Ask Ethan: Could the CMB arise from galaxies, not the Big Bang?

Every once in a while, an idea comes along that turns our conventional understanding of the Universe on its head. We have our current understanding of the Universe: the laws that govern reality, the conditions and properties of the physical Universe, and the initial conditions that gave rise to the cosmos we observe today. But sometimes, surprises fall into our lap, and they cause us to re-evaluate everything we think we know. However, new information very, very rarely negates our former picture and conception of reality; it normally only refines and enriches it. For as much as we look forward to the possibility of major scientific revolutions, most improvements in our understanding are incremental, not an overthrow of everything we previously thought.

Still, there’s always something to watch out for: contrarians to the scientific mainstream, eager to undermine the currently robust scientific consensus, who prop up specious claims and build an easily-collapsible house of cards by making sensational allegations that are completely divorced from reality. Recently, a number of people, including Tom Tracy, Rob Chapman-Smith, and Jonathan Dugan have all contacted me about the same claim as seen from multiple different sources: that the early galaxies seen by JWST can contribute to some or even all of the energy traditionally attributed to the cosmic microwave background. As Jonathan skeptically inquired,

“Is this real enough to look interesting, or is this wild guesswork and cosmology norm-breaking/clout chasing to get attention? I.e., interesting or nah?”

We have to be careful not to fool ourselves, but that’s a knife that cuts both ways. Let’s take a look at this claim in detail: the background, the motivation, the underlying science, what’s been previously established, and then evaluate the full suite of what we know together. That’s how we’ll answer the only question a legitimate scientist should care about in the end, “What is actually true?”

COBE, the first CMB satellite, measured fluctuations to scales of 7º only. WMAP was able to measure resolutions down to 0.3° in five different frequency bands, with Planck measuring all the way down to just 5 arcminutes (0.07°) in nine different frequency bands in total. All of these space-based observatories detected the cosmic microwave background, confirming it was not an atmospheric phenomenon, and that it had a cosmic origin. The magnitude of the fluctuations within it are 1-part-in-30,000 only; not greater.

For background, let’s start by understanding that the cosmic microwave background radiation, or CMB for short, is taken to be the leftover radiation from a very early, hot, dense stage in cosmic history: specifically, the epoch when the primeval plasma — the ionized atomic nuclei and electrons and much, much greater numbers of energetic photons — expanded and cooled enough so that neutral atoms could form. When the nuclei and electrons combine to form atoms, the photons, previously in thermal equilibrium through their collisions with those charged particles, now stop bouncing off of other particles and instead free-stream in a straight line for all eternity.

Encoded in that radiation is the gravitational potential of whatever region they formed in. On average, the Universe started out uniform, with the same density and temperature everywhere. However, in some regions, there’s an overabundance of matter-and-radiation: an overdensity. In other regions, there are similar underdensities. When neutral atoms form:

• photons flowing out of average density regions have average temperatures, as they climb out of average-sized gravitational potential wells,
• photons flowing out of underdense regions have above-average temperatures, as they have less-than-average climbing to do,
• and photons flowing out of overdense regions have below-average temperatures, with greater-than-average climbing to do.

When we see the spectrum of acoustic peaks and valleys, and maps of overdense and underdense regions in the CMB, it’s this primordial information that gets encoded onto the radiation.

Regions of space that are slightly denser than average will create larger gravitational potential wells to climb out of, meaning the light arising from those regions appears colder by the time it arrives at our eyes. Vice versa, underdense regions will look like hot spots, while regions with perfectly average density will have perfectly average temperatures.

That’s the basic idea of the CMB. On average, it’s of uniform temperature and energy everywhere. Superimposed atop those “average” properties are fluctuations: gaussian (following a bell curve) in distribution, small in magnitude as seeded by inflation (at the 1-part-in-30,000 level, on average), and processed by the force of gravity and radiation pressure to produce the spectrum of fluctuations we see by the time neutral atoms form. Then, once that radiation is released and begins free-streaming, it’s only primarily affected by:

• the gravitational redshift of intervening structures as they form,
• any energetic boosts they receive from hot and/or moving gas that they encounter,
• and any foregrounds, like from intervening galaxies or our own Milky Way, that the CMB photons run into on their way to us.

For a nice layperson-friendly review of the CMB, you can take a look at this article I wrote back in January of 2025.

But what if there’s something else at play? What if the CMB isn’t what we think it is, and what if it doesn’t tell us the properties of the Universe the way we think it does? That’s primarily the idea of the underlying paper that’s cited by the various question-askers that have reached out to me this week. The paper, to be published in a special issue of the scientific journal Nuclear Physics B in August of 2025, is relatively simple. Instead of this conventional story, the authors contend, there’s an additional, previously overlooked foreground effect: the massive early galaxies found by JWST. And these galaxies, the authors contend, change everything.

This image shows a three-filter NIRCam view of galaxy MoM-z14: the new record holder (as of May 16, 2025) for the most distant galaxy ever discovered. Invisible at wavelengths below 1.8 microns, JWST has measured its spectrum and detected several emission lines, cementing its status as arising from when the Universe was a mere 282 million years old. This galaxy is bright and luminous, but is also compact; only ~10% of galaxies from the first ~550 million years of cosmic history have non-negligible amounts of dust.

They point to the fact that these early-type galaxies are massive, that they evolve and grow up quickly, and then they assert that the traditional picture of structure formation in the Universe — known as hierarchical mass assembly (where small-scale structures form first, then merge together to form larger ones — is incompatible with the galaxies that we see. They say that we see too many heavy elements that require too many generations of stars for the standard cosmological picture to hold, and that we need either additional non-standard ingredients (like intense and sustained star-formation rates and a top-heavy, galaxy-wide initial mass function of stars, where the average newly formed star back then is much more massive than the average newly formed star today is), or the Universe doesn’t make sense.

But then, they go on to say, based on that recognition, that there must have been these enormous clouds of matter-and-gas that must have collapsed to form those early-type galaxies. Those clouds must have been huge: more than two million light-years in diameter apiece. And these clouds of material must have formed very early on: when the Universe was between 180 and 270 million years old, or earlier than even the most distant galaxy ever detected by JWST is found. And then, spectacularly, they go on to claim that the energy from these early-type galaxies is an important source of CMB foreground contamination, representing anywhere from a low of 1.4% up to a possible high of 100% of the total CMB energy.

That’s the claim of the paper’s authors, and those claims are repeated and amplified in press releases and multiple videos.

This deep-field view of the Universe showcases a portion of the COSMOS-Web field acquired with JWST. In this field are a wide variety of galaxies, where the reddest, most dot-like galaxies represent some of the most distant, earliest galaxies ever seen. By examining a large number of galaxies, both dust-rich and dust-poor of a variety of masses, from the first ~1.5 billion years of cosmic history, we’ve been able to finally understand how galaxies form and grow up in our Universe.

But now, let’s take a look at the underlying science behind those claims.

First off, sure, these early-type galaxies do exist, and they were quite a surprise to cosmologists when we first found them, and tremendous effort has gone into understanding just what these galaxies are and where they came from over the past three years: since the JWST science era began in mid-2022. After all, initially, the abundance of these galaxies was something that puzzles most astronomers and astrophysicists, forcing us to ask, “Why are there so many of these bright, early-type galaxies, with more than 100 of them observed for every one we would have predicted beforehand?”

However, this isn’t the existential crisis many thought it would turn out to be. We’re now a full three years into the JWST era, and we’ve conducted an enormous amount of science on these galaxies. It turns out that a combination of four factors explains their abundance:

1. JWST was cleaner than expected, causing distant objects to appear brighter than anticipated,
2. structure-formation grows earliest in the rarest, most severe overdensities, something that wasn’t modeled properly in early simulations,
3. star-formation doesn’t happen in a continuous, steady, maximum-rate-obeying fashion, but rather occurs in bursts — often unsustainable bursts — that exceed the theoretical maximum rate, enhancing the brightness of these objects temporarily (i.e., when we observe them),
4. and many of these bright, early-type galaxies are enhanced by the activity of a central supermassive black hole, which doesn’t have anything to do with starlight at all.

Accounting for these four factors brings the Universe we observe back in line with what we expect.

This image shows 15 of the 341 hitherto identified “little red dot” galaxies discovered in the distant Universe by JWST. These galaxies all exhibit similar features, but only exist very early on in cosmic history; there are no known examples of such galaxies close by or at late times. All of them are quite massive, but some are compact while others are extended, and some show evidence for AGN activity while others do not. In general, the earlier galaxies are dust-poor and aren’t extended, while the later-type galaxies are more dust rich and more likely to be AGN-enhanced.

But there’s more to the story, too, as I wrote about in detail just a few days ago. Part of the explanation for what we’re seeing is that these early-type galaxies fall into two main varieties:

• galaxies with ~100 million stellar masses or fewer worth of stars, which are extremely low in dust and tend to be the “bursty star-formation” galaxies we see,
• and galaxies with ~300 million stellar masses or above worth of stars, which tend to be dust-rich and are often brightness-enhanced by an active supermassive black hole.

Moreover, that first population of galaxies represents — and hold your breath here, folks — nearly 90% of all JWST galaxies found from the first ~550 million years of cosmic history or earlier. Meanwhile, if we look at the next ~1 billion years of cosmic history, from 0.5 to 1.5 billion years of age, those low-dust, low-mass galaxies represent only about a quarter of the galaxies we see. We need to remember this; it’s going to come up again in just a bit.

So we have early galaxies, they do have stars, and those stars do inject energy into the Universe. But we next have to recognize that even though these early galaxies are, in fact, tremendously important sources of injected energy, that doesn’t necessarily mean they’re going to impact the CMB in a meaningful way. For starters, “energy” is not the same as “energy found at a particular set of wavelengths,” and this is a huge problem for light from these early-type galaxies. Starlight is mostly ultraviolet, visible light, and near-infrared light, but even at these very early times the authors consider, between a redshift of 15 and 20, that only corresponds to far-infrared wavelengths, or temperatures of around 40-60 K.

This Herschel image of the Eagle Nebula shows the heat-based emission of the intensely cold nebula’s gas and dust as only far-infrared views can capture. Each color shows a different temperature of dust, from around 10 K for the red, up to around 40 K for the blue. The Pillars of Creation, identifiable just below and to the left of center, are among the hottest parts of the nebula as revealed by these wavelengths. Note the extreme non-uniformity in temperature of the heated dust, showcasing its non-blackbody nature.

This is very important. It tells us that the starlight from these very early galaxies — from just 180-270 million years after the Big Bang, assumed to be there even though they’re beyond the limits of what JWST has already seen — could not be contributing to the CMB energy directly. Instead, that starlight would have to be:

• absorbed by some sort of uniformly-distributed matter,
• and then re-radiated with a perfectly blackbody spectrum at the appropriate temperature corresponding to the CMB’s exact temperature at that specific redshift/distance/epoch in cosmic history,

or it would fail to match the observations we already have in hand for the CMB.

There are three reasons why this is very, very difficult to admit as a viable scenario.

1. Starlight doesn’t follow a single blackbody spectrum, but rather is the sum of a great many blackbodies: both for each individual star and for aggregate populations of stars found together.
2. The matter that could absorb and then re-radiate starlight wouldn’t be of uniform temperature or uniformly distributed; it’s dust, and dust grains only get generated by galaxies with large enough stellar masses, which is only a small fraction of the galaxies produced at the earliest times.
3. And perhaps most importantly, we observe the CMB’s temperature fluctuations on a variety of angular scales: from large (~60-90° scales) down to scales of ~1 arc-minute, at minimum. But these individual early-type galaxies are less than 1/100th of an arc-minute, meaning they’re less than 1/10,000th of the angular area that the smallest resolved regions of space possess as far as the CMB is concerned.

The Sun’s actual light (yellow curve, left) versus a perfect blackbody (in gray), showing that the Sun is more of a series of blackbodies due to the thickness of its photosphere; at right is the actual perfect blackbody of the CMB as measured by the COBE satellite. Note that the “error bars” on the right are an astounding 400 sigma. The agreement between theory and observation here is historic, and the peak of the observed spectrum determines the leftover temperature of the cosmic microwave background: 2.73 K.

Let’s look at each of these more closely.

Above, you can see a graph for the Sun that shows its spectrum: the amount of emitted energy as a function of wavelength. Even ignoring the absorption lines because of different neutral atoms absorbing the overall starlight, you can clearly see that it doesn’t follow the same curve as a perfect blackbody: an ideally black object (a perfect absorber and perfect radiator) heated up to a certain temperature. This is because the Sun’s photosphere is partially transparent; it’s the sum of several blackbodies of different temperatures, from the coolest outer layers to the warmer, more inner layers. Stars also come in different temperatures and varieties; they do not radiate uniformly.

On the other hand, the CMB is the most perfect blackbody ever measured in the Universe. If it was in any way due to emitted-and-reflected starlight, we would’ve noticed deviations from this blackbody spectrum that are far in excess of what’s actually observed. This is actually a very old idea brought up by the Big Bang’s initial opponents: Fred Hoyle and Geoffrey Burbidge among them. For nearly the last 40 years of his life, Hoyle would go around giving talks questioning the CMB’s cosmic origin, claiming “we live in a fog” like it was some kind of great mystery. However, the science didn’t support Hoyle’s position, and his later-career research was entirely abandoned, rightfully, by the community upon his death. The CMB cannot be affected in any substantial way from early, reflected, or absorbed-and-re-emitted starlight.

This plot shows galaxies from the first ~1.5 billion years of cosmic history, color-coded by redshift and plotted by their metallicity (x-axis) as a function of the dust-to-stellar mass ratios (y-axis) found within them. The majority of low-metallicity galaxies are also dust-poor and are known as GELDAs, dominating the very early Universe, while later-time, more dust-rich galaxies are much more enriched in heavy elements. In the very early Universe, only very small pockets of space possess any meaningful amounts of dust; it requires sufficiently massive galaxies with enough stars to produce it.

The second point is about dust and its distribution in the Universe. In order for the energy from any early galaxy to be shifted into the range where it can mimic or contribute to the CMB, there needs to be matter capable of absorbing and re-emitting that energy. As the authors of the paper note themselves, there needs to be some (mysterious?) mechanism in place to distribute cosmic dust uniformly and to thermalize it: making it all radiate at the same exact (CMB-at-that-epoch) matching temperature.

But neither of those things reflect physical reality at all.

The cosmic dust that exists, as we’ve learned from viewing these early-type galaxies, only grows to become the type of dust grains that we’d need to absorb and re-radiate starlight — the type of dust that we find in the interstellar medium, for instance — in large, massive galaxies: galaxies with several hundred million solar masses worth of stars.

That dust isn’t uniformly distributed at all, but rather remains concentrated in and around those massive, early-type galaxies.

And, moreover, as we’ve learned from looking at heated dust in star-forming regions near and far, the way that dust gets heated by starlight occurs in a wildly non-uniform way, causing dust to radiate at a huge variety of temperatures, completely ruling out a blackbody description.

This graph shows the angular scales of CMB fluctuations as measured by Planck, ACT, and SPT down to the smallest angular scales ever probed: about 2 arc-minutes in angular scale. For contrast, the little red dot galaxies seen are all on sub-arc-second scales, more than 100 times smaller in angular size and 10,000 times smaller in angular area than the smallest measured scales of the CMB.

And finally, there’s the aspect of scale: of the angular size of the observed fluctuations in the CMB, versus the angular size of these early-type galaxies. The smallest angular scales we’ve observed of fluctuations in the CMB on come not from Planck (which only got down to scales of about ~4 arc-minutes) but from the Atacama Cosmology Telescope and the South Pole Telescope, soon to be surpassed by the Simons Observatory, which will get us all the way down to about 1 arc-minute. But these early-type galaxies are less than 1/10,000th of that area in terms of angular scale, and so they wouldn’t be able to contribute in any noticeable way to the actual CMB fluctuations we measure. The scales are all wrong.

Now, with all of this, we can finally draw our conclusions. The CMB truly is of cosmic origins, and the contentions of the paper that these massive early-type galaxies can play a substantial role is not borne out by the evidence. Early-type galaxies do produce energy, but in the wrong wavelength range and on the wrong angular scales to meaningfully contribute to the CMB. The idea that the entirety of the CMB could be explained by them is dead on arrival; we wouldn’t get a blackbody spectrum or the spectrum of fluctuations that we see. Even the basic concept — that these early-type galaxies are a meaningful CMB foreground — is highly suspect. Time and further research will tell, of course, but our question-asker’s speculation, that “is this wild guesswork and cosmology norm-breaking/clout chasing to get attention,” is a shoe that seems to fit much more aptly than anything the authors (or their promoters) have asserted.

Send in your Ask Ethan questions to startswithabang at gmail dot com!