Michael Nielsen on Collaboration, Quantum Computing, and Civilization’s Fragility

Michael Nielsen — scientist, writer, and co-author of the field’s canonical textbook on quantum computing — joins Tyler Cowen for Episode 213 to range across the beauty of the universe, what makes a good collaborator, the honest status of quantum computing, the fragility of civilisation, and why non-legibility is not the enemy of knowledge but its penumbra.

Key ideas

1. Good collaborators are selected for complementary skills, not similarity. Nielsen’s theory of his own collaborators is that they share curiosity, openness, ambition, and prosociality — but each has at least one strong skill he lacks. Diversity of capability matters more than diversity of background, and younger mentors (people he actually learns from, not students who work for him) are the key to intellectual longevity in any field.
2. Quantum computing is genuinely hard but will almost certainly arrive. Nielsen stopped working on it in 2007 after fifteen years; he now watches from the side. With order-100-qubit systems (a qubit is a quantum bit — a unit of information that can represent both 0 and 1 simultaneously, unlike classical bits) that can apparently be manipulated at will, he is confident the field will eventually deliver. The most interesting near-term application is accelerating materials discovery through high-throughput quantum simulation.
3. Open science requires institutional change, not just access. Preprint culture in physics versus biology is a cultural contingency, not a structural inevitability — it took Paul Ginsparg recruiting Andy Strominger and Ed Witten to launch arXiv by social pressure. Openness per se is too weak a concept; what matters is the platform, the norms, and the kind of artefact made open. Jupyter Notebooks shared openly drive more progress than open journal articles because they are active materials, not static text.
4. Civilisational fragility is the deepest problem. Nielsen is more worried about biosafety and computer security than nuclear proliferation; the Nuclear Non-Proliferation Treaty provides some cartel control that will eventually erode. He agrees with Nick Bostrom’s vulnerable world hypothesis — that advancing technology will lower the cost of catastrophic attack — and sees near-universal surveillance as the uncomfortable corollary. The exit from fragility is getting a million people off Earth, not twenty.
5. Non-legibility and legibility grow together. Nielsen practises spaced repetition (a memorisation technique using timed review intervals to reinforce long-term retention) not to master a fixed body of knowledge but to expand what he can perceive at the edge. Making things legible — photographing, note-taking, memorising — does not reduce mystery; it pushes the penumbra of the illegible further out. Travel works the same way.

Content

On the beauty of the universe

The conversation opens with a question Nielsen says should have had a Metaculus prediction market: why is the universe beautiful to human eyes? He is drawn to but not satisfied by selection explanations. Physicists assert that simplicity tracks beauty, but nobody knows why simple rules should govern reality at all. Nielsen notes that visual beauty has a plausible evolutionary handle — we are programmed to find novelty attractive, and spiral galaxies register as novelty — but the same handle does not explain why the sounds of the universe leave people cold. Birdsong is beautiful, he concedes, but we are closely related to birds. The universe’s aural output, rendered as music, is ‘noticeably not particularly attractive.’

The James Webb Space Telescope, he argues, is one of the most expensive image factories ever built, which implies people are willing to pay for images of the universe even if they are not buying paintings of it. The most extraordinary single image he can name is the map of the 3-degree microwave background — a photograph of the universe as a whole, expressing structure at the scale of creation.

On collaborators and younger mentors

Nielsen spent the first half of his life as the youngest person in his circles — new fields like early quantum computing and open science were dominated by people with no sunk costs, who tended to be young. In San Francisco, surrounded by ambitious twenty-five-year-olds, he finds himself in the same position in his second half.

His theory of good collaborators names five traits: curiosity, openness to experience, prosociality, ambition, and imagination — but beyond those, he selects for at least one strong skill he does not himself have. The resulting collaborators are nearly always younger, and he regards them as mentors rather than as students. His empirical finding from studying older physicists is that those who aged well had younger mentors, not just younger employees. The mechanism is a network effect: if your circle is no longer at the frontier, that infects you; if your mentors are twenty-five and working in the newest ways, you get to partake of their momentum.

He describes his own style as ‘extremely disagreeable, but in a polite way’ — and observes that his younger mentors enjoy disagreement. People who feel threatened when you offer an alternative framing are not going to be good collaborators.

On quantum computing

Cowen notes that experts give wildly inconsistent accounts of quantum computing’s status. Nielsen’s answer is deliberately agnostic: he stepped off the field in 2007 after fifteen years on it, and is not confident in real-time assessments. What he can say: it is definitely not solved; it is an extremely difficult problem; order-100-qubit systems now exist and can apparently be manipulated at will, which suggests progress is real and the field just needs time. He refuses to commit to a year for the first useful result, though he thinks the most plausible near-term application is accelerating materials discovery — using quantum simulation to run experiments at the quantum scale that are currently prohibitively slow or systematically wrong.

On quantum money (the old Stephen Wiesner idea of currency whose quantum-mechanical properties make it uncounterfeitable), Nielsen thinks something like it may eventually arrive; he suspects quantum coherence will prove more stable than expected. He is sceptical of Roger Penrose’s theory that the human brain is a quantum computer — ‘the answer, unfortunately, is not really, no’ — but leaves room for unanticipated quantum structure in the brain that does not touch consciousness. He regards physics’ failure to break theory over the last fifty years as a structural problem: the theories have been too successful, so there is nothing to falsify, and quantum gravity remains stuck. One appealing feature of quantum computing is that pushing into genuinely novel physical regimes may finally provide something that breaks.

On open science

Nielsen was central to the open science movement and co-authored with Patrick Collison on the slowdown in scientific progress. He gives a structural account of preprint culture’s uneven spread: it is contingent on individuals, not fields. Ginsparg launched arXiv by going narrow — he recruited high-status physicists (Strominger, Witten) and built out from that social seed. The same mechanism was available to biology but did not fire. COVID accelerated biology’s shift, and Nielsen reads this as confirmation that it was always a cultural problem, not a structural one.

On whether openness can materially accelerate science, Nielsen is precise: the concept is too vague. Open journal articles do less than open code and data. Jupyter Notebooks — open, executable, reproducible — drive more progress than open PDFs because they are active rather than static. He thinks a 2× improvement in scientific pace from openness alone is not achievable, but a much larger gain is conceivable once norms and institutions adapt. He is puzzled by academics who defend $400-a-year textbook royalties while his own freely distributed neural networks book has benefited him far more — in reputation, opportunity, and indirect income — than any royalty stream would have.

On civilisational fragility

Cowen opens this section with the vulnerable world hypothesis — Nick Bostrom’s frame for technologies that lower the cost of catastrophic attack to the point where current governance cannot contain them. Nielsen sees biosafety and computer security as more immediate concerns than nuclear proliferation; the Nuclear Non-Proliferation Treaty provides a functioning cartel of fissile material that will eventually erode but is holding for now. Still, the direction of travel is not reassuring.

Near-universal surveillance follows, Nielsen thinks, almost necessarily: the ability to supply justice depends on understanding what has occurred. He would not press a button to ban universal surveillance permanently, because opaque systems are also exploitable by the powerful. The right frame is Madisonian — competing powerful institutions with strong checks — but he is not confident the United States is capable of this.

Getting civilisation off Earth is, for him, a genuine priority. He has Casey Handmer’s book on Mars colonisation in mind — ‘way too optimistic in many of its assumptions’ — but agrees on the scale: a million people, not twenty, before it counts as civilisational redundancy. Economic growth will reduce the relative cost over time, but the scientific unknowns (effects of Martian regolith on human biology, long-term impacts of low gravity, radiation above the Van Allen Belts) remain genuinely open.

On legibility, memory, and travel

Nielsen practises spaced repetition as one of several general-purpose attention strategies — not because memory is intrinsically valuable but because it causes him to pay closer attention. Taking photographs, taking notes, and committing things to memory are all versions of the same move: they install another way of perceiving. He frames the same logic as his reason for appearing on the podcast: Tyler Cowen will ask questions nobody else asks, which makes him pay attention differently.

He resists the tension Cowen sees between legibility and mystery. Making things legible — learning a culture, memorising a concept — does not reduce the illegible; it expands what you can see at the edge. The penumbra moves outward. Travel works the same way: the world has eight billion people, each of whom could teach you a great deal, and visiting places makes more of that depth accessible rather than closing it down. On great talent-identifiers, he notes they are idiosyncratic by necessity — the frontier of knowledge is not yet commoditised, and seeing it requires being near the edge oneself.

Related

• Michael Nielsen — guest
• Tyler Cowen — host
• David Deutsch on Multiple Worlds and Our Place in Them — also Conversations with Tyler; related themes of physics, reality, and civilisational thinking