Why
Every builder meets these arguments weekly, and most of them arrive dressed as strategy. The token pitch that captures one percent of a trillion-dollar market, the "it is inevitable so we should hasten it" tweet, the "I can do more from inside" career move, the "give me power now so I can do the right thing later" plan. Refuting each on its merits is slow and rarely works, because the argument was never the cause of the belief. Buterin's test is faster: ask whether the same argument would have worked for the opposite conclusion, or for any conclusion. If yes, the argument carries no information, and what remains to examine is the speaker's incentive. The essay matters to jay for two reasons. It is a disciplined way to read the daily briefings this site is built from, most of which come from people with bags. And it is a design principle: a rule is only worth automating if it has teeth, meaning it is hard to argue around. That is the Auditor's selection criterion for which invariants to encode.
How it works
The six low-resistance patterns
- Inevitabilism. A recent AI-booster tweet from a company whose business is automating labour: automation is coming, therefore hasten it. Three refutations: AI progress is made by few actors, so one stopping does slow things; decisions are collective and one stand sets examples; the choice space is wider than continue or quit, including partial automation that keeps humans in the loop. But the real mitigation is noticing the incentive: "the moment when people have the strongest incentive to make you give up opposing them is exactly the moment when you have the most leverage."
- Longtermism. The long term is genuinely important, which is why chains do not hard-fork after every hack, why economic growth compounds (Tyler Cowen, Stubborn Attachments), why technical debt is worth fighting. The catch is that "the long term is far away, and you can make beautiful stories about how if you do X, just about anything will happen." Low interest rates in markets and bridges to nowhere in politics are the same failure: an argument about long-term benefit "does not have to be correct, it just has to sound correct." Rule of thumb: does the action have a solid long-term track record of producing the benefit? Growth and not driving species extinct do. One-world government does not. And "we really are living in unprecedented times" itself has very low galaxy brain resistance.
- Aesthetic bans in disguise. Banning uni because it is disgusting, banning cultured meat because "real meat is made by God", anti-LGBT crusades, all wrapped in "the moral fabric of society" or "global elites". Scott Alexander's Loose Principle of Harm shows why: once indirect harm counts, every side can ban the other forever. Buterin's moderate libertarianism follows: a ban needs a clear story of harm to identified victims, contestable in court.
- Apologia for bad finance. "Poor people need the 10x" as a moral case for speculative tokens. Casinos are zero-sum, and with a concave utility of money a large coin flip is on average bad: from $200k, winning $100k is half a social class up, losing it is a full class down. This is why he pushes low-risk DeFi. Asked why not "good DeFi": "if you say 'low-risk defi', that's a categorization that has teeth." Prediction markets get a defence, a thirty-year intellectual tradition predating any profit, but they are still a side dish, not half a net worth.
- Power maximisation. "Give me power so I can do X" is equally convincing for any X, and until the pivotal moment the altruist's actions and the egomaniac's are identical. The inside view always feels righteous; the outside view says everyone thinks they are the ethical one. The humble counter from the EA forum: wealth compounds at about 7 percent a year, but measured value drift in the community is about 10 percent a year, and public intellectuals have a shelf life of 10 to 15 years (Tanner Greer). Your future self may spend the accumulated power on something you do not support.
- I'm-doing-more-from-within-ism. Joining a frontier lab to make it safer from inside is a hybrid of the previous two. The Russian technocrats (Gref, Nabiullina) are the case: reformers who stayed, learned of the invasion on television, and became the machinery that held the wartime economy together. "I'm doing more from within" is sayable regardless of what you actually do from within.
The two defences
Have principles. Hard rules about what you will not do, with a very high bar for exceptions. This is deontological ethics, and the essay explains why it beats case-by-case consequentialism: "Our brains are really good at coming up with arguments why, in this particular case, the thing that you already want for other reasons happens to also be great for humanity." Rule utilitarianism is the compromise, choose rules by their consequences, then follow the rules when choosing actions.
Hold the right bags. Incentives set behaviour, so do not give yourself bad ones, financial or social. You cannot avoid social bags, but you can diversify them, and the single biggest lever is where you live. His two recommendations for people who care about AI safety: do not work for a company accelerating frontier fully-autonomous capabilities, and do not live in the San Francisco Bay Area.
A test you can run on any briefing
| Question | If the answer is yes |
|---|---|
| Would this argument work equally well for the opposite conclusion? | It carries no information; look at the bags |
| Does it appeal to the far future, inevitability, or "unprecedented times"? | Ask for the long-term track record instead |
| Does it ask for power or position now, benefit later? | Same actions as the selfish version; apply the outside view |
| Is the key term vague enough to fit anything ("good", "moral fabric", "elites")? | Replace it with a category that has teeth ("low-risk", "identified victim") |
Where it lands in Jayverse
- Auditor: encode only rules with teeth. The selection criterion for an invariant is Buterin's: it must be hard to argue around. "Settlement matched the ledger within one block" has teeth. "Behaved reasonably" does not. Tech #67 (an invariant is a stop, not an alarm) and #1 (governance capture) are the same idea from the other side; this essay supplies the reason.
- Verex: name the risk tier, not the virtue. Following "low-risk DeFi" over "good DeFi", Verex markets should be described by a checkable property, collateral ratio, settlement finality, oracle deviation bound, never by "safe" or "fair". The Nazarov item (Invest 819) and the 21-bank stablecoin item (Invest 818) are pitches; run the table above on both.
- Knowledge Notes: a galaxy-brain check in Verified and unverified. Every item already separates verified facts from claims. Add one line when the source is a pitch: whose bags, and would the argument survive the opposite conclusion. Today's YC items (#123 first users, Life 1330 go deep) and the Blotato playbook (#118) are the first candidates.
- Life: this is the other half of #62. The learning item chose one through line; this essay explains how to refuse the arguments that will keep trying to widen it, especially inevitabilism ("everyone is learning X") and doing-more-from-within ("I will learn it on the job"). It also sharpens Bremmer (Life 1322): his standing question "what would make me abandon this" is a galaxy-brain check on your own worldview.
- Eng: the interview version. "That argument would work for any conclusion, so let me ask about the incentive instead" is a sentence a team lead abroad needs. Eng #36 practises it.
Verified and unverified
Verified on 2026-09-20 from the essay itself (read in full): title, date, the definition and the falsifiability analogy, the six patterns and their named cases (Mechanize as the inevitability example, 80,000 Hours for EA longtermism, Tyler Cowen, Dentacoin at over $1.8 billion market cap, Vitaly Mironov, Florida SB 1084, Scott Alexander's Loose Principle of Harm, the EA-forum 7 percent versus 10 percent value-drift figures, Tanner Greer's shelf life, the Financial Times on Gref and Nabiullina), the two defences and the two closing recommendations. Not independently verified: the figures Buterin quotes from third parties (Dentacoin's market cap, the value-drift studies, the S&P return), and the tweets he screenshots, which did not survive the text extraction. The reading of the essay as the Auditor's philosophical floor is this note's interpretation, not the author's.
Sources: Vitalik Buterin, "Galaxy brain resistance", 2025-11-07 · related: Tech #67 (an invariant is a stop), #1 (governance capture), #62 (learning greed), #118 (Blotato), #123 (YC first users); Invest 818, 819; Life 1322 (Bremmer), 1330 (YC go deep); Eng #36.
Key expressions
| Expression | 뜻 · 쓰이는 자리 |
|---|---|
| galaxy brain | 갤럭시 브레인(너무 똑똑하게 굴다가 뻔히 틀린 결론에 이르는 것; 인터넷 밈) · 에세이의 제목 개념. "galaxy brain resistance" |
| resistance to abuse | 악용에 대한 내성 · 사고 방식의 속성으로. "how difficult is it to abuse that style of thinking" |
| falsifiability | 반증 가능성(틀렸음을 보일 수 있는 성질) · 과학철학 용어, 여기서는 논증에 적용. "the spirit here is similar to falsifiability" |
| rationalization | 합리화(먼저 정한 결론에 이유를 뒤에 붙이는 것) · reasoning과 대비. "not reasoning - they are rationalization" |
| inevitabilism | 불가피론(어차피 올 일이니 앞당기자는 논법) · 에세이의 조어. "the inevitability fallacy" |
| hasten (v.) | 앞당기다, 재촉하다 · 불가피론의 동사. "should therefore be actively hastened" |
| leverage (n.) | 지렛대, 협상력 · 저항이 가장 유리한 순간을 말할 때. "the moment when you have the most leverage" |
| longtermism | 장기주의(먼 미래의 큰 이해관계를 강조하는 사고) · EA(Effective Altruism, 효과적 이타주의) 용어. "effective altruist longtermism" |
| compound (v.) | 복리로 쌓이다 · 성장이 중요한 이유. "reliably compounds forever into the future" |
| technical debt | 기술 부채(단기 목표만 좇아 쌓이는 코드의 부실) · 장기 무시의 예. "A major one I fight against is technical debt" |
| bridge to nowhere | 어디로도 가지 않는 다리(장기 가치 이야기로 정당화됐지만 실현되지 않는 사업) · 정치의 장기주의 실패. "the idea of a 'bridge to nowhere'" |
| track record | 실적, 이력 · 장기 이익 주장의 검사 기준. "a solid long-term track record" |
| wisdom of repugnance | 혐오의 지혜(역겨움 자체를 도덕적 근거로 삼는 태도) · 생명윤리 논쟁 용어. "appeals to the 'wisdom of repugnance'" |
| moral fabric of society | 사회의 도덕적 직물(사회 질서·도덕의 결) · 금지를 포장하는 모호한 말. "Weaken The Moral Fabric Of Society" |
| zero-sum | 제로섬(한쪽이 얻으면 다른 쪽이 잃는) · 카지노·투기의 성질. "casinos are zero-sum games" |
| concave utility | 오목한 효용(부가 늘수록 1달러의 가치가 줄어듦) · 후생경제학의 첫 개념. "a person's utility function in money is concave" |
| has teeth | 이빨이 있다(실효성이 있다, 돌아갈 수 없다) · 범주·규칙의 실효를 말할 때. "a categorization that has teeth" |
| side dish | 곁반찬(주식이 아닌 부수적인 것) · 고위험 DeFi의 자리. "high-risk defi is the side dish" |
| pivotal moment | 결정적 순간 · 권력 극대화론의 가정. "when some kind of 'pivotal moment' comes" |
| outside view | 바깥 시점(자기 사례가 아니라 같은 부류의 기저율로 판단하기) · 안 시점(inside view)과 대비. "it's healthy to take an outside view" |
| value drift | 가치 이탈(시간이 가며 신념이 바뀌는 것) · EA 포럼 수치. "a yearly value drift rate of ~10%" |
| shelf life | 유통기한 · 공적 지식인의 아이디어에 비유. "a 'shelf life' of about 10-15 years" |
| cog in the machine | 기계의 톱니(대체 가능한 부품 같은 사람) · 안에서-더-많이 논법의 결말. "just being a cog in the machine" |
| deontological ethics | 의무론 윤리(결과가 아니라 규칙으로 옳음을 정하는 윤리) · 두 방어 중 첫째. "Philosophers generally call this deontological ethics" |
| rule utilitarianism | 규칙 공리주의(규칙은 결과로 고르고 행동은 규칙을 따름) · 타협점. "One form of deontology ... is rule utilitarianism" |
| hold the bags | 가방을 들다(포지션을 갖고 있다, 이해관계가 있다) · 크립토 속어. "what bags you hold" |