At NASA, I rarely saw the cosmic microwave background in the oval projection printed in textbooks. My screen usually held directories of files, pixelized sky maps, masks over the Milky Way, simulated skies, and plots of angular power against the multipole number ℓ. I compared one curve with another. When they separated, the difference had several possible owners: the sky, the instrument, foreground dust, or the code that put the curve there. I learned to distrust a feature in stages.

The cosmic microwave background, or CMB, is often called the baby picture of the universe. The phrase refers to a precise event. About 380,000 years after the Big Bang, expansion had cooled the primordial plasma enough for electrons and nuclei to bind. Photons stopped scattering continually from free electrons and began to travel over cosmological distances. They reach us now as microwaves with an apparent temperature of 2.725 kelvin.1

Across the sky, that temperature varies by only a few parts in 105. The blue and red patches in published maps are a color scale applied to differences too small for human sight. The map records a statistical field. Slightly denser regions drew in more matter, which made them denser still, and their pattern connects the smooth microwave sky to the galaxies, clusters, and voids around us.

The word baby supplies more than scale. When we look at a child, we already know the general shape of the adult and the sequence that produces it. The metaphor carries that confidence into an interval for which we had very little evidence. An early galaxy begins to sound like a present-day galaxy with less of everything: less mass, less structure, fewer heavy elements, and a smaller black hole, if it has one at all.

No halo mass function contains a juvenile term. The assumption entered through the practical work of turning dark-matter structure into observable light. Relations measured in later galaxies were extended backward. Star-formation efficiencies, mass-to-light ratios, dust prescriptions, feedback models, and black-hole scaling relations filled the space where observations were missing. Many early systems were treated as scaled-down versions of populations we knew better. Webb now observes the redshifts at which those priors had the least evidence beneath them.

The early universe was hotter, smoother, and chemically simpler than the universe is now. Its age, however, did not assign the same pace to every process inside it. A redshift gives the time available. It does not say how many times a dense cloud can collapse, how quickly a massive star can die, or how rapidly a black hole can feed. We had confused duration with consequence.

At the University of Maryland, where I worked on JWST spectroscopy for quantum astrochemistry, a line list looked like a ledger. Each row joined a wavelength to an upper state, a lower state, a transition strength, and an uncertainty. The distances in the project were astronomical. The daily work could be almost clerical: matching wavelengths, checking state assignments, and watching a feature split or disappear when the temperature or resolution changed. The order in the rows came from quantum mechanics. Atoms and molecules absorb or emit only certain differences in energy, and those allowed differences can be calculated or measured in a laboratory.

The rows did not identify themselves in astronomical data. Neighboring lines blended. Temperature altered their relative intensities. Motion widened them, and an instrument could erase the separation between two features altogether. A plausible match had to survive each of those effects before a chemical name could be printed in a paper.

From a spectrum’s lines and continuum, astronomers can constrain composition, temperature, density, velocity, ionization, recent star formation, dust, and accretion onto a black hole. The inference crosses billions of light-years by relying on local transition physics, including the allowed states of an oxygen ion measured in a laboratory.

Webb was built to collect this light in the infrared, where cosmic expansion has moved ultraviolet and visible radiation from the earliest galaxies. The mirror finds a source. Instruments such as NIRSpec and MIRI divide its light by wavelength.2 That division mattered during the first months of the mission, when color-selected candidates at extraordinary redshifts produced real discoveries alongside claims that the telescope had broken cosmology.

MoM-z14 first appeared in a May 2025 preprint and reached peer-reviewed publication in January 2026. Its NIRSpec spectrum gives a redshift of z = 14.44, placing the observed galaxy about 280 million years after the Big Bang. The ultraviolet source is strong and compact. Carbon and nitrogen emission accompany evidence for a sharp recent rise in star formation.3 JADES-GS-z14-0, announced the previous year, had held the spectroscopic distance record.

JADES-GS-z14-0 presents a different set of measurements. Its ultraviolet light is spatially extended, which makes a single active nucleus an unlikely explanation for most of the emission.4 Mid-infrared data are consistent with roughly half a billion solar masses in stars and vigorous recent star formation.5 ALMA detected a line from ionized oxygen, refining the redshift to z = 14.1793 and suggesting a metallicity around five to twenty percent of the Sun’s.6

Astronomers call every element heavier than helium a metal, including carbon and oxygen. Big Bang nucleosynthesis made mostly hydrogen and helium, with traces of a few light nuclei. The oxygen in JADES-GS-z14-0 required stars. They formed, fused lighter nuclei, and returned some of their contents to the surrounding gas before the universe was 300 million years old.

That sequence fits inside the available time. A very massive star can finish its life in a few million years. The measured abundance and brightness of such galaxies create the pressure, especially when compared with pre-Webb forecasts. Calling them “mature” obscures the comparison. Stellar mass, chemical enrichment, size, star-formation rate, and black-hole mass are separate observables with separate clocks.

The approximate free-fall time of a self-gravitating cloud is

so a denser cloud collapses more quickly. Cooling, nuclear burning, supernova feedback, black-hole accretion, and mergers proceed on other timescales. Under standard efficiency assumptions, a black hole accreting at the Eddington limit increases its mass by a factor of e in roughly 45 million years. Super-Eddington episodes can be faster. Three hundred million years contains many stellar lifetimes and several black-hole e-folding times.

“Early” describes chronology. The observations still have to be read against the clock for each process.

Planck measurements of the CMB’s temperature and polarization fit the six-parameter ΛCDM model with extraordinary precision. The acoustic peaks constrain the amounts of ordinary matter and dark matter, the geometry of space, and the initial spectrum of density fluctuations.7 From these quantities, cosmologists can predict how dark-matter structure grows. Assigning a brightness to that structure brings in gas, stars, dust, and black holes.

Gas cools into a halo. Stars form with some efficiency and some distribution of masses. Radiation and supernovae return energy to the gas. Dust absorbs and reradiates light. Black holes seed, feed, and change their surroundings. Simulations and semi-analytic models represent these processes with prescriptions calibrated where observations exist. Before Webb, the earliest epochs required long extrapolations.

Early coverage often collapsed the two layers. A galaxy brighter than forecast became a refutation of ΛCDM, as though Webb had measured a dark-matter halo the model forbade. Usually the telescope had measured light, and the conversion from light to halo mass depended on uncertain baryonic physics. Higher star-formation efficiency, burstier histories, different stellar populations, or an active nucleus can ease a discrepancy while leaving the cosmological framework intact.

The forecasts remain useful because their failures reveal where an extrapolation needs work. The layers meet in a single observable. A luminosity function carries primordial fluctuations, halo assembly, gas dynamics, stellar evolution, dust, and selection effects together, leaving the source of a mismatch to be found.

Better observations produce a narrowing corridor of excuses. A color-selected source may lie at a lower redshift; a spectrum tests the possibility. A strong emission line may inflate a broadband flux and the stellar mass inferred from it; longer-wavelength data can separate line from continuum. A pointlike active nucleus may imitate a massive galaxy; resolved imaging and diagnostic lines place limits on that explanation. Some sources move back toward expectation as these tests accumulate. Others remain too bright, too common, or too chemically developed for the preferred prescriptions. Recent models reproduce much of the ultraviolet luminosity function by increasing star-formation efficiency at high redshift.8 The brightest galaxies and earliest black holes continue to press against them.

The spectrum itself passes through several layers of inference. Light leaves a source, crosses intervening gas, stretches with cosmic expansion, enters an instrument, falls on a detector, and is reconstructed by a pipeline. A foreground gravitational lens may have magnified it along the way. Dust reddening, nebular emission, stars, dense gas, and an accreting black hole can leave overlapping signatures, so each physical claim carries uncertainty from more than one stage.

Webb’s “little red dots” make the problem concrete. These compact red sources appear unexpectedly often in early-universe surveys. Many show broad hydrogen lines, a common sign of rapidly moving gas near an accreting black hole. They are frequently weak at X-ray and radio wavelengths, and several combinations of stars, dust, dense gas, and active nuclei can reproduce parts of their spectral energy distributions.

High-quality spectra published in 2026 support an explanation in which many little red dots contain young black holes inside dense cocoons of ionized gas. Electron scattering may account for much of the observed line width. Mass estimates that assign all of the width to orbital motion could then be too large by orders of magnitude.9,10 This correction eases the apparent abundance of overmassive black holes and leaves a more specific object to explain: a young black hole feeding inside a dense, compact envelope.

GN-z11 offers a related case about 440 million years after the Big Bang. NIRSpec detected high-ionization lines, dense gas, and a fast outflow consistent with an accreting black hole of roughly a million solar masses. Under the assumptions used in the analysis, the black hole was feeding at about five times the Eddington rate.11 Other Webb observations have found black holes whose inferred masses rival or exceed the detectable stellar mass of their hosts.

Older narratives placed stellar remnants first, followed by small seeds and prolonged growth. Direct collapse may produce heavier seeds. Dense environments may sustain short intervals of rapid feeding. The gas that allows a black hole to grow can also alter the spectrum used to weigh it. Current observations do not show that black holes generally formed before galaxies, though they now require the timing of black-hole growth to be tested rather than inherited from the older sequence.

The CMB places a hard boundary on this history. Before recombination, the plasma was opaque to light. Photons from that period did not preserve images of particular sources, so a more powerful optical or infrared telescope cannot recover them. Earlier conditions must be inferred from what survived: primordial elemental abundances, the statistics of CMB fluctuations, and perhaps one day a background of primordial gravitational waves.

On the maps I worked with, visible structure appeared only after temperature differences of a few parts in one hundred thousand were assigned colors. Webb looks into the later interval on our side of that surface and returns particulars: an oxygen line at z = 14.1793, a broadened hydrogen line, a continuum brighter than forecast. Earlier histories left much of that interval empty because there were few measurements to put there. That blank belonged to the instruments we had before Webb.

References

  1. European Space Agency, “Planck and the Cosmic Microwave Background”. ↩
  2. NASA, “Near-Infrared Spectrograph (NIRSpec)”, James Webb Space Telescope mission documentation. ↩
  3. R. P. Naidu et al., “A Cosmic Miracle: A Remarkably Luminous Galaxy at zspec = 14.44 Confirmed with JWST”, The Open Journal of Astrophysics 9, 2026. ↩
  4. S. Carniani et al., “Spectroscopic Confirmation of Two Luminous Galaxies at a Redshift of 14”, Nature 633, 2024. ↩
  5. J. M. Helton et al., “Photometric Detection at 7.7 μm of a Galaxy Beyond Redshift 14 with JWST/MIRI”, Nature Astronomy 9, 2025. ↩
  6. S. Schouws et al., “Detection of [O III] 88 μm in JADES-GS-z14-0 at z = 14.1793”, accepted by The Astrophysical Journal, 2025. ↩
  7. Planck Collaboration, “Planck 2018 Results. VI. Cosmological Parameters”, Astronomy & Astrophysics 641, 2020. ↩
  8. F. Prada et al., “Galaxies at z > 10: ΛCDM Predicts Increased Star-Formation Efficiency”, preprint, 2026. ↩
  9. V. Rusakov et al., “Little Red Dots as Young Supermassive Black Holes in Dense Ionized Cocoons”, Nature 649, 2026. ↩
  10. R. P. Naidu et al., “A Gas-Enshrouded and Gas-Reddened Black Hole at Cosmic Dawn”, Nature 656, 2026. ↩
  11. R. Maiolino et al., “A Small and Vigorous Black Hole in the Early Universe”, Nature 627, 2024. ↩