Panel: Foundations of Quantum Science — Annotated Lecture Notes & Key Takeaways

These are my own notes from watching the session, reconstructed and paraphrased. They are not a transcript, and anything clumsy in the explanations below is mine, not the panel's.

I went into this expecting a definitional slog — four people arguing about what the word "foundations" even means in quantum science. What actually happened was more interesting: four origin stories, a running argument about whether quantum theory is finished business or barely begun, and a lot of loose ends that nobody tied up. The session ran about 44 minutes and was moderated by Peter Fisher, who heads MIT's Office of Research Computing and Data and used to run the physics department. It was billed as the closing session of a day-long event, and the format was roughly thirty minutes of moderated discussion followed by audience questions.
 

Fisher Sets a Loose Frame: Bridge Rules and Four Very Different Machines

Fisher opened by admitting the title was chosen to be vague enough that the panel could go wherever it wanted. His own working definition was decent, though: the irreducible ideas and rules of quantum mechanics, how we understand them now, and how we might come to learn them in the future. That third clause quietly disappeared for the rest of the hour, which I'll come back to.

The line that stuck with me came early (1:12). Fisher once told MIT's former provost that the rules of quantum mechanics are about as complicated as the rules of contract bridge — and that the interpretation of both is far more complicated than the rules themselves. It's a nice way to separate "can you compute with this" from "do you know what it means," and that split became the spine of the whole discussion. He also set expectations: the theory is our best one so far, it's been tested harder than anything else we have, and the hard part isn't the machinery.

Then he introduced the four panelists with a self-deprecating joke about lowering the average IQ on stage by 20% just by standing there. The roster: Moungi Bawendi in chemistry, who worked on quantum dots — Fisher described one as an almost one-dimensional quantum circuit; Peter Shor in mathematics, who invented the factoring algorithm that reframed cryptography and laid the groundwork for quantum error correction; Wolfgang Ketterle, who first saw gas atoms condense into a single quantum object; and Pablo Jarillo-Herrero, who found superconductivity in a system that is nothing more than two sheets of carbon atoms. Fisher's pitch was that each of these discoveries spawned a field, which is a fair claim and also a hint about where the panel was heading: toward platforms, not philosophy.
 

How Four People Actually Opened a Field

Pablo went first, and his story has a structure worth noticing because it's the opposite of the usual "theory predicted it" narrative. Graphene is a single layer of carbon atoms. Stack two layers with a slight rotational offset — 1.1 degrees, the so-called magic angle — and the combination superconducts, even though neither sheet does on its own. Here's the part I found genuinely interesting: the angle was predicted by theory, but the superconductivity was not. Theory got the geometry exactly right and completely missed the phenomenon sitting inside it. Seven years on, the trick (now called twistronics) has turned into a way to realize most of the phases of quantum matter that show up in solids. He positioned it between conventional quantum materials and ultracold atoms in optical lattices: less control than the cold-atom people have, but enough to observe a great many phases. I'd have liked an example of a phase that twisted bilayer graphene reveals and that the other two platforms can't touch — that comparison was asserted, not shown.

Shor's account is the one with the cleanest dramatic arc. He found in 1994 that a quantum computer could factor large numbers, which mattered enormously because internet security rests on that being hard. The pushback was immediate and, on its face, fatal: you can't measure a quantum computer without disturbing it, and you can't copy a quantum state, so how could you ever correct errors? He then worked out how to do quantum error correction, which turned "impossible in principle" into "an engineering problem." Fisher asked what he pictured a quantum computer to be while doing this work, and the answer was a bit of a swerve. Shor described starting from quantum Turing machines, realizing that physicists couldn't get their heads around them, and reformulating everything as a circuit model — a bundle of two-state qubits, gates made by bringing pairs together, and a program as a list of which pair interacts when. Physicists found that picture easy. The two models are equivalent in power, so nothing changed at the level of abstraction. What changed was who could work with it. That's a genuinely useful lesson about framing, though he never did answer what he personally visualized.

Ketterle's version was disarmingly blunt: he took atoms and made them very cold. As a young assistant professor he simply asked why they couldn't go colder, what the practical and theoretical limits were. Some colleagues thought it wasn't useful, and he admitted he hadn't fully worked out what it was for. The payoff came later, when Bose-Einstein condensates turned into a zero-entropy quantum substance that other people shaped into analog quantum simulators and even models for black holes. His broader claim is the most quotable idea of the session: quantum mechanics doesn't live in the abstract, it needs real platforms and real systems. He noted that people had dismissed BEC as trivial because superfluid helium was richer, and his response is that doing something already known inside a brand-new system can change the paradigm anyway. Same phase transition, same universality class — but a different-looking system makes people think differently, and you figure out what's new afterward. I buy the sentiment. It's also the least defended claim anyone made: no example was offered of a case where this actually happened beyond his own.

Bawendi asked a question that sounds simple and isn't: what happens when you take something intrinsically quantum, like an atom, and put a bunch of them together? Enough atoms and you get bulk matter — wood, silicon — whose behavior looks classical. An electron in an atom is a quantum object; an electron in a circuit is a particle moving around a wire. So at what count does the quantum-ness drain out? Answering that meant building things too small to be bulk, which put him right at the birth of nanoscience: hundreds or thousands of atoms, bigger than a molecule, smaller than a lump. He was candid that his career depended on making something well-defined enough to vary the size and watch properties change. The result was semiconductor quantum dots, where you can put in one electron, manipulate it, add more so they interact, and couple dots together. He was also the only panelist who talked about engineering constraints in the real world: biology lives in water and at room temperature, and getting inorganic quantum objects to interface with living systems is not trivial. He tied this to future quantum sensors for things like brainwaves.
 

Out of Equilibrium, Simulators, and a Slow-Motion Universe

Fisher steered the middle of the session toward quantum simulation, noting that Feynman thought it was the low-hanging fruit of quantum information science, and that the line between simulation and computation now looks blurry. Pablo picked up non-equilibrium physics as the promising frontier. In solids, everything couples to everything — electrons, phonons, magnons, photons — so isolating one degree of freedom is miserable, and dynamics are messy in a way that ground-state properties aren't. Simulators, he argued, give you exquisite control over parameters, mostly through lasers, so you can follow entangled many-body dynamics precisely.

Fisher then tried to connect this to Shor's world, asking whether quantum error correction could be viewed as an emergent many-body property. Shor's answer was a nice piece of intellectual history: Kitaev showed in the late 1990s that you could do error correction by simulating topological quantum materials, and the theory is beautiful — but essentially nobody building error-corrected machines uses it. They use more straightforward codes. The obstacle is that the topological route requires interacting more than two qubits at once. He sounded optimistic that progress on multi-qubit interactions might make it practical, but this is clearly an open bet rather than a plan.
The most memorable stretch of the session was Ketterle on why cold atoms are good for non-equilibrium work. His starting point is almost philosophical: nothing would ever happen if systems stayed at equilibrium, which he called entropy death. The problem is that in ordinary solids relaxation is absurdly fast, so you can't watch it happen. At nanokelvin temperatures, everything slows down — at (21:06) he offered the image of sound traveling a millimeter per second, so his voice would need an hour to reach the back of the room. With microsecond or millisecond resolution you can watch processes that would play out in femtoseconds in a solid, and you can even create negative-temperature states.

Fisher interrupted to say that "negative temperature" was freaking people out, and the clarification is worth having: you don't reach negative temperature by cooling past zero. You heat past infinite temperature, and the sign flips on the other side. Negative temperatures are hotter than positive ones, not colder. The panel never explained why anyone would want such a state, which was a small missed opportunity.

Bawendi then did what Fisher jokingly called a pivot that keeps one foot planted, and made the best point of the session in my view. Chemistry, unlike solid-state physics, approaches equilibrium slowly — and that's lucky, he said, because otherwise we'd all be dead. Large, strongly coupled systems in the presence of dissipation do something complicated that we don't have a vocabulary for. His proposal was that simulators could model arrays of coupled quantum dots and let us find out what happens, and crucially that their value isn't just producing results. It's giving theorists a way to think, building the kind of intuition we have for classical systems from everyday life and simply lack for quantum ones. I wish he'd given one concrete example of a two-body result that fails at many-body scale, because the claim "we understand two, we don't understand many" is exactly the kind of thing that lands harder with a case study.

Pablo extended this into what he called the key foundational question: how does quantum mechanics behave as you add particles, with all of them interacting? Condensed matter lives here all the time — unconventional superconductors, fractional Chern insulators, non-abelian statistics. His numbers were the useful part. Ten to the twenty-third particles is a mess, but a lot of emergent behavior already shows up at a hundred or a thousand, which is far beyond any classical computer and possibly within reach of quantum simulators in a few years to a decade. Note what just happened to the word "foundations": it quietly shifted from interpretation of the theory to computational intractability. Nobody flagged the shift.
 

Is Quantum Mechanics Understood? A Real Disagreement Nobody Resolved

Fisher asked for near-term unanswered questions, explicitly setting quantum gravity aside with a jab that their colleague Daniel just needs 100,000 qubits.

Bawendi answered with scaling and non-equilibrium, and was refreshingly honest about the state of play: his group creates systems out of equilibrium, watches them evolve, doesn't understand the results, publishes them with simple models that they know aren't really working, and needs better ones. He thinks the better models will eventually be built bottom-up, but you have to learn the effective interactions first so you can make the right approximations.

Then Ketterle drew a hard line. He wants to separate the physics of quantum mechanics from its philosophy. When people say nobody understands quantum mechanics, he rejects it outright — he does it every day and is in control of what he's doing. He went further and made it a civic argument: it's dangerous for physicists to claim they don't understand their own most-tested theory at a moment when science isn't always treated as guidance in public life, because it elevates speculation above testable, falsifiable hypotheses. Pressed, he conceded one genuine open question — whether something at macroscopic scale affects coherence in large superposition states.

Fisher agreed and invoked David Kaiser's essays on how the pendulum has swung between philosophical debate and what Kaiser calls the "shut up and calculate" regime, then emphasized that quantum mechanics has passed every test thrown at it. "You heard it here first," he added.

Shor said he seconded all of that, and then quietly undercut it in a way nobody acknowledged. His framing: quantum mechanics has no mystery at its foundation, but when you apply it to a well-understood theory, that theory stops being well understood. Three examples. The strong Church-Turing thesis — roughly, that no reasonable model of computation beats a Turing machine by an exponential factor — is now false (32:42). Classical channel capacities add when you combine two channels; quantum ones don't. Feedback doesn't increase the capacity of a classical channel, but it can for a quantum one. His closing question was whether there are more theories that quantum mechanics wrecks on contact.

That's a much sharper claim than "the theory is fine, the philosophy is hard," and it's in real tension with Ketterle's "fully understood and fully controlled." Ketterle's point is about the theory's internal consistency and predictive record; Shor's is about the damage done to neighboring fields. Both can be true, but the panel let them sit side by side without ever noting the friction — which is my single biggest criticism of the session.

Pablo rounded it out by aiming at biology: physics to quantum chemistry to quantum biology. His read on "nobody understands quantum mechanics" is that people mostly mean it's counterintuitive when applied to macroscopic objects and daily experience. He acknowledged that the chain from a hundred interacting particles up through chemical and biological systems to full-sized humans will take a very long time.
 

Theory Versus Experiment, Quantum Biology, and What Got Left Hanging

An anonymous audience member asked how much of the next decade's progress would come from theory versus materials, fabrication, and measurement. Ketterle said most discoveries are experimental because nature is more creative than we are — watching beats scribbling equations — and hazarded a ratio around 80/20 in favor of experiment (36:56). Pablo, an experimentalist, agreed in spirit but called it a partnership: theorists supply the intuition about what to try next and then help explain what came out, so the either/or framing is unfair. Shor answered as a computer scientist, pointing out that some algorithms were found by experimenters and took theorists decades to prove, and that without theory you're just trying things at random. Fisher closed the thread by invoking Thomas Kuhn and The Structure of Scientific Revolutions, calling him an unsung hero at MIT. Kuhn is many things; unsung is not one of them. The reference added flavor rather than resolution.

The last question — what is quantum biology? — produced the weakest stretch of the session. Ketterle said he finds coherent superpositions in aqueous solution hard to imagine, and that quantum biology usually means specific moments inside biological molecules where the physics is genuinely quantum, photosynthesis being the standard case when light is absorbed and an electron is released. Pablo added that the quantum/classical boundary is a moving target that keeps sliding as our measurements get more sensitive, since everything is quantum and "classical" is just a label for behavior we understood first.

Then Shor speculated that there are probably many fundamentally quantum processes in biology that haven't been found, mainly because biologists don't think to look, and that evolution is clever enough to exploit quantum where it helps. That may well be true. It's also a possibility claim with no evidence attached, and the jump from "one step in photosynthesis involves quantum mechanics" to "evolution probably uses quantum all over" happened without anyone building the bridge. As a listener, that's where I stopped nodding.

Bawendi got the last word and brought it back down to earth. Molecules are intrinsically quantum, stack enough of them and you get a human, and any biological process fast enough will be quantum. His example was lovely: we can detect single photons with our eyes, which makes the back of your eye a quantum sensor wired to a nerve signal, and he suspects we do a lot of quantum sensing without noticing. Fisher signed off with a joke about xenon isotopes in fruit flies and a suggestion that he might try it on his cat, which the panel talked him out of.

Two absences are worth naming. First, the measurement problem was gestured at repeatedly and never actually examined — it was mostly used as a foil for "we understand this fine." Second, Fisher's opening definition promised how we might learn quantum mechanics in the future, and nothing in the hour addressed teaching, intuition-building in students, or public understanding. For a session on foundations, that gap felt like the most foundational omission of all.

Still, as an hour of quantum science, it delivered something a textbook can't: four people describing how a field actually gets opened, mostly by building a machine nobody had built before and then figuring out what it was for.