The key implication of Cosmological Natural Selection

Probably the most important single insight I've ever had gets its own post, at last (Author: Julian Gough)

Probably best known for writing the ending to the video game Minecraft, Julian Gough is an award-winning Irish writer whose books, published in 36 languages, span serious literary novels, satire, science fiction, children’s books, and more recently cosmology. These essays are part of his experimental online non-fiction project The Egg and the Rock, which explores and extends the work of physicist Lee Smolin (and others), by applying the principles of Darwinian evolution to universes.

---

---

There is an extraordinarily important implication of Cosmological Natural Selection that I feel I haven’t made sufficiently clear in my writing so far. I MENTION it frequently, but always in passing; I’ve never stopped and shone a spotlight directly on it. But it answers three extremely important and fundamental questions, and so it needs its own standalone post; something I can simply direct people to, when they ask

• “Why do you think this theory is true?”

• “What does it imply?”

• “What does it predict?”

The failure to highlight this vital insight is partly because of the unusual circumstances of its birth: it only occurred to me in those frantic days, in June and early July of 2022, during which I was thinking through Cosmological Natural Selection from first principles. I had a hard deadline: I needed to generate, refine, and write up my predictions, then post them online, and email them to my subscribers, before the James Webb Space Telescope unveiled its first data on July 12th, 2022. That breakthrough insight became the basis of my (successful) predictions, but I didn’t have the time to explore it at length, and draw attention to it, in the way it deserved. It just became a hastily-written paragraph in a predictions post that was emailed to my 146 subscribers of the time.

This Substack now has over 10,000 subscribers, so 98.5% of them didn’t read that original breakthrough post. Yet it contains the Big Original Idea that led to Three-Stage Cosmological Natural Selection; the key insight from which all else follows. My successful predictions about the early universe emerge directly from it (and thus my grants and funding flow from it); the Blowtorch Theory of structure formation is built directly on top of it; the big piece on spiral galaxy formation that I’m currently working on starts with it; and, above all else, it’s the key piece of evidence that this theory may be true.

So, I’m going to, finally, separate it out here as a short(ish) post.

BACKGROUND: THE EARLY HISTORY OF COSMOLOGICAL NATURAL SELECTION

(Yes, I need to recap again for new subscribers – numbers have gone up a lot recently. And yes, if you know the history of Cosmological Natural Selection already, then click here to skip to the “implications” section.)

The brilliant, quirky American physicist John Wheeler (1911-2008) was co-author of the definitive textbook on gravity, and accidental co-inventor (along with a heckler at one of his talks) of the term “black hole”.

In the mid-1970s, Wheeler became fascinated by the fact that cosmology now had two mysteries involving singularities – points where the density of matter went off-the-scale, and where both of our best mathematical theories (general relativity and quantum mechanics) ceased to function, and started spitting out infinities.

One was a black hole, where mass/energy collapsed rapidly to a point, and mysteriously vanished from the bubble of space-time that is our universe.

The other was the Big Bang, where mass/energy emerged mysteriously from a point, and rapidly expanded to form the bubble of space-time that is our universe.

Wheeler pointed out that they looked awfully like the two sides of one process or event, joined at the singularity. One side simply mirrored the other, as though contraction had flipped or “bounced” at the singularity to become expansion.

(A BRIEF ASIDE: This, by the way, is not unprecedented in physics. For example, and to massively oversimplify: during the collapse of a star, the net effect of all the relevant forces flips, in a way that abruptly reverses the inward collapse of the star into the outward expansion of a supernova explosion. The forces themselves don’t flip. It’s just that their various values, and the way they interact at extreme densities, happen to be tuned to values that can generate this extraordinarily powerful reversal – which allows the periodic table of elements, generated by fusion deep in the heart of the star, to be distributed back up out of the star’s enormous gravity well and into the interstellar medium, where it can help build more stars, planets, and ultimately life. Yes, if you’re looking at our universe through an evolutionary lens, that looks awfully like the kind of fine tuning – to otherwise unlikely values, with functional consequences for reproductive success – that you would end up with after a long evolutionary process. But we are getting ahead of ourselves… END OF BRIEF ASIDE.)

As an elegant way to solve the two problems, Wheeler proposed that mass/energy in a parent universe collapsed to form a black hole (thus leaving the bubble of space-time that was the parent universe). It then “bounced” at the singularity, and expanded in a Big Bang, to form a new universe – budding off a new bubble of space-time, outside of, and separate from, its parent universe.

That is, parent universes reproduced through black holes, which “bounced” to form Big Bangs. Each Big Bang was therefore the birth of a new child universe, which grew up to produce more black holes; more child universes.

In the 1990s, another brilliant American theoretical physicist, Lee Smolin, saw a marvellous implication of Wheeler’s idea that had been completely missed by everybody (because theoretical physicists don’t usually read much evolutionary biology; Smolin had been reading some, for fun). Smolin realised that if there was slight variation in the basic parameters of matter in the child universe (rather than the large and random variation assumed by Wheeler), you would automatically get Darwinian evolution of universes. That’s because slight variation allows for inheritance; and some of those slight variations would lead to more black hole production; more reproductive success; more offspring. Those successful variations would therefore be inherited more widely than less successful variations. The successful variations would vary again (slightly) in the next generation, with some variants being even more reproductively successful (and some less); and off we go – Darwinian evolution of universes, with the evolutionary ratchet optimizing for reproductive success. Over time, the majority of universes will come to be fine-tuned for very high levels of black hole production (compared to the evolutionary starting point, where the numbers could be as low as one or two). This seems plausible: our universe has already produced over forty quintillion black holes, just from stellar collapse. (That’s 4×10¹⁹; a four and 19 zeros! For how they worked that out, see Sicilia, Lapi, et al, 2022.) Smolin’s simple and straightforward evolutionary theory of universes became known as Cosmological Natural Selection (CNS), laid out in this paper, and this book.

So, Smolin had uncovered a powerful evolutionary implication, hidden inside Wheeler’s simple model of the reproduction of universes.

But there is a further, huge, implication hidden inside Smolin’s new and improved theory. Lee Smolin didn’t see it at the time because he was a theoretical physicist, not an evolutionary biologist, and so not saturated in the logic of evolution (from which the implication emerges). He was also handicapped by the fact that astronomy had only observed half a dozen or so supermassive black holes at the time he wrote his original paper and book. It was still therefore blithely assumed, by astronomers and cosmologists and theoretical physicists alike, that supermassive black holes must simply be a bunch of (much smaller) stellar-collapse black holes merged together. It just hadn’t occurred to anybody that black holes that large might have a separate, far simpler, formation mechanism.

But that’s exactly what an evolutionary theory of universe implies. This is an extremely important point, but people usually skip over it when I make it, because it’s usually embedded in a more complex post, alongside a lot of other new information; and so it just whizzes past, as one point among many. This time, I’m going to walk through the logic in excruciating detail. And when I’m done, don’t just click through to the next thing in your scroll. Take a moment, and sit with it. Think it through. Test it in your mind.

And if, at the end of all that, you think it’s true, internalise it. Bring it to the conversation around these issues. Because if it’s true, the boundaries of various scientific fields are currently in the wrong place to explore this new knowledge. A lot of things will need to change. And you can help change them.

Photo of bubbles by Kind and Curious on Unsplash. Yes, I was looking for a simple (OK, crude) visual metaphor for Cosmological Natural Selection, and universes as separate bubbles of spacetime. No, it wouldn’t actually look like this. Nice picture though.

COSMOLOGICAL NATURAL SELECTION’S KEY IMPLICATION, AND THUS PREDICTION

The original reproductive mechanism for universes can’t have been stellar-collapse black holes, because stars are complex, orderly structures made of complex, orderly matter; and all the complex, orderly systems we know of are the result of evolution. (Just as the evolutionary history of biological life can’t have started with complex, orderly eukaryotic cells, with their sophisticated, interacting organelles; just as the evolutionary history of communications devices can’t start with the iPhone 17 Pro Max. ) Looked at through this lens, the periodic table itself, with its dozens and dozens of ever more complex elements, clearly must be the result of an evolutionary process. (It is highly suggestive that the elements get more ragged and unstable towards the upper, more complex, and thus more recently evolved, end.) The original ur-matter, in the original ur-universes, however, must have been as simple as it gets. Very pure, very simple matter, with no structure, and building no structure.

Small, efficient, stellar collapse black holes, therefore, generated at the end of a star’s rich and complex lifecycle, are – must be – a later evolutionary breakthrough.

So, think this through from first principles. If our universe is the result of a Darwinian evolutionary process, then every single one of the earlier universes in our direct evolutionary line must have reproduced successfully, by definition.

But the only thing that is unavoidably, unsimplifiably required to make a black hole (to reproduce) is for mass/energy to collapse.

The original reproductive mechanism, therefore, must have been the direct collapse of extremely primitive and unstructured matter to form small numbers of large, crude, and therefore supermassive black holes. Far more massive, and therefore far less numerous, than the forty quintillion stellar-collapse black holes we already have so far in our specific universe. (That’s unavoidable, because the more massive a black hole is, the larger the percentage of its universe it must take up, and therefore the less of them there can be.) Evolution will thus blindly drive towards the ever-more efficient production of ever-smaller black holes - with each still able to produce a full-sized universe, thanks to the fact that the positive mass energy and negative gravitational energy in a universe net out to zero, and so full-sized universes can be built for free.

But we have discovered, in the three decades since Lee Smolin published the first, wonderful, tentative (and tragically overlooked) version of Cosmological Natural Selection, that our specific universe in fact contains roughly a trillion supermassive black holes – one at the centre of every galaxy. (It sounds like a lot, but divide that trillion into forty quintillion: there’s just one supermassive black hole for every forty million stellar collapse black holes to date.) However, though relatively few in number, those supermassive black holes are often millions or even billions of times more massive than the average stellar-collapse black hole. They must, therefore, have formed by that original reproductive mechanism; direct collapse of extremely large amounts of primitive and unstructured matter.

Why? Because evolution is frugal; she doesn’t bother to come up with a much more complicated way of doing something for which she has already come up with a perfectly good mechanism, unless that innovation leads to far greater reproductive success. Yes, evolution – blindly exploring the possibility space for the basic parameters of matter, through generation after generation of universe – eventually came up with more complex forms of matter, and thus stars, and stellar-collapse black holes, and conserved that breakthrough, because they were a huge improvement on a small number of immense direct-collapse supermassive black holes. With stars, you could get millions or even billions of stellar-collapse black holes / offspring / child universes, out of the same mass that once (in earlier generations) produced just a single supermassive black hole, and thus a single child universe.

Yet supermassive black holes clearly still have a function in our universe, given that there’s one at the center of every galaxy; given that they have been conserved by evolution. (And I explore that function in my first Blowtorch Theory post.) They are clearly still required, to help build out the larger, later, more complex structures (just as, say, complex contemporary multicellular organisms still require the presence of the primitive, ancient mitochondria they long ago engulfed, in order to build and assemble themselves). If supermassive black holes have been conserved by evolution, then their formation mechanism, direct collapse, must also have been conserved. (Similarly, mitochondria still reproduce by splitting themselves in two, as they always did – in a cycle that is not synchronised with the reproduction of their host cell.)

Meanwhile, how did mainstream cosmology think supermassive black holes formed, before the James Webb Space Telescope data? Well, they had half a dozen theories, which means they had no theory. This is the understandable and inevitable consequence of having no meta-theory; no helpful and constraining framework (such as evolution) capable of extracting meaning from data, and imposing meaning (or, if you prefer, discipline) on hypotheses. Such a meta-theory gives you something against which the real-world likelihood of hypotheses could be tested. In the absence of such a meta-theory, the only test cosmologists could apply to their ideas was (and is), are they mathematically possible (that is, without mathematical contradictions). But there are countless mathematically possible formation mechanisms, especially when dark matter is added as an optional, and tuneable, ingredient.

(ANOTHER BRIEF ASIDE: Quantum mechanics, incidentally, is in a similar pickle, which is why theoretical physicists, over the past century, have come up with SO MANY utterly wild, purely mathematical theories, attempting to advance the field, without ever actually getting anywhere. As Adam Forrest Kay puts it, in his superb book Escape from Shadow Physics,

“…without pictures of hidden reality, the only guidance is mathematical rigor. This makes the search space explode, because then all steps that do not lead immediately to contradiction are on equal footing.”

– Adam Forrest Kay, Escape From Shadow Physics: Quantum Theory, Quantum Reality and the Next Scientific Revolution

But that’s a whole other story… END OF ANOTHER BRIEF ASIDE.)

And so, in the decades before the James Webb, mainstream cosmology kept coming up with exciting new hypotheses for supermassive black hole formation.

DARK MATTER MINI-HALOS GENERATE POPULATION III STARS!

The most widely accepted theory was that the seeds for supermassive black holes were formed by “Population III” stars, the annoying term for the very first stars, made of only hydrogen and helium, which were believed to be unusually massive and shortlived: after one burned out and collapsed to form a large black hole (of maybe a hundred solar masses), it would continue to swallow a huge amount of gas and grow fast. And maybe merge with other Population III black holes, if that was still necessary to get the mass right (i.e., to keep the maths from becoming contradictory). As nobody had ever seen a Population III star, or knew when they first formed (or, indeed, IF they formed), you had a lot of mathematical wriggle room with this one. Plus, oh yes, you needed a dark matter mini-halo to form them. More mathematically wiggly fun.

DENSE STAR CLUSTER MERGER MANIA!

Another popular idea was that, in dense clusters of stars, many massive stars would sink to the center, collide, and merge to build an extremely massive star that would quickly collapse to form a black hole with a mass of between a thousand and a hundred thousand solar masses. (That’s an IMBH [Intermediate Mass Black Hole], with a mass of 10³–10⁵ M☉, if you speak maths and acronym.) Several of those intermediate mass black holes would then merge, and also accrete more gas, to finally form a supermassive black hole.

PRIMORDIAL BLACK HOLES FORM IN THE FIRST SECOND!

Another option was primordial black holes: black holes which might form in the absurdly early universe, like, the first second after the Big Bang.

(A BRIEF ASIDE FOR THE MATHS BROS, SKIP IF YOU HATE MATHS: “The first second” is, in fact, surprisingly, a fairly precise term here. When radiation dominates – i.e. in the immediate aftermath of the Big Bang – the horizon mass scales ≈ 10⁵ M☉ × (t/1 s). Primordial black holes forming at t ≈ 1 s naturally sit at ~10⁵ M☉, which would make a nice, heavy seed for rapid growth into a supermassive black hole. (Earlier times make for much smaller primordial black holes; e.g., t ≈ 10⁻⁵ s → ~1 M☉.) END OF BRIEF ASIDE FOR THE MATHS BROS.)

How did these primordial black holes theoretically form? Through mechanisms for which we had no evidence, but which – given how little we know about the Big Bang itself, and how much you are therefore free to make up – you could easily make mathematically non-contradictory. Formed that early, they had a looooong time to drink gas and grow large. Also, they could merge, why not.

SELF-INTERACTING DARK MATTER GRAVOTHERMALLY COLLAPSES!

Want more? Self-interacting dark matter could undergo “gravothermal collapse” (don’t ask). I mean, it’s self-interacting dark matter! It can do anything it likes. Mathematically speaking, it could probably bake you cookies and bring them to you in bed. So, gravothermal collapse, why not.

MIGRATION-TRAP PHYSICS!

Or how about migration-trap physics – an idea borrowed from the theories used to explain how planets form in protoplanetary discs. The idea was that many small stellar-collapse black holes would get embedded in gas, which would drag them closer, pair them up, merge them, merge that merged pair with other merged pairs… and eventually you end up with a supermassive black hole.

You will note that many of these approaches to supermassive black hole formation put forward the same kind of passive, bottom-up, hierarchical, merger-driven process that was assumed for galaxy formation. This reveals the unspoken meta-theory under which cosmology was operating: it’s all arbitrary and random and means nothing. Any large complex, organised structure is staggered into, almost accidentally, through a random walk, by matter with arbitrary characteristics.

DARK MATTER ANNIHILATION HEATS THE FIRST PROTOSTARS!

Oh, but we are not finished. How about dark-star seeds! The big idea here was that dark-matter annihilation (don’t ask) would heat up the first protostars. As we know nothing at all about dark matter, you had plenty of room to make this mathematically non-contradictory.

Was there a clear winner from all this? No.

In fact, by the 2020s, you had comprehensive reviews like Inayoshi, Visbal & Haiman’s The Assembly of the First Massive Black Holes, in 2020, arguing for a multichannel model, where you had seeds forming from Population III stars, clusters smashing lots of stellar collapse black holes together, some primordial black holes, and maybe a few dark-matter-driven dark stars.

In other words, no theory. (Because, no meta-theory. No frame. Or rather, a nihilistic, meaning-denying meta-theory that couldn’t help you decide between options.)

But here’s something that makes the whole situation even more interesting, and revealing; in the years since Smolin’s original paper, Did the Universe Evolve?, some gutsy astronomers and astrophysicists – Martin Rees, Avi Loeb, Priya Natarajan, Volker Bromm, Marta Volonteri, and others – did work out that the formation of supermassive black holes by direct collapse was technically possible in our universe. They published lots of papers saying so, like this one, and this one, and this one…

However, those intrigued by direct collapse were outnumbered by those proposing all the formation mechanisms laid out above.

So direct collapse was lying there, on the table, as an option, since the 1990s. Why wasn’t it embraced?

Because cosmologists were stuck inside the old paradigm, the old meta-theory, the old frame-story: that our universe is a random one-off, with arbitrary properties, self-assembling slowly and passively and basically randomly – and certainly not fine-tuned by evolution for reproductive success through black hole production. Their meta-theory, their frame-story, therefore pushed them towards bad ideas, and away from good ones.

Three-stage cosmological natural selection, however (the key theory in the field I am starting to call Evolutionary Cosmology more generally) – provides a framework that allows you to simply and cleanly pick the killer formation mechanism – direct collapse – out of the police lineup (or identification parade, if you are in the UK or Ireland) of formation mechanism suspects.

That’s because if supermassive black holes – the earliest and most primitive structures, through which the earliest universes reproduced – are still to be found in our universe, then their production in our universe should precede that of any complex structures that evolved far later. The simple, original, reproductive mechanism shouldn’t depend on – and certainly couldn’t emerge from – later, more complex structures. That’s simply the unavoidable logic of an evolutionary history. And so, direct-collapse supermassive black holes, which must have been the earliest form of reproduction for the most primitive universes, with no structures required for reproduction, will come into existence in our specific universe before any complex structures. They will not be generated by more complex structures, and they are highly unlikely to be dependent on any more complex structures that only emerged later in the evolutionary history of universes. That’s just an unavoidable piece of evolutionary logic.

(Yes, it is possible that evolution has enriched and complexified the original mechanism of reproduction so thoroughly that it is now dependent on structures that evolved later. But the fact that there is an original mechanism of reproduction, and it is simple direct collapse, moves the odds firmly against that.)

This lets you dismiss all those formation mechanisms which rely on (early) star formation to drive (later) supermassive black hole formation. It simply cannot happen in that order. Likewise, no bottom-up, multistep, hierarchical merger model can work to explain supermassive black holes. The big primary thing can’t be built out of little secondary things. (Once formed, a supermassive black hole can later be fed stars and elephants and iPhones, sure. But it can’t be initially formed by them.)

Similarly, primordial black holes (the only other formation mechanism to make it through that filter) fails, because it requires a whole bunch of complicated (fine-tuned) tweaks to the conditions surrounding the Big Bang in order to work. And such fine-tuned tweaks can’t have PRECEDED the earliest reproduction of universes (as would be required if they ENABLE that reproduction); such fine-tuning could only be the result of later evolutionary pressure. Meanwhile, a simple, smooth Big Bang doesn’t give you – can’t give you – primordial black holes. So, that mechanism, too, fails the Evolutionary Cosmology test.

DO YOU SEE THE POWER OF THIS?

Right now, in cosmology, we don’t have a way to discriminate between mathematically plausible theories. But three-stage cosmological natural selection (and indeed the entire nascent field of Evolutionary Cosmology) gives you a way to discriminate between mathematically plausible theories, because you can do a second check – after “is it mathematically plausible?” – which is, “is it evolutionarily plausible?”; or, more precisely, “is it compatible with the evolutionary history of our universe, given that our universe reproduces in this fashion?”

This is how, back in 2022, I was able to out-predict the entire field of cosmology. Not because I’m cleverer, or more hardworking, or have any particular virtues as a scientist or thinker. But simply because I was working with, I was extending, a better meta-theory. A better framework.

It’s very, very telling that the first use of this principle, when predicting what the James Webb would see, gave such a strong positive result.

And so I think the keystone prediction of Cosmological Natural Selection, when applied to our specific universe and its development, is this: supermassive black holes should be presumed to form first, from smooth gas, by direct collapse, before galaxy formation. They should not be presumed to depend for their formation on any more complex structures, such as stars, or galaxies, which – doing far more complex things, with far more complex matter – must have evolved later in the evolutionary history of universes.

And the supermassive black holes found in our specific universe today therefore are most likely to have formed when conditions in our specific universe most resembled those found in the earliest, most primitive universes; just after the Big Bang, and before star and galaxy formation; so, well inside the first couple of hundred million years. Back when the gas was still smooth enough for direct collapses of large areas, without small local density fluctuations, and thus without breaking up into stars.

This insight, this prediction, emerges purely from evolutionary logic, not from mathematical reasoning or from physics. (Though, of course, it obeys all physical laws, and requires no new particles or physics. It’s a surprisingly conservative theory, simply applying Darwin to universes.) It was first made before the James Webb Space Telescope had sent back any data. And it appears to describe, remarkably accurately, what the James Webb is seeing in the early universe. See, for example, this astonishing paper from last year, describing a huge population of extremely early galaxies, or protogalaxies, dominated by their central supermassive black holes:

“Little Red Dots: An Abundant Population of Faint Active Galactic Nuclei at z ∼ 5 Revealed by the EIGER and FRESCO JWST Surveys”, by Matthee, Naidu, et al, 2024.

Or this great paper on UHZ1, the supermassive black hole that weighs as much as all the stars in its galaxy put together:

First Detection of an Over-Massive Black Hole Galaxy UHZ1: Evidence for Heavy Black Hole Seed Formation from Direct Collapse, by Priya Natarajan (hurray! take that victory lap!), Fabio Pacucci, et al, 2023.

Or this paper on an even larger, sleepy, early, supermassive black hole:

A dormant overmassive black hole in the early Universe, by Ignas Juodžbalis, Roberto Maiolino, William M. Baker et al, 2024.

And so on, and on.

WE ARE ALL ON THE SAME TEAM

Three-Stage Cosmological Natural Selection gives a wider conceptual framework for the pioneering work done on direct collapse black holes over the years by a handful of brave cosmologists and astronomers, and it explains why their colleagues turned out to be wrong to think of direct collapse as just one possible formation channel among many (with various ways of merging stellar collapse black holes as the leading candidates): direct collapse will not turn out to be a rare, unlikely event (even though Wikipedia still says “Direct collapse black holes are generally thought to be extremely rare objects in the high-redshift Universe”); direct collapse will turn out to be ubiquitous, because evolution has fine-tuned our universe to enable it.

The two approaches, those of the pioneers of direct collapse and mine, turn out to be totally complementary. We can only help each other. This is a delightful win/win situation, where the new paradigm doesn’t need to smash the old paradigm; it can just help the old paradigm solve all its problems. Reframe, and explain. We get to keep all the old data, gathered under the old paradigm; we will just understand it better now.

Let’s end with last month’s article in Quanta, scrambling to explain a recently discovered “naked” supermassive black hole, in the early universe, that doesn’t seem to have a galaxy around it yet.

“This new black hole, which is as heavy as 50 million suns and is dubbed QSO1, clashes with the old, provisional account of the galaxy formation process, which did not start with black holes. Black holes were thought to have come along only after a galaxy’s stars gravitationally collapsed into black holes that then merged and grew. But Maiolino and his colleagues described a solitary leviathan with no parent galaxy in sight.
The question now is how this black hole came to exist.”

– Quanta Magazine, “A Single, ‘Naked’ Black Hole Rewrites the History of the Universe”, by Charlie Wood, September 12th 2025.

Obviously, I think I know the answer. (Having predicted exactly what they are seeing now back in 2022.)

It is amusing to see that, three years into the new James Webb Space Telescope era – after three years of discovering ever-greater domination of galaxy formation by ever-larger, ever-earlier supermassive black holes – so many cosmologists, astronomers, and science journalists remain oblivious to the fact that there is a perfectly respectable and conservative theory that predicted all this...

OK, that was fun. Glad I finally got this piece written.

Now, ponder it. Stress-test the logic. See if you think it’s true. And if it is, internalise it. Then go read Blowtorch Theory, if you haven’t already, for more – much much more – information on this evolutionary approach. Or watch this video, if video is your thing.

And I’ll go back to writing my epic spiral galaxy formation piece.

---

---