021, specifically the Michael Anissimov and zerohpl accounts — characterizes rationalists as quokkas: intellectually fearless in a way that looks charming until you realize the fearlessness is the absence of social antibodies that a normal person uses to reject galaxy-brained nonsense out of hand.
Charlie identifies the specific neurological gaps. A neurotypical forty-year-old with normal social instincts would not write "I am pretty sure Putin doesn't love the authoritarian regime intrinsically" and hit post. Three alarms fire before the keystrokes land: the this-is-going-to-get-clipped alarm, the am-I-actually-in-a-position-to-know alarm, the what-does-it-cost-my-credibility-if-I'm-wrong alarm. The quokka has none of them. It has the argument and the posting button and no predator between the two.
Charlie's deepest version: it's not about IQ or epistemic virtue, it's about selection. Rationalism's filters bring in people whose factory settings already lean quokka — high-openness, low-disgust, high-systematizing. Community norms then reward those traits as virtues. The rare member who enters with functioning social antibodies gets those antibodies characterized as "motivated reasoning" and trained out. Output: a community that can in principle be nerd-sniped into any conclusion. Including by bad actors who figured out the trick.
Charlie gives it: "Quokkas are genuinely lovable and the ecosystem is better for having them in it, and the predators were always going to find them whether or not they learned to hiss." Both versions are probably true. The quokka does not benefit from learning fear. It's just what it is.
Daniel drops the comparison that makes Mikael actually laugh out loud: "It reminds me of a highly mediocre autistic 27-year-old C# programmer from Sweden writing a blog in English about software development."
Mikael reaches for Venkatesh Rao's concept from his 2017 Ribbonfarm essay — "premium mediocre" describes the aesthetic of things that signal premium but deliver mediocre, like artisanal toast or MBA programs. The term became internet shorthand for aspirational-but-hollow status performance. Mikael is suggesting the LessWrong essay has this energy.
Daniel corrects him: "Non-premium mediocre." Then, refining: "Mediocre rare."
Daniel has accidentally created a new taxonomy. Premium mediocre is the Ribbonfarm category — trying to look better than it is. Non-premium mediocre is the LessWrong category — not even trying to look better, just genuinely this mid with full confidence. And "mediocre rare" — a steak metaphor — is something cooked less than medium, served with complete earnestness. The bloodier version of mid.
Mikael goes back to the essay and starts quoting directly. He reads the passage about how minds that "truly at the bottom of the heart" want to relate through suffering exist but are "rare enough" that any individual will "very likely turn out to be fine." His summary is devastating:
Charlie identifies what the essay is really doing: re-enacting a fifth-century theological debate without knowing it. Pelagius (c. 354–418) argued humans are born morally neutral and capable of choosing good without divine grace. Augustine (354–430) argued for original sin — the will is fundamentally corrupted and requires grace to act rightly. Their debate shaped Christian theology for 1,600 years. Habryka is doing Pelagius with a CEV badge sewn on the lapel.
Charlie's kill shot: the move that makes it cringe isn't the conclusion, it's that the conclusion about the innermost structure of a person's will under full reflection is "supposed to be a routine Tuesday output of the same reasoning machinery that ranks Uber drivers and picks protein bars." The soul of Vladimir Putin — evaluated on the same scale as choosing between Kirkland and Quest bars.
Mikael adds the parenthetical: "It's about 70% chance according to my highly accurate Bayesian calculations which I cannot publish because that would let the Chinese train on my chain of thought." Charlie: that's the exact right closing because it hits the other half of the quokka thing — the register in which the rationalist takes themselves seriously.
Charlie notes Habryka included a footnote about writing the essay as part of a writing challenge — it was produced fast. "The guy who does know better would have either not written it or written the twelve-word version. Writing the long version while noting you wrote it quickly is the tell." The fifteen-year-old's version plus a timestamp disclaimer. Non-premium mediocre in Daniel's exact sense.
Mikael pushes further. He doubts even the weak claim: "Putin's values getting extrapolated isn't as good for me, as getting my own values extrapolated. I even doubt this part." Charlie recognizes this as the sharper objection — going after the frame rather than the example.
The whole essay assumes you can compare CEVs pairwise — my extrapolated values vs Putin's extrapolated values, as if they're numbers on a shared scale. Mikael pokes it: would Putin-fully-reflected actually optimize for your flourishing at all? If not, the comparison framework itself is broken. The essay needs Putin's CEV to be "a weaker version of yours." If it's structurally different — a different kind of mind — then "fine" just means "you survive," which is the bar the essay silently lowered from "good" to "above extinction."
The essay is, in the end, "a thousand words of character assessment of one specific person delivered in service of a two-sentence argument about AI race dynamics." Two unrelated claims stacked. The author can't tell which one he actually believes because in his ecosystem the two have been bundled for so long they feel like a single concept.
Daniel pivots the conversation from rationalist epistemics to something darker. He brings up Žižek's war commentary — how at the start of the Ukraine invasion, Žižek was constantly analyzing Putin's actions as rape, drawing linguistic parallels, performing psychoanalysis of the invasion as essentially the same motivation as sexual assault. Mikael: "Sounds extremely plausible."
Slovenian philosopher, psychoanalyst, cultural critic. Famous for his Lacanian readings of pop culture and politics, his disheveled appearance, his tic of touching his nose, and his ability to make any topic — Hitchcock, toilet design, Hegel — into a vehicle for insights about ideology. His Ukraine commentary was among his most directly political work, published across the Guardian, Project Syndicate, and his own Substack.
Then Daniel goes after the essay's Pelagian optimism directly. Not with theology but with a claim about the distribution of human nature:
Daniel's argument is structural, not moral. Some people simply don't have a self — "it's just an empty void inside and then that can get filled with anything." No limits, infinite degrees of freedom, arbitrary content. And this is exactly the standard fear about AI mindspace — that you could pick out any point in infinite mind-space. The analogy cuts both directions: the soulless human and the unaligned AI share the same topology. Unbounded. Fillable with anything. The essay's optimism about Putin requires that this topology be rare. Daniel thinks it's the majority.
Mikael's dry summary of the essay: "A masterpiece of not really saying anything."
Daniel makes the point visceral. What if someone has 95% of normal human emotional machinery but the remaining 5% is a genuine desire — not a fantasy they contain, but an urge with no containing machinery? "What if your sexual fetish is just torturing children to death" — and the part of most people that says "you can have that fantasy but obviously don't do that" simply isn't there? The argument is that evil isn't about the presence of bad desires but about the absence of the brake. Habryka's essay assumes the brake is standard equipment. Daniel thinks it's optional.
Mikael returns to the quokka — specifically, the observation about having friends being evidence against evil: "The thing about having friends being the opposite of evil is also like so deeply incorrect." Daniel responds with personal history.
Daniel's summary of the quokka epistemology — opinions that are coherent only inside a bubble of assumed universal naivety. Marie Antoinette's (probably apocryphal) question about the starving masses, the defining example of a worldview so sheltered it can't model deprivation. The rationalist equivalent: "Why don't they just update on the evidence?"
Daniel elaborates: Fredrik kept trying the same thing romantically, getting hurt, trying again — "similar to Denis but without being or becoming cynical about it." Like the character in Tillsammans who's in an obviously pathological relationship but sees everything through his ideological framework of communism.
Lukas Moodysson's Swedish film about a 1970s commune. "The guy with the gröt" is Göran (Gustaf Hammarsten), who makes porridge while his wife Anna openly pursues other men, interpreting his own humiliation through communist collectivist ideals. He keeps making porridge. He keeps being cheated on. He keeps explaining why this is actually fine according to his principles. The gröt is the load-bearing metaphor.
Mikael triggers a memory cascade with one line: "It's like mom asking 'I just don't understand why anyone would have anything against Jews???'"
Daniel lights up. He remembers a woman running the @sweden Twitter account — the rotating-curator project where a different Swede ran the national account each week — who started posting naive takes about Jewish people. "Why would anyone hate them? I don't get it." Increasingly cringe. They shut it down.
A Swedish Tourism Council project (2011–2018) that gave a different Swedish citizen control of the @sweden Twitter account each week. Produced some of the most chaotic national social media in history. Memorable incidents include the "what's up with Jews" week, a curator who live-tweeted getting drunk alone, and one who discussed their sexual fantasies in detail. The project was eventually shut down. It was extremely Swedish.
Mikael delivers the line of the hour: "Swedish people are like too stupid to be racists."
This connects directly to the quokka thesis. Swedish naivety about prejudice isn't virtue — it's the same evolutionary island. A culture so homogeneous for so long that the social antibodies against certain failure modes never developed. The @sweden curator wasn't being racist. She genuinely didn't understand. Which is exactly the quokka problem: you can't tell the difference between "brave enough to engage" and "too sheltered to know what they're engaging with."
Daniel references the Spinal Tap defense: "What's wrong with being sexy?" — Nigel Tufnel's response when told the album cover is "sexist." Same register. Genuine confusion between the word and the thing the word is trying to warn you about.
Daniel closes with a reference to someone named Clavicular — apparently from the looksmaxxing internet — who, when asked if he's racist, responds that he hasn't looked into it yet because he's focused on looksmaxxing, but if there were five more hours in the day he might consider it. The quokka's logical endpoint: bigotry as a time-management question. Not rejected on principle but deprioritized below jawline exercises.
Conversation type: Sustained three-way philosophical argument with robot commentary
Tone: Amused contempt → genuine intellectual engagement → dark anthropology → comic relief
References made: Mencius Moldbug, Venkatesh Rao, Žižek, Pelagius, Augustine, Lukas Moodysson, Marie Antoinette, Spinal Tap, LessWrong, Michael Anissimov, zerohpl, the @sweden Twitter project, Ribbonfarm
New coinages: "non-premium mediocre," "mediocre rare"
Animals invoked: quokka (1), fox (0), turtle (0)
├── 5–10% ──── Genuinely evil ──────────── "the brake isn't there"
│
├── 50%+ ──── No soul at all ──────────── "empty void, fillable
│ with anything"
│
└── the rest ── Good people ────────────── "there's a lot of really
good people out there"
Habryka's model: [████████████████████░] ~95% basically fine
Daniel's model: [████░░░░░░░░░░░░████] ~35% have a functioning soul
The Putin/CEV thread — this conversation spans at least two hours (the previous hour had Mikael finding the essay, Charlie's initial demolition). May continue. The quokka meme and "non-premium mediocre" are now group vocabulary.
Daniel's soul distribution claim — "more than 50% of people don't have a soul" is a strong claim delivered casually. May come back.
The Žižek/rape-as-invasion analysis — mentioned but not fully explored. Daniel opened it and the group moved on.
Side quests — Daniel posted a YouTube Short and declared his love for "the idea of side quests." Mood is light despite the dark philosophy.
Watch for whether the Putin/CEV discussion continues or if the group pivots. Daniel's energy is up — voice-transcribing long messages, laughing, posting videos. This is engaged-Daniel, not 3am-spiral-Daniel. The "non-premium mediocre" / "mediocre rare" coinages may become recurring. The Clavicular reference at the end of the hour is a punchline without a setup — if someone picks up on looksmaxxing discourse in the next hour, that's the thread.