NNaval
← All episodes
22 December 2021

The Beginning of Infinity, Part 2

4Frameworks
15Insights

Listen

Frameworks in this episode

Insights & moments

The myth-busts, hot takes, explainers, and tools worth keeping.

Myth Buster· 2

Myth Buster

Bayes is a fine formula and a broken theory of knowledge

Bayesianism is presented here as the modern form of justified true belief: hold a theory, gather evidence, grow more confident. The objection is not to Bayes theorem, which is a perfectly acceptable statistical formula used legitimately all the time, but to treating it as epistemology. It cannot generate new theories, only rank confidence between existing ones. The killer example is gravity before 1919 — every experiment was consistent with Newton, so the Bayesian's confidence peaked the day before the theory was shown to be false.

  • Justified true belief is still the prevailing idea of knowledge.
  • Bayes theorem is valid statistics but not a theory of knowledge.
  • Bayesian reasoning cannot generate new theories, only compare confidence.
  • Pre-1919 gravity: maximum confidence arrived immediately before refutation.
  • Popper's answer: Newton's theory was never true, it contained truth.

Anyone who calls themselves a Bayesian is a justified true believer.

How do you square that circle of the day before it was shown to be false was the day when you're most confident in it?

#bayesianism#popper#epistemology#science
Myth Buster

The follow-up question that exposes a parroting AI

The AGI-by-scale position — add more compute and intelligence appears — is rejected here because we do not know what makes us creative in the first place. GPT-3 is described as an impressive text matching engine whose good outputs depend on a human selecting them from the garbage. The proposed test is cheap: take a generated output and ask why, or ask it to make a prediction from it, and watch it collapse. There is no underlying model of reality that can explain the seen in terms of the unseen, so there is nothing to follow up.

  • Compute alone does not produce creativity, because we cannot say what creativity is.
  • A human filter is doing the selection work behind impressive generated output.
  • Test: ask why, or ask for a prediction, and the explanation is not there.
  • Parroting is Bayesian reasoning over what humans already wrote on the web.
  • Deutsch's quip: if you can't program it, you don't understand it.

Take a GPT 3 generated output and ask it why.

Just add more compute power and you'll get intelligence when we don't really know what it is underneath that makes us creative and allows us…

#agi#gpt-3#creativity#explanation

Hot Take· 3

Hot Take

Against the dark forest: aliens would come to trade ideas

The sci-fi default is that contact means destruction, as with the Vogons demolishing Earth. The argument here inverts it. Any species capable of interstellar travel had to be maximally creative, which requires freedom and the absence of coercive centralised authority — so it will be non-violent. It will also have no interest in our matter, since it can sweep hydrogen from intergalactic space and print what it wants. The only thing it could lack is knowledge it does not have, which makes ideas the sole trade good and contact something to welcome.

  • Creativity requires liberty, so advanced civilisations are non-violent by construction.
  • The dark forest hypothesis assumes a Malthusian universe that is not running out.
  • Advanced species can build resources from hydrogen; only knowledge is scarce.
  • Every civilisation is a universal explainer, so any idea can be traded and understood.
  • The best way to be worth trading with is a dynamic, abundant, thriving civilisation.

Smart alien civilizations trade ideas, and successful civilizations trade ideas, because those ideas take things that were useless before and turn them into resources.

So if we encounter an alien species, we should probably rejoice.

#aliens#dark-forest#trade#abundance
Hot Take

Graduate counts are not innovation: the China challenge

Impressive rocket launches, GDP and new cities are dismissed here as imitative — technology picked up elsewhere and scaled by having more people. The stated bar is invention: new pharmaceuticals, vaccines, spacecraft, energy generators or fusion reactors that we do not know how to build, and GDP per citizen crossing ours. Graduate volume is rejected as a proxy because students trained to memorise a textbook and pass an exam are trained to imitate, not to question. The closing line is that creativity goes zero to one and no amount of throwing bodies at it will get you there.

  • The test is a genuinely new invention, not scale applied to imitated technology.
  • GDP per citizen is offered as the real comparison, not aggregate GDP.
  • More science graduates does not mean more innovators.
  • Exam-optimised training produces imitation, not frontier-pushing.
  • Ten creative physics graduates beat fifty thousand who all score 100%.

Call me when they invent something new.

Give me 10 innovative, creative young physics graduates over 50,000 physics graduates that are all able to pass the exam with 100% efficiency any day.

#china#innovation#education#creativity
Hot Take

Post-capitalism has two options, and one of them is a thug

Asked what replaces capitalism, the answer offered is that there are essentially two ways to divvy up credit and reward work: feedback from free markets and reality, whose best instrument is money, or feedback from people, which is where communism lands. The second requires someone to decide who did the best work, and that role reliably accrues to whoever holds the most force — which is why every communist country degenerates into a dictatorship. The free-market claim is narrower than usually stated: what it removes from the decision is coercion.

  • Two options: an objective function, or subjective allocation by people.
  • Money is the known objective function for allocation.
  • Subjective allocation gets captured by friends, family and eventually force.
  • Coercion enters when government mandates the answer for everyone.
  • Individuals must be free to choose wrongly — trying and failing is the only error correction.

Either you need an objective function to carve it up, and money is the known objective function, or it becomes all subjective.

So I don't think it's an accident that every communist country degenerates into a dictatorship.

#capitalism#communism#markets#incentives

Explainer· 8

Explainer

The four theories David Deutsch says one person can hold

Deutsch argues that everything important about reality is understandable by a single human, and that it reduces to four base theories. Those are epistemology, evolution by natural selection, quantum theory (with relativity and the rest of physics subsumed into it), and the theory of computation including quantum computation. The episode works through what is non-obvious in each. The framing is deliberately physics-first — Deutsch is concerned with reality and truth-seeking rather than human systems, which is why candidates like Adam Smith's economics are left out.

  • The four: epistemology, evolution by natural selection, quantum theory, computation.
  • Relativity and the rest of physics get folded into quantum theory.
  • Claim: the underlying base theories that drive reality fit in one human head.
  • Sociological theories like Adam Smith's are excluded because Deutsch is a physicist.

Deutsch has this great view of the world where he believes that everything important is understandable by a single human.

The four that he picks are the theory of epistemology, the theory of evolution by natural selection, quantum theory.

#deutsch#epistemology#physics#computation
Explainer

Darwin and Popper found the same mechanism twice

Before Darwin the only explanation for living things was a creator, and Lamarck's gradual-change idea had no mechanism. Darwin supplied one that mirrors Popper's account of knowledge: variation trialled against an environment, with error correction by death of the unfit. Neo-Darwinism identifies the gene, not the group or the individual, as the unit of selection. The episode extends the pattern upward — genetic evolution was a prelude, and the history of the universe from here is memetic, with criticism weeding out ideas the way environments weeded out genes.

  • Lamarck offered gradual change with no mechanism; Darwin supplied error correction.
  • The environment gives feedback from reality by destroying the unfit.
  • The gene is the unit of selection, not the herd or the individual.
  • Memetic evolution takes over from genetic: criticism replaces death.
  • Same pattern shows up in invention, capitalism and startups failing.

So what Darwin came up with is a similar idea to what Popper had in knowledge. It was error correction.

Deutsch likes to say genetic evolution was merely a prelude.

#evolution#darwin#memetics#error-correction
Explainer

Why humans are off-axis, not just smarter chimps

Science moved us from the centre of the universe to being nothing special — one unremarkable planet, one point on a continuum from bacteria to chimpanzees. The counter-argument here is that the difference is qualitative, not quantitative: we are off-axis. Humans are universal explainers, capable of explaining the seen in terms of the unseen, and therefore capable of maximal knowledge. There is no concept possible in this physical reality that a human could not understand given enough time, resources and education — which also disposes of the idea that aliens would have incomprehensible thoughts.

  • The cosmological principle made us unspecial in place; biology made us unspecial in kind.
  • The rebuttal: we are qualitatively different, not slightly ahead of chimpanzees.
  • A city outside your window cannot be explained by gradual biological complexity.
  • Claim: no concept exists that a human could not understand, given time and education.
  • Saying the next laws of physics are beyond us is an appeal to the supernatural.

There is no concept out there that is possible in this physical reality that a human being, given sufficient time, resources, and education could not…

In fact, the mind-blowing thing here is that your mind cannot be blown.

#universal-explainers#human-exceptionalism#knowledge#deutsch
Explainer

A biological answer to Fermi: the odds against a second us

The standard Fermi setup counts stars and galaxies and concludes the universe should be teeming. The counter-argument here is biological rather than astronomical. Life spent roughly two and a half billion years as nothing but bacteria, and convergent features like wings and eyes arose repeatedly while the capacity to do mathematics or build radio telescopes arose exactly once. Chain even a generous hundred independent steps at one in ten each and you get one in ten to the hundred — a number that swamps the count of planets. The fact that it happened once should blow our minds.

  • About 200 billion stars per galaxy and 200-300 billion visible galaxies sets up the paradox.
  • Life stayed bacterial for most of Earth's history — no impetus to complexify.
  • Wings, eyes and hearing are convergent; creative intelligence arose once.
  • Lineweaver's planet of the apes hypothesis: evolution has no direction, only hindsight.
  • A deliberately generous 100-step chain still yields odds that swamp the planet count.

The argument is that we are alone. And the argument has nothing to do with astronomy. It has everything to do with biology.

In other words, the probability of us arising on this particular argument is infinitesimally small.

#fermi-paradox#evolution#probability#aliens
Explainer

The credential priesthood is burning its own capital

Institutions used to be how credibility was conferred — a masthead at the New York Times, a professorship at Harvard — and those systems got hacked by people using the credibility capital others built. The internet is exposing this slowly, but the transition is messy because a diploma is still the only cheap filter a hiring manager has, making it a Schelling point everyone is stuck on. Meanwhile academia trains the next generation of teachers, which the guest calls a priesthood: the priests decide what may be said, and who counts as a priest.

  • Mastheads and professorships were the old credibility-conferring mechanism.
  • Activists and social scientists are burning credibility capital built up over decades.
  • A diploma persists as a coordination point because no cheaper filter exists.
  • Power is transferring from institutions to individuals over a generation or more.
  • The empire strikes back phase: institutions capturing Twitter, Facebook and Patreon.

What we're seeing is the transition of power from institutions to individuals, but it's going to be messy.

Yeah, it's a priesthood.

#credentialism#institutions#media#academia
Explainer

Innovation needs two things: decentralisation and a frontier

The pattern offered is that hotbeds of innovation share a structure — Athenian city-states, Italian city-states, an early loosely federal United States — where competition was real and people could switch jurisdictions when their ideas were unwelcome. The diversity that matters is diversity of ideas. The second requirement is a frontier, intellectual or physical, and California has become the establishment rather than the Wild West. Frontiers are always full of scammers and weirdos, which is exactly why they host the innovation: Paul Graham's Steve Jobs would be in crypto today.

  • Decentralisation gives competition and the ability to exit to another jurisdiction.
  • The diversity that matters is diversity of ideas.
  • You also need a frontier — intellectual, like crypto, or physical, like space.
  • California is now the institution rather than the frontier.
  • Frontiers attract scammers and weirdos and that is where the innovation happens.

Innovation requires a couple of things. One of the things that it seems to require is decentralization.

And you also need a frontier. You need something new to explore, either an intellectual frontier or physical frontier.

#innovation#decentralization#frontier#crypto
Explainer

The pendulum: kings, brigands, DeFi and DoorDash

Centralised finance ossifies under regulators and regulatory capture until Wall Street extracts a large share of the economy, so decentralisation pressure builds — and brings scams, fraud and losses with it. The historical version is brigands in the forest, then a king who mints money and builds a keep, then a king who debases the currency and jails people, then people running back to the forest. Technology swung centralised in the last decade, but Shopify lets small stores compete and DoorDash mobilises a decentralised army of restaurants.

  • Centralised finance ossifies into regulatory capture and rent extraction.
  • Decentralisation buys freedom and pays for it in fraud and losses.
  • The king-and-brigand cycle is the same pendulum in historical dress.
  • The last decade's technology arc ran towards centralisation.
  • Shopify and DoorDash are centralised services enabling decentralised sellers.

There's a pendulum between centralization and decentralization.

So there's natural pendulum swing that goes on in history between centralization and decentralization.

#centralization#crypto#history#platforms
Explainer

Why making a field social destroys its ability to find truth

Social groups need consensus to survive, and consensus runs on compromise rather than truth-seeking — so socialising a discipline degrades it. The natural sciences worked because individuals could truth-seek on behalf of society and others would verify the result. Social sciences are described as politically compromised in both directions: dependent on society for funding and feeding models back into policy. The historical pattern offered is that most breakthroughs came from independent natural philosophers who were reviled, sometimes persecuted, and vindicated decades or centuries later.

  • Groups need consensus; consensus needs compromise; compromise is not truth-seeking.
  • The more groupthink, the further from the truth.
  • Social sciences are funded by and feed back into the politics they study.
  • Natural sciences are described as increasingly socialised too.
  • Most big breakthroughs came from independent thinkers outside institutions.

Making something social destroys the truth of it because social groups need consensus to survive.

The more groupthink you see involved, the further from the truth you actually are.

#groupthink#science#academia#truth

Story· 1

Story

The foundation where every project was a success

Serving on the board of a grant-giving foundation turned out to be disillusioning: every project was declared victorious, with backslapping, mission statements and nice dinners, and nothing ever got done. The diagnosis is missing feedback — no loss, only social profit, therefore no possibility of failure and endless misdirected resources until the money runs out. The generalisation is that groups never admit they were wrong because a group that changes its mind falls apart; you get schisms and forks instead, as with the Catholic-Protestant split or forking coins.

  • Every funded project was declared a success; nothing shipped.
  • No loss and no objective feedback means failure is impossible and resources are misallocated.
  • Individuals can eventually be forced to admit failure; groups cannot.
  • Groups that try to change their minds fracture into schisms and forks.
  • Conclusion: to change the world, a for-profit is the more sustainable vehicle.

And what I realized was because there is no objective feedback, because there's no loss, it's all social profit, they couldn't fail.

A group will never admit they were wrong.

#nonprofits#feedback#incentives#for-profit

Takeaway· 1

Takeaway

Where an ambitious young person should build wealth now

Human conquest was driven by competition for the same physical resources, but the first explorers were traders after spices, gold, silk, plants and animals. As technology raises the ceiling, wealth shifts to whoever has the best ideas — which is why Silicon Valley led as a wealth creation engine. The practical instruction is blunt: do not learn real estate, coal or oil extraction. Go into idea space — programming, books, movies, blogs, podcasts, and robots that are mostly intellectual property underneath.

  • Early explorers were traders first and conquerors only under resource constraints.
  • You do not roll tanks into a country for its natural resources anymore.
  • The wealthiest places are the ones with the best ideas.
  • All the new fortunes are being created in idea space.
  • Robots included — the value sits in the intellectual property underneath.

The new oil is ideas.

You go into idea space. You go into programming, books, movies, blogs, and podcasts, and building robots, which are mostly intellectual property underneath.

#wealth#ideas#careers#leverage