Two Questions, Out Loud

May 17, 2026 · 18min read · updated May 18, 2026

The first time Jim Rutt mentioned the two questions, I was barely paying attention. He was interviewing someone — I don’t remember who, I don’t remember the episode — and the topic drifted, as topics on his show drift, into one of his standing obsessions. He said something like the Fermi paradox is one of the two questions I really can’t stop thinking about, and the guest nodded politely, and they moved on. I noted it as the kind of thing a smart American podcaster says, the way other smart American podcasters say things about cognitive biases or about Stoicism. Of course, I thought, Fermi. I was a nineties nerd the way you got to be a nineties nerd growing up in the interior of Brazil: TV Cultura and TV Escola via satellite dish (I was the only kid I knew who watched TV Escola outside of school, which wasn’t a matter of isolation — our gate stayed open all day and the yard was full of neighborhood kids all the time, I just sometimes preferred to be inside watching the show), Superinteressante magazine read cover-to-cover the afternoon it arrived in the mail (with particular affection for Luiz Barco’s column — a USP mathematician who also did math on public television, when didactic math on public television was a thing), dubbed reruns of Sagan’s Cosmos (the books I read in fragments, later), and a few hours a week on the forum of the Sociedade da Terra Redonda, running on phpBB, whose administrator was always asking for money to pay the server. STR was where things started getting too fringe to fit in Super — and it scratched the nerd itch much better than Super ever did. Fermi is furniture in my head; it has been there since I was twelve. Hearing it from Rutt was like hearing someone reach for an old standard at the piano. Pleasant, recognizable, no information.

Two weeks later he said it again, on a different episode with a different guest. The same two questions. Fermi, and a thing about minimum metaphysical commitments needed to do science, which sounded interesting in the moment and then I forgot. I noticed I was noticing. Some weeks after that, a third time. By the fifth time — I am not sure it was the fifth, but it was that order of magnitude — I had stopped finding it boring and was starting to find it strange.

Most intellectuals don’t do this. Most intellectuals are moved by the next shiny thing. You watch them across years and they’re always orbiting whatever conceptual fashion is in the air — complexity, AI, prediction markets, longevity, post-rationalism, fill in the blank. Their pivot questions change with the season. Last year I was thinking about X, this year I’m thinking about Y — that’s the structure, and it’s the structure for almost everyone, including me. Rutt has had two questions for over a decade. He brings them up on roughly half his episodes. He admits, on the air, that he brings them up too often. He keeps bringing them up. It’s giving monastic discipline, but secular — the practice is the practice, the practice does not get replaced because some new practice is trending on substack.

It dawned on me, somewhere around the fifth or seventh mention, that the consistency itself was a kind of argument. Not an argument for his particular questions, though if you stay with those questions for ten years they had better be good ones, and his are. The argument was about the structure of having two of them. You only get one life. You get to ask, what, a hundred big questions across a working lifetime, and most of those you’ll abandon halfway? Rutt picked two. He’s been working them for over a decade. He’ll be working them when he’s seventy. There’s a whole life shape to this that I had never seen modeled at close range before.

Which left me, slowly at first and then less slowly, with a question I had not particularly wanted: what are mine?

Not in a casual way — oh, what are some questions that interest me. Of course I have questions that interest me. I am a nerd, I have many questions, and I have nothing if not a long-standing reading list. The question Rutt’s repetition put to me was sharper than that. What are the two — give or take one — that you would be willing to fix for ten years? That you would mention on roughly half your conversations and not be embarrassed by your own consistency? That are worth narrowing your inventory to, in exchange for the depth you might get from staying with two instead of touring through forty? That is a different question and it does not have a flippant answer.

I have been ruminating on it for a few months. This post is what came out.

Rutt, briefly, for those who don’t know him#

Jim Rutt ran Network Solutions in the 90s and chaired the Santa Fe Institute in the 2010s. He is now retired in the way certain Americans of a certain generation are retired, which is to say he spends thirty hours a week running a long-form interview podcast where he reads the guest’s book, talks to them for two and a half hours, and asks better follow-ups than most academic peer reviewers manage. Gray beard, glasses, baseball cap, recording from somewhere in the Shenandoah Valley. The voice of a man who has explained the same thing to junior engineers many times and stopped pretending it was their fault for not getting it. He is allergic — allergic, in the medical sense — to lazy metaphysics. He has a running bit where he announces he is going to draw a pistol if anyone uses the word metaphysics without enough care. The pistol is imaginary. The bit is funny. He has been writing, over the last two years, a careful philosophical essay called A Minimum Viable Metaphysics, in which he tries to nail down exactly what commitments are necessary to do science. It is now on version 2.0. He has admitted in public that 3.0 is coming.

The pistol and the essay are not in tension; they are the same person. People who most hate the bad version of something are usually the people who end up writing the rigorous version.

His two questions#

The Fermi paradox you probably know in outline. Universe big. Universe old. Galaxy has a hundred billion stars. Even if a vanishingly small fraction produces intelligent life, the math says there should be a lot of intelligent life by now, and at least some of it should have left traces detectable across light-years. There are no such traces. The sky, as the astronomers say with admirable understatement, is silent.

Rutt got Fermi early — somewhere in the seventies, when he was a teenager, decades before Webb’s book existed (which would require time travel and a less minimum viable metaphysics than even he would tolerate). The book later gave the obsession a shape, but the obsession was already a tenant. He says on episode 258 with Webb himself: I have asked, at this point, probably half my guests about Fermi. And I just keep asking. I can’t stop. It is exactly that confession that made me start to take the obsession seriously. He knows it’s a tic and the tic, for him, is the practice.

The second one is younger and quieter. The Minimum Viable Metaphysics program is Rutt’s attempt to identify the smallest possible set of philosophical commitments a person needs in order to do science, in the engineering sense of do science — go to the lab, write a paper, predict a number, get the number within error bars of the measurement. Most professional philosophers, when confronted with this kind of question, want to give you the entire palace: presentism or eternalism on time, identity theory or property dualism on consciousness, structural realism or pragmatism on physical objects, and on for thirty other questions you didn’t know you were supposed to have opinions about. Rutt’s move is to say I don’t need any of that. He wants the minimum. The current version is four commitments: there is a reality, the universe started asymmetric, some part of nature is lawful, and time involves genuine becoming rather than just a clock-index inside a static block. Four. Get out of bed. Do physics.

I find both of these questions excellent. I want to say why, briefly, before saying why neither is mine.

Why they’re good questions#

The Fermi paradox is a probabilistic calibration exercise pretending to be an astronomy question. Every solution to it — the Great Filter behind us, the Great Filter ahead, the dark forest, the simulation, the rare-earth hypothesis, the zoo hypothesis, the they’re here and we don’t see them hypothesis — is a position on how to reason under massive uncertainty when your data set is one. Your prior matters. The evidence is sparse. You have to integrate over many parameters whose error bars are several orders of magnitude wide. That word is doing a lot of work when I say integrate: nobody knows how to do this honestly, and the whole field is, in effect, a community of smart people taking their best guesses out for a walk and seeing where they go.

The Minimum Viable Metaphysics is the same kind of question in a different costume. It’s an engineer’s question pretending to be a philosopher’s question. The engineer asks: what’s the smallest scaffolding I can put up and still hold the load? The philosopher’s traditional move is to keep adding things — every metaphysical position implies five more — until the whole edifice is too large to lift. Rutt’s move is the inverse. Tell me the four bricks. The rest is decoration. You may disagree about which four; you may want five, or three. But you’ve agreed to a game that has a chance of ending. Without that move, metaphysics is Sisyphean by design.

Both questions also reward the kind of person who actually thinks about them, instead of having opinions about them. Fermi rewards the calibrated. MVM rewards the disciplined minimalist. They’re the kind of questions that, ten years in, you’ve gotten somewhere with — not all the way, never all the way, but somewhere — and the somewhere is itself a feature.

So they’re good. They’re old, they’re well-trodden, and they’re good. None of this is news to a nerd. The news, for me, was not the questions; it was that someone had picked these two and was staying with them.

What took me longer#

It took months of intermittent rumination before I let myself write down what mine were. Months, for two reasons.

The first is that what are my two questions turns out to be a hard question to answer honestly. The easy version is what topics interest you most — for which I have a long, undisciplined answer that drifts across legal theory, philosophy of mind, complexity, probability, Brazilian institutional history, and at least three things I’d rather not list. The hard version is what questions would you fix and not abandon. Almost everything on the interest-list fails the not-abandon test. I read about something for a year and move on. The questions that actually survive that filter are smaller in number, and they are the ones I keep coming back to under the surface of whatever I’m reading at any given moment.

The second is that I was suspicious of the exercise. Picking two pivot questions sounded like the kind of LinkedIn-grade self-branding move that people make to seem deeper than they are. My two questions are: what is consciousness, and how can we build a more equitable society. No. That’s giving thought-leader. Rutt is not doing that — he genuinely cannot stop thinking about Fermi, he has been writing the metaphysics essay for two years, and his obsessions are confessional rather than performative. I had to make sure mine were too. The way I checked was simple: were these questions I’d been ruminating on for years already, without knowing they had a name? If yes, fine. If I was inventing them for the post, no.

The embarrassing thing is that they were not new. I had just failed to admit they were central. Written down as plainly as I can:

Are probability distributions real?

What is the best definition of reality?

Read them again. Both look smaller than they are. The first sounds like a question for an undergraduate statistics seminar; the second sounds like the kind of question one asks at a dinner party after the second bottle of wine. Their interest, for me, is in what each is hiding, and how each turns out to need the other to even get started.

The first one, on inspection#

The distribution question lands differently depending on which word you stress. Stress real and you have a tired ontological dispute about whether mathematical objects exist mind-independently — a fight going on since Plato that nobody has ever won decisively. Stress probability and you have a more interesting fight about whether probabilities describe the world (frequentist), describe our state of information (Bayesian), or describe an irreducible feature of nature (the question quantum mechanics keeps asking and refusing to answer). Stress distributions and you get the question I actually care about.

When we model some phenomenon as drawn from a distribution — heights from a Gaussian, particle decays from an exponential, queue waiting times from a Poisson — are we describing something about the distribution that is out there in the world, or are we using distributions as a bookkeeping convenience for our ignorance about underlying mechanisms that are, in fact, deterministic and individual?

I do not know the answer. Nobody knows. That’s the math not mathing. What I do know is that the question won’t leave me alone, and that the directions it pushes me — toward the foundations of probability, toward the interpretation of quantum mechanics, toward computational physics, toward whatever it is the Rosencrantz project is becoming — are the directions I find myself walking in when no one is watching. It is the place Rutt’s MVM 2.0 explicitly punts. He writes, in the Lawfulness section, “Both deterministic and stochastic interpretations of quantum mechanics remain viable. The experimental evidence does not yet distinguish between them conclusively.” He’s right not to commit; minimum-viable means leaving open what you can leave open. But the leaving open is exactly what I cannot put down. The frontier of his program is the start of mine.

The second one, which took longer#

The second one took three tries to write down honestly. I wrote it as what is reality? and crossed it out — too big, no traction, the kind of question a college freshman writes on a napkin and cannot read back later. I wrote it as is the world fundamentally physical? and crossed it out — old debate, well-trodden, nothing new I can add. I wrote it as what’s the relationship between mind and world? and crossed it out — same problem, slightly different costume.

What I actually wanted to ask, and what I finally wrote down when I let myself be honest, was this:

What is the best definition of reality?

Which sounds, on its face, even worse than the first three drafts. Wittgenstein would refuse the question — real is a word in ordinary use, it has its meaning from its use in language games, you don’t define it, you watch what work it does and leave it alone. I have some sympathy for that move. I also have a day job, where for almost twenty years I have made my living from the technical fact that words like good faith and public interest and reasonable and necessary do not, in fact, have stable meanings, and that my job is to negotiate, parecer by parecer, what they should be made to do in a particular concrete case. I am a working semantic engineer is one way to put it. Less generously, I’m paid to gaslight nouns into doing things they didn’t sign up for.

After enough years of this, when I sit down and look at a word like real — the most general, most overloaded, most contested term I encounter in any context — what I want to know is not what it means. I know it doesn’t have a single meaning. I want to know what job we would like it to do. Which is a Carnapian question more than a Wittgensteinian one — explication, Carnap called it, where you take a fuzzy ordinary-language concept and propose a sharper successor that does the same job better. Wittgenstein would withdraw permission and Carnap would hand it back. I am going with Carnap. The first question — are probabilities real — turns out to presuppose the second. You cannot answer whether a distribution is real until you know what real is for.

The shape of this#

Written down next to each other, the two questions do something I noticed only after I had put them on the page: they don’t sit in sequence. They argue.

The naive read is that one determines the other. You define real — that’s the second question — then you check whether distributions satisfy the definition. Clean order, second-then-first. But that is not what happens when I actually try to think about it. What happens is the reverse. I look at distributions doing absurd amounts of work — quantum mechanics, statistical thermodynamics, evolution, epidemiology, finance, machine learning, queuing theory, insurance, the whole inventory of things humans have built that actually work. That’s giving load-bearing math. And I look at serious people refusing to call them real. Frequentists who say a distribution is not an object, it is a limit of relative frequencies. De Finetti opening his treatise with probability does not exist. Non-realists about quantum mechanics insisting the wavefunction is a calculation tool, not furniture of the world.

The pressure that puts on the second question is not subtle. If a thing works that hard — predicts, calibrates, pays mortgages, vaccinates populations, builds reactors — and still fails the real filter, then what is the filter protecting? What does real refuse to grant, that distributions are doing without? Or, less charitably: is the definition of real we are using just too small?

The first question, by being empirically recalcitrant, forces the second to justify itself. The second determines the first, but the first also interrogates the second. The two questions don’t wait their turn. They argue with each other, and the argument is the content.

I don’t expect to answer either question. Neither does Rutt. The point of declaring pivot questions is not to be on a path toward answering them — it is to admit that you have them, that they organize your reading and your projects and your stray thoughts in the shower, and that the rest of what you do makes more sense if someone knows what they are. Here is where I have been stuck for years. Ask me how I’m doing on it. I’m not doing anything. I’m still stuck. I will be stuck on this at sixty. It is fine.

It is, in the end, a question about commitment. The world has too many interesting questions and one life is short. To pick two and stay with them is to refuse the rest. I have spent twenty years drifting from one fascinating-thing to another, and the drifting has not been wasted, but it has not produced what staying with two things for a decade would have produced. Rutt has Fermi and MVM. I have probability distributions and the definition of reality — each of us with one question about what’s out there and another about what we should require of any answer to the first. I am not going to figure out whether that is coincidence or whether pivot-question shape just tends to land there. The duality is the kind of thing that gets noticed and left alone.

What I’m sure of is that I should have written these down years ago.

For further reading#

• Jim Rutt, A Minimum Viable Metaphysics, v2.0 — the essay. Read this if you want to see what an engineer-philosopher’s commitments look like when written by someone who refuses to add anything that isn’t load-bearing.
• Stephen Webb, If the Universe Is Teeming With Aliens… Where Is Everybody? — the book that gave Rutt’s lifelong Fermi obsession a shape, and which lays out roughly seventy-five candidate solutions to the paradox. Even the bad ones are interesting.
• The Jim Rutt Show, EP 258 with Stephen Webb — the Fermi confessional. Rutt at his most open about the obsession.
• The Jim Rutt Show, EP 328 with Brendan Graham Dempsey — the metaphysics episode, where Dempsey interviews Rutt on his own program. The pistol joke is in there.
• Bruno de Finetti, Theory of Probability — for the subjectivist position on what probabilities are. De Finetti’s probability does not exist is the cleanest possible statement of the position my first question is trying to evaluate.
• Rudolf Carnap, Logical Foundations of Probability — for explication as a method. Read the introduction, skip the technical bulk; the introduction is what you want.
• Ludwig Wittgenstein, Philosophical Investigations — for the position Carnap is in tension with, on what it means to even ask what a word should mean.