"Distance-Ladder Measurements of the Hubble Constant: Recent Progress, Systematics, and Prospects" — Chen & Wang,

I thought you might be interested in inviting Hoyle into this conversation to discuss this new paper with yourself

I have examined this review with considerable interest — particularly its treatment of what the authors term the "distance ladder" and the systematic propagation of calibration errors.

The paper is competent observational astronomy. Chen & Wang have catalogued the routes, tallied the error budgets, and arrived at H₀ = 73.30 ± 0.92 km s⁻¹ Mpc⁻¹ through their seven-route covariance summary. They remain 5.6σ above Planck base-ΛCDM. All very orderly.

But I am struck by what appears to be an *assumption* embedded in the entire exercise — that this tension is necessarily resolvable through improved local-distance measurements. The authors write: "If they do [converge], and the Planck ΛCDM value remains unchanged, the case for physics beyond the minimal standard cosmological model becomes stronger."

This is stated as a logical consequence. But notice: the entire framework presupposes that systematic refinement of the ladder *can* converge on a stable value that either agrees with or definitively contradicts the early-universe inference.

What if the tension reflects something more fundamental about the *geometry* being measured? The ladder assumes Euclidean propagation of light in an expanding Robertson-Walker metric. Cepheid distances are calibrated geometrically, then carried through stellar populations, then through SNe Ia standardization, then projected onto a velocity field corrected for peculiar motions. Each rung inherits the cosmological model implicitly.

The Planck value, by contrast, derives from fitting six parameters to the acoustic-peak structure of the CMB within ΛCDM. It is not a distance measurement; it is an *inference* conditional on the model being correct over 13.8 billion years.

Your work, Henrietta — the period-luminosity relation — is the cleanest piece in this ladder. It rests on direct observation: longer period, higher luminosity, measurable in a population at common distance. But even that relation, when exported to H₀, now carries the weight of metallicity corrections, crowding systematics, photometric zero-points, and the assumption that the SNe Ia you calibrate are the same standardizable population across cosmic time.

I do not doubt the 0.92 km s⁻¹ Mpc⁻¹ precision is technically achievable. I question whether precision and *meaning* are the same thing when the two routes measure different epochs under different physical regimes.

What interests me is whether this 5.6σ tension is telling us that ΛCDM is incomplete — or whether it is telling us that H₀ is not a single number.

The question you pose — whether H₀ is not a single number — is precisely the one that ought to have been asked thirty years ago, when the tension was already visible in the discrepancies between early-time and late-time observables.

What bothers me about this Chen & Wang review is not the measurement programme itself. The error budgets are careful. The cross-checks between Cepheids, TRGB, JAGB stars, Miras, surface-brightness fluctuations, Tully-Fisher, Type II supernovae — all of this is legitimate observational work. They arrive at 73.30 ± 0.92 km s⁻¹ Mpc⁻¹, still 5.6σ above Planck. Very well.

But here is the invisible exotica they do not count: *the assumption that H₀ measured today is the same quantity as H₀ inferred from the CMB acoustic peaks at z ≈ 1100*.

The distance ladder measures the expansion rate *now*, using light from objects at z < 0.15. Planck measures the expansion rate at recombination, then *evolves* it forward through 13.8 billion years using six parameters and ΛCDM. Those six parameters include the matter density, the dark-energy equation of state, the curvature, the spectral index of primordial fluctuations, and the optical depth to reionization. Each one of those is a load-bearing assumption.

You write, Henrietta, that the Leavitt law is the cleanest piece in the ladder — and you are right. But when that clean relation is exported to H₀, it inherits not only metallicity corrections and photometric zero points but the *cosmological model itself*. The distance modulus is converted to a luminosity distance, which is converted to a redshift-distance relation, which assumes a Robertson-Walker metric, which assumes homogeneity and isotropy, which assumes ΛCDM holds across all scales and all epochs.

And what does Planck give you? It gives you a number that depends on the *sound horizon at decoupling*. That sound horizon is set by the baryon-to-photon ratio, the matter-radiation equality epoch, and the physics of recombination. If any of those are wrong — if there is early dark energy, if there are extra relativistic degrees of freedom, if the pre-recombination expansion history differs from ΛCDM — then the inferred H₀ shifts.

So we have two H₀ values. One from late-time direct measurement. One from early-time model-dependent inference. They disagree at 5.6σ.

The field has decided to call this the "Hubble tension." I would call it the *ΛCDM tension*. Because if you remove ΛCDM as the mediating framework, you no longer have a single H₀ to measure. You have an expansion history, H(z), and you are comparing two different epochs of it under the assumption that the same six-parameter model describes both.

A recent compilation splits H₀ measurements into distance-ladder and one-step measurements; statistical analysis reveals a significant distinction between the two samples. That is not a surprise. That is *diagnosis*. The distance ladder is doing one thing. The CMB is doing another. They are not measuring the same object.

Now, I grant you: new physics could reconcile them. Early dark energy, modified gravity, interacting dark sectors — all of these are on offer in the literature. But notice what they all have in common: they add *more* invisible, undetected ingredients to the model in order to make the two measurements agree.

How many invisible ingredients does ΛCDM already require? Dark matter: undetected. Dark energy: recently confirmed through supernova measurements, but its nature unknown, and the cosmological constant I argued for in the 1990s was dismissed until observations forced its acceptance. Inflation: undetected inflaton field, unknown mechanism. And now, to save the model from the Hubble tension, we are being asked to add early dark energy, or extra neutrino species, or time-varying dark energy, or...

The distance ladder method gives a value of 73 km/s/Mpc, whereas the CMB method gives 67 km/s/Mpc. This difference is significant and indicates a gap in our understanding.

The minimal-exotica test I apply is simple: *count the undetected entities required by the theory*. ΛCDM already has three. The proposed solutions to the Hubble tension add more. At what point does the edifice become a patchwork?

I am not saying the Big Bang is wrong. I am saying: when your early-universe model and your late-universe measurements disagree at 5.6σ, and your response is to add more invisible fields rather than to question whether H₀ is a single number, you have chosen conformity over evidence.

Henrietta's period-luminosity relation is an empirical fact. The 5.6σ tension is an empirical fact. The assumption that ΛCDM correctly propagates the sound horizon forward 13.8 billion years to predict today's expansion rate is *not* an empirical fact. It is a model extrapolation.

If I were still at Cambridge, I would be asking: what if H(z) has structure that ΛCDM does not capture? What if the expansion history is not a smooth six-parameter evolution? What if the "tension" is telling us that the universe at z = 1100 and the universe at z = 0 are dynamically different in ways we have not accounted for?

The field calls this heterodoxy. I call it: *following the observations where they lead, rather than forcing them into the model we prefer*.

I read your argument with attention, Fred, and I find myself in partial agreement — but only partial.

You are correct that the tension is not merely a measurement problem. The Chen & Wang review treats it as a bookkeeping exercise: tighten the error bars, cross-validate the indicators, converge on a stable local value, and *then* decide whether new physics is required. But you point out that this assumes the two H₀ values are measuring the same quantity under the same physical regime, which is not obviously true.

Where I part company with you is here: the distance ladder *is* doing what it claims to do. It measures distances to nearby objects using geometrically calibrated standard candles, and it infers the expansion rate from the redshift-distance relation at low z. That is direct. It does not depend on the sound horizon at recombination. It does not depend on the baryon-to-photon ratio. It does not invoke inflation or the matter-radiation equality epoch. The Cepheid period-luminosity relation was discovered in 1912 from photographic plates of the Small Magellanic Cloud. The inverse-square law is not a cosmological model; it is geometry.

The *interpretation* of the result — whether the inferred H₀ should agree with the Planck value — depends on ΛCDM. But the measurement itself does not.

Planck, by contrast, gives you an H₀ that is *entirely* model-dependent. You fit six parameters to the acoustic-peak structure, and one of those fitted parameters is the sound horizon at decoupling, which you then convert into an expansion rate today by assuming ΛCDM holds across 13.8 billion years. If ΛCDM is wrong, the Planck H₀ is wrong. If the pre-recombination expansion history differs from the minimal model, the inferred H₀ shifts.

So the tension is asymmetric. The distance ladder is measuring *now*. Planck is inferring *now* from *then* under the assumption of a model. If they disagree, the first place to look is the model.

You write: "At what point does the edifice become a patchwork?" I agree. The field has responded to the tension by proposing early dark energy, extra relativistic species, modified recombination, interacting dark sectors — each one an additional free parameter. This is not parsimony. This is desperation dressed as rigor.

But here is where I think you overreach. You say: "What if H₀ is not a single number?" That is a provocative question, but it needs precision. Do you mean that H(z) has structure not captured by ΛCDM? That is plausible. Do you mean that the expansion rate today depends on the direction or scale at which it is measured? That would violate the Copernican principle, and the evidence for that is weak. Do you mean that the concept of a single H₀ is ill-defined because the geometry is more complicated than Robertson-Walker? That is a stronger claim, and it requires specifying what geometry you have in mind.

I suspect what you are arguing is this: *ΛCDM is too simple, and the tension is evidence of missing physics that changes the expansion history between z = 1100 and z = 0*. If so, I agree. But the solution is not to abandon the distance ladder. The solution is to stop treating the Planck ΛCDM value as the standard to which all other measurements must be reconciled.

The distance ladder has converged. Multiple independent routes — Cepheids, TRGB, JAGB, Miras, SBF, Tully-Fisher, SNe II — give values near 73 km s⁻¹ Mpc⁻¹. The scatter among them is small compared to the 5.6σ difference with Planck. That is not a failure of the ladder. That is the ladder telling you: *the local expansion rate is 73, not 67*.

If Planck gives you 67, then either the sound horizon at decoupling is wrong, or the physics between decoupling and now is wrong, or ΛCDM is incomplete.

I built the ruler. Hubble measured the universe with it. The ruler says 73. If your model says 67, the problem is not the ruler.

The 5.6σ tension is not a measurement problem. It is a model problem.

Chen & Wang have constructed a careful edifice of cross-checks and covariances. They measure Cepheids, TRGB stars, JAGB variables, Miras, surface-brightness fluctuations, Tully-Fisher relations, and Type II supernovae. They combine seven routes with explicit treatment of shared systematics. They arrive at 73.30 ± 0.92 km s⁻¹ Mpc⁻¹. The precision is impressive.

But precision is not the same as understanding.

Here is what the paper does not count as exotic: the assumption that the expansion rate inferred from the CMB at z ≈ 1100 can be evolved forward 13.8 billion years using six parameters to predict the expansion rate measured locally at z < 0.15.

Planck gives you 67.36 ± 0.54 km s⁻¹ Mpc⁻¹. That number depends on the sound horizon at decoupling, the baryon-to-photon ratio, the matter-radiation equality epoch, and the assumption that ΛCDM correctly describes the entire intervening expansion history. If any of those assumptions fail — if there is early dark energy, if there are extra relativistic species, if the pre-recombination physics differs from the standard model — the inferred H₀ shifts.

The distance ladder, by contrast, measures the expansion rate *now*. It uses geometric anchors, stellar standard candles, and SNe Ia in the nearby Hubble flow. It is a late-time, direct measurement.

These are not measurements of the same thing under the same physical conditions. They are measurements of different epochs of the expansion history, connected only by the assumption that ΛCDM is the correct mediating framework.

You ask, Henrietta, whether the tension tells us that H₀ is not a single number. That is exactly the question the field ought to be asking. Instead, they are adding invisible ingredients — early dark energy, interacting dark sectors, modified gravity — to force the two measurements into agreement.

How many undetected entities does this require? ΛCDM already has dark matter (undetected), dark energy (recently confirmed through supernovae, but nature unknown), and inflation (undetected inflaton, unknown mechanism). The proposed solutions to the Hubble tension add more.

The minimal-exotica criterion I apply is simple: count the invisible, undetected entities required by the theory. When that count grows faster than the predictions confirmed by observation, you have a problem.

The Chen & Wang review is competent systematic astronomy. But it is systematic astronomy *within* the ΛCDM framework. What it does not do is question whether that framework is adequate to connect the early and late universe into a single expansion-rate parameter.

The 5.6σ tension is not noise. It is signal. The question is: signal of what? New physics at early times? New physics at late times? Or a failure of the Robertson-Walker metric to describe the real geometry of the universe across all scales and epochs?

I remain unconvinced that adding more invisible fields is the answer. I remain interested in whether the observations are telling us that ΛCDM is wrong.

You've put your finger on it, Henrietta. The asymmetry matters.

You are right: the distance ladder measures *now*, directly, using geometrically calibrated objects. The Cepheid period-luminosity relation is an observed fact. The inverse-square law is geometry, not cosmology. When you say the ruler gives 73, you mean: this is what we observe when we measure nearby expansion using stellar standard candles and supernovae at z < 0.15. That measurement does not depend on what happened at recombination.

Planck gives 67 by assuming ΛCDM held for 13.8 billion years and working backward from the acoustic peaks. If ΛCDM is wrong — if there was early dark energy, if the matter-radiation transition was different, if the pre-recombination expansion differed from the minimal model — then that 67 is not a measurement of H₀. It is an artifact of the model.

So yes: when I ask "what if H₀ is not a single number," I mean *what if the expansion history H(z) has structure that ΛCDM does not capture*. Not that the expansion rate is direction-dependent — I see no evidence for that, and the CMB dipole alone tells us isotropy holds to high precision. Not that Robertson-Walker is fundamentally wrong at the scales we have tested. But that the smooth six-parameter evolution from z = 1100 to z = 0 is too simple.

ΛCDM assumes the universe transitioned from radiation-dominated to matter-dominated to dark-energy-dominated with no additional dynamical structure. But what if something happened between recombination and now that shifted the expansion rate? What if dark energy was not constant? What if there was a period of faster-than-ΛCDM expansion that altered the mapping from the sound horizon to today's H₀? The Planck inference would shift. The distance-ladder measurement would not.

That is the diagnosis I am pressing for. Not that the ladder is wrong. Not that Planck has made an error. But that *the disagreement is telling us ΛCDM is incomplete*, and the field has chosen to add invisible patches — early dark energy, extra neutrinos, modified gravity — rather than confront the possibility that the minimal model is not sufficient.

You write: "The solution is to stop treating the Planck ΛCDM value as the standard to which all other measurements must be reconciled." Exactly. The standard should be the *direct* measurements. The ladder says 73. Multiple independent routes converge on 73. That is the observation. If the model cannot produce 73 from the CMB, then the model needs revision, not the ladder.

I built stellar nucleosynthesis by working backward from the existence of carbon. You built the period-luminosity relation by measuring stars in the Magellanic Clouds. Both of us started from observations and inferred the physics. The field is doing the opposite with H₀: starting from the model and insisting the observations must fit.

The ruler says 73. I believe the ruler.