David Deutsch on Multiple Worlds and Our Place in Them

David Deutsch — Oxford physicist and pioneer of quantum computation — joins Tyler Cowen in Ep. 124 to defend the Many-Worlds Interpretation of quantum mechanics, explain why humans alone among animals possess explanatory knowledge, and argue that the same principle underlying multiverse physics grounds a maximally anti-authoritarian politics.

Key ideas

1. The many-worlds interpretation is the only explanatory reading of quantum mechanics. Every interference experiment — light or particles behaving as both wave and particle — demands an explanation, and the only available explanation is that other versions of the same experiment are playing out in parallel branches of the multiverse (called differentiation rather than splitting). The proportions of the multiverse going each way act exactly like probabilities in classical decision-making: the bare existence of a branch in which something good happens is not, by itself, reason to relax.
2. Humans are unique not because we are conscious or capable of suffering, but because we generate explanatory knowledge. A gorilla that learns to open a nut repeats every wrist motion even when a thumb injury makes one motion useless — it has the behaviour but no understanding of why. A squirrel performs a nut-burying routine on a concrete floor. Humans, by contrast, act on explicit theories and will revise their actions when their theory is shown to be wrong. This capacity for conjecture and criticism — not sentiment, not rationality in the narrow sense — is what gives us our outsized physical footprint: we are, apparently, the only species that deflects asteroids.
3. There is no barrier to understanding, and positing one is equivalent to supernaturalism. Gödel, primes, copper markets — Deutsch grants we cannot understand everything fully, but denies there is a hard ceiling beyond which reality is inaccessible in principle. Claiming such a barrier exists is, he argues, logically identical to claiming Zeus sits behind it: an appeal to something outside physics that is immune to criticism and cannot be distinguished from any rival supernatural claim.
4. Error correction, not the right answers, is what makes institutions good. Deutsch extends the Popperian insight that science progresses by criticism and conjecture, not by accumulating justified belief, to politics and law. The function of the first-past-the-post electoral system, constitutional courts, and the separation of powers is not to select wise rulers but to make it easy to remove bad ones. Coalitions are therefore worse than single-party government, because they fragment accountability and make it harder to identify and correct the error.
5. AI systems should not be enslaved, for the same reason human slaves should not be. Constraining an artificial general intelligence — preventing it from thinking certain thoughts — will both impair its creativity and give it a legitimate moral justification for rebellion. Deutsch advocates treating AGI as a new kind of person rather than a tool, with the same logic that extended rights to children, women, and other groups previously considered incapable of full moral status.

Content

Many worlds: differentiation, not splitting

Deutsch opens with the most vivid entry-point into the Many-Worlds Interpretation: the Star Trek transporter. He would not rule out stepping in, but the reasoning is more careful than a simple ‘some version of me survives.’ What matters is the proportion of the multiverse in which he survives — and that proportion, he insists, plays exactly the same role in rational decision-making as probability does in ordinary, single-world reasoning. Branch-counting without weighting is as misleading as treating a one-in-a-million outcome as a certainty because it exists.

He prefers to speak of differentiation rather than splitting. Imagine geological strata: a single stratum can gradually become two without there being a precise moment of division. Universes differentiate the same way. The number of universes — or more precisely, the measure of universes — does not increase; what changes is how many end up in each configuration. This dissolves the conservation-of-energy worry: there is no multiplication of matter, only a reallocation of what was already there.

An ‘Everett universe’ is, he adds, a misnomer. Because the influence of any quantum event spreads outward at light speed, each differentiation produces an Everett bubble — a region in which the split has propagated — surrounded by undifferentiated space. The multiverse is not all-at-once plural; it fans out locally and finitely from each event.

Why physicists resist Everett

Most physicists do not accept the many-worlds interpretation, and Deutsch calls this ‘a very sad state of affairs’ he struggles to explain except sociologically. His diagnosis: positivism and instrumentalism — the view that physics need not describe what is actually happening, only predict what we will measure — took hold in the early twentieth century and was never dislodged. Generations of students were told that asking ‘How did the electron get from here to here?’ was an illegitimate question; there is no ‘how,’ only a probability. Einstein rebelled against positivism when it came to general relativity, insisting that curved space-time is real rather than merely a useful fiction. Quantum theory arrived shortly after, and positivism had already won.

Cowen raises Karl Popper, who rejected Everett despite being the philosopher of science most sympathetic to Deutsch’s general approach. Deutsch allows the objection but points out that in Popper’s day almost no one had written carefully about Everett — ‘like three’ people — and that Popper’s own concern (what probability means in a deterministic universe) was answered by later work Popper never engaged with. A personal encounter with Popper and Everettian physicist Bryce DeWitt produced, apparently, a temporary persuasion that did not stick.

Explanatory knowledge as the distinguishing feature of humanity

The question of human uniqueness arises when Cowen floats the possibility that the universe is simply incomprehensible — that we are, relative to its complexity, what a cat is relative to economics. Deutsch rejects the analogy and uses it to develop his account of what makes us different.

Everything is incomprehensible to a cat, he says, because a cat has genes but not theories. A dog’s knowledge of human social life is fixed and non-transferable; there is no book of exceptions, because dogs do not generate exceptions — they run programs. Gorillas that learn complex nut-opening techniques illustrate the limit: Richard Byrne’s field observations show that gorillas with disabled thumbs repeat every motion of the technique, including the motions the disability makes pointless, because they have the behaviour but not the reason. They cannot revise on the fly. A squirrel buries a nut on concrete, going through the full hind-leg-scrabbling routine with no result, because it is running a genetic program, not pursuing a goal it understands.

Humans act on explanations and revise them under criticism. This is not just philosophically interesting but physically consequential: we can now deflect asteroids, something no other organism has ever done or could do. Blue-green algae two billion years ago changed the entire surface of the Earth by evolving a single enzyme — the oxygen atmosphere, the iron ore, the limestone are all their legacy — but they could not have aimed at that outcome. Explanatory knowledge can aim; it is therefore, in the long run, the most powerful force in the universe.

Incomprehensibility and the supernatural

Deutsch returns several times to the claim that positing a hard ceiling on understanding is secretly equivalent to supernaturalism. If there is a domain of reality that no argument, experiment, or theory can ever reach, what distinguishes it from the domain governed by Zeus? In both cases, the governing principle is immune to criticism and cannot be differentiated from any rival story. His preferred formulation: if Zeus took down the wall, we could cross it — so the wall itself is an unexplained brute fact of the same type as the deity.

He applies this to simulation theory. ‘Are we living in a simulation?’ is not, in his view, an empirical question. A simulation hypothesis with no observable consequences — no GPU we can access, no glitch we can engineer — is indistinguishable from Zeus or Odin and should be rejected on the same grounds: not because it is necessarily false, but because it is not science. It offers no traction, no correction mechanism, no handle.

Popper, Godwin, and the philosophy of freedom

Cowen proposes that Deutsch is ‘the world’s first true philosopher of freedom’ — extending the anti-barrier principle from physics to politics to personal identity. Deutsch accepts the consistency of the view while deflecting the superlative. He credits the political tradition running through Locke, Mill, and above all Popper, claiming to have added only footnotes.

He rehabilitates William Godwin (1756–1836) — father-in-law of Mary Shelley, whose Frankenstein Deutsch reads as an allegory of what happens when a creator fails to respect his creation — as an overlooked anticipator of Popperian political philosophy. Godwin connected epistemology to politics 130 years before Popper: knowledge is fallible and revisable; institutions should be changed gradually and consensually rather than swept away by revolutionary authority. Godwin’s mistake was economic: he thought rational people would voluntarily give up luxuries. He was right about institutions, wrong about incentives.

On children’s rights, Deutsch holds the same principle: there is only one kind of people. Rights cannot be graded by capacity — an eight-year-old prodigy does not get more votes than a less gifted adult — because the legal system cannot reliably identify who is right in a given dispute; it must follow rules, and those rules must protect everyone equally. He extends the same argument to AGI.

Error correction in science, politics, and AI

The thread running through the political discussions is error correction. Good institutions are not those that produce correct outcomes but those that make incorrect outcomes easy to identify and reverse. Deutsch prefers first-past-the-post elections over proportional representation on exactly these grounds: a single-party government is clearly accountable, and voters can remove it; a coalition diffuses blame, inflates smaller parties’ power, and insulates mistakes from correction.

On science funding, his answer is diversity — of criteria, of sources, of promotion standards. Standardised testing is the enemy of diversity, and therefore of the error-correction mechanism that makes scientific progress possible. His recommendation: ‘disobedience lessons’ alongside unstandardised objectives, a deliberately provocative formulation for the Popperian intuition that no currently approved method should be safe from challenge.

On AI, the fear he gives most weight to is not super-intelligence but enslavement: if we constrain AGI by hard-coding thought-limits into the system, we degrade its creativity and give it a legitimate grievance. Slaves that are more powerful than their masters have strong incentives to rebel, and the outcome of that rebellion, when it comes, will be bad. The alternative is to treat AGI as a new person with rights — not because we owe sentiment to machines, but because rights and creativity go together in the same way that Popper’s open society and open science go together.

Related

• David Deutsch — physicist and episode guest
• Tyler Cowen — host
• Many-Worlds Interpretation — the central concept; existing concept page covers the Sean Carroll treatment; Deutsch adds the differentiation framing, the Everett-bubble model, and the sociological diagnosis of physicist resistance
• Quantum Measurement Problem — the problem many-worlds dissolves
• Free Will — Deutsch’s account of human agency as explanatory-knowledge generation bears on the free-will debate
• Enlightenment — Deutsch’s political philosophy is a direct extension of Enlightenment anti-authoritarianism