Hartnett's Carmelian Cosmology

Summary

• John Hartnett's Carmelian cosmology is the most technically ambitious young-earth cosmology ever published, and its author has now withdrawn it in print at least five times.  The dates are 7 February 2015, 19 April 2019, 11 March 2021, 23 June 2021 and 6 December 2024.  Two of those statements retract the underlying physics, not only the creation model built on it.

• Hartnett is a real experimental physicist.  He worked on ultra-stable cryogenic sapphire oscillators at the University of Western Australia and later at the University of Adelaide, and he published his Carmelian work in the International Journal of Theoretical Physics and Foundations of Physics, not only in creationist journals.  Whatever is wrong with the model, it is not that an amateur wrote it.

• Moshe Carmeli's theory is genuine published physics.  Carmeli held the Albert Einstein Chair in Theoretical Physics at Ben-Gurion University.  His "cosmological special relativity" adds a fifth coordinate, the recession velocity of a galaxy, and treats the Hubble time τ = 1/H₀ the way ordinary special relativity treats the speed of light.  It was not invented to serve creationism.

• What Hartnett claimed it bought was real, and two of the three claims were checkable.  Spiral-galaxy rotation curves without dark matter.  The high-redshift supernova magnitudes without dark energy, at a matter density Ωm ≈ 0.021 that matches the measured baryon density.  And, in Starlight, Time and the New Physics (2007), enough time dilation on Day Four for light to cross ten billion light years while less than a day passed on Earth.

• The theory assumes the Hubble law rather than deriving it.  Hartnett states this himself in his own paper: "The Hubble law is assumed as a fundamental axiom for the universe."  A model that puts the redshift-distance relation in by hand cannot claim that relation as a success.

• His bounded-universe result is the single most useful thing in this file, and it cuts against him.  Hartnett solved the field equations for a finite universe with our galaxy at the centre and reported that it "is identical with the unbounded solution."  Conner and Page reached the same verdict about Humphreys' bounded universe in 1998.  Two young-earth cosmologies, built independently, both turn out to be observationally indistinguishable from the standard picture they were meant to replace.

• The outside criticism is thin but not absent, and one piece of it is fatal to a specific claim.  Alan Macdonald showed in 2006 that the time-redshift formula in a paper Hartnett co-authored with Carmeli fails an elementary consistency check: it does not give t = t₀ at z = 0.  Ahmet Öztaş and Michael Smith tested the Carmeli distance relation against 580 Union2.1 supernovae in 2014 and found that it fits, along with several other models with "drastically different physical realities."  Fitting the same data is not evidence for a theory when the data cannot tell the theories apart.

• The criticism that actually killed it came from Hartnett.  He caught Carmeli making an invalid simplification inside a hard integral, found a mistake in the gravitational-lensing chapter of Carmeli's last book while editing it for World Scientific, deleted sections of that book that "did not make sense," and never found the five-dimensional solution his creation scenario needed.  In February 2015 he wrote that Carmeli's cosmological special relativity "I believe is fundamentally flawed."

• He treated accelerated expansion as a success in 2005 and doubted it in 2015.  His TJ paper credits Carmeli with predicting acceleration "two years before it was measured."  Ten years later his Creation magazine article asks "does the evidence confirm that the universal expansion is accelerating?"  By December 2024 he had rejected metric expansion altogether, which is why he says Carmeli's cosmology is no longer available to him: "Carmeli's cosmology describes only an expanding universe."

• Where he stands now is on Jason Lisle's anisotropic synchrony convention, and he has said so continuously since 2019.  As recently as 25 June 2026 he wrote that "the ASC model, as initially promoted by Dr. Jason Lisle, is the best solution."  He remains a young-earth creationist.  He did not leave the position;  he left the model.

• The lesson is worth stating flatly.  Of the young-earth cosmologies with any technical content, the decaying speed of light was refuted by a creationist's reanalysis, Humphreys moved off the white hole, and Hartnett withdrew the Carmelian model himself.  Three of the field's flagship answers to distant starlight were abandoned by the people best placed to judge them.  Ministry websites still promote some of them.  Hartnett complained about exactly that in 2021.

Who John Hartnett is, and why that matters

Start with the credentials, because the usual reflex on hearing "young-earth cosmology" is to assume the author is not a physicist.  In this case the author is.

John Gideon Hartnett spent most of his career in the Frequency Standards and Metrology group at the University of Western Australia, building cryogenic sapphire oscillators.  Those are the devices that keep time for very-long-baseline radio astronomy and for tests of fundamental physics, and they are among the most stable frequency sources ever built.  His cosmology preprints from 2004 and 2005 carry that address.¹  He moved to the University of Adelaide in 2013 and worked there with Dr James Chappell on higher-dimensional generalisations of the Minkowski metric.²  His experimental publication record is in mainstream journals and has nothing to do with creationism.

His cosmology work was published the same way.  The rotation-curve paper is in the International Journal of Theoretical Physics.³  The supernova fit is in Foundations of Physics.⁴  The bounded-universe solution is in the International Journal of Theoretical Physics again.⁵  Preprints of all of them sit on arXiv under his own name and university address.¹⁴⁶  He also has a paper in Astrophysics and Space Science that killed the best surviving version of the redshift-quantisation claim, which he had previously championed himself.⁷⁷

That last item is worth pausing on.  Hartnett looked for periodicity in Sloan Digital Sky Survey quasar counts, found it, then found the same periods in the survey's own reliability flag, and published the negative result:  "these redshift periods must result from some selection effect, as yet undetermined.  It does not signal any intrinsic (quantized) redshifts in the quasars."⁷  He did the same thing in 2018 with the missing Sunyaev-Zel'dovich shadow, an argument he had made for Creation magazine in 2006 and retracted in Journal of Creation twelve years later:  "At best this no-shadow argument is now equivocal and hence I suggest that it should no longer be used as an argument against the big bang hypothesis."⁸⁸

So the pattern this article documents is not new for him.  Hartnett has a habit of publishing the result that goes against his own earlier position.  That is the behaviour a skeptic should want from anyone, and it is why the retraction record below deserves to be read as evidence about the model rather than as an embarrassment to be explained away.

What Carmeli's theory actually says

Moshe Carmeli was born in Baghdad on 15 June 1933 and died in Beer Sheva on 27 September 2007.⁹  He took his doctorate at the Technion under Nathan Rosen, who had been Einstein's assistant, and became Albert Einstein Professor of Theoretical Physics at Ben-Gurion University in 1979.⁹  He wrote standard textbooks on group theory, gauge fields and spinors.  In 1997 he published Cosmological Special Relativity: The Large Scale Structure of Space, Time and Velocity, and over the following decade extended it into a general-relativistic version he called cosmological general relativity.¹⁰¹¹

The idea is easier than the algebra makes it look.

In ordinary special relativity you have three space coordinates and one time coordinate.  The speed of light c converts between them, and it is a hard limit:  nothing goes faster.  The invariant interval is ds² = c²dt² − dx² − dy² − dz².

Carmeli built a parallel structure in which the fourth coordinate is not time but the recession velocity of a galaxy in the Hubble flow.  The conversion constant is not c but the Hubble time τ = 1/H₀, which he called the Hubble-Carmeli time constant and evaluated at about 4.28 × 10¹⁷ s, or 13.54 billion years.⁶  The invariant interval becomes τ²dv² − dx² − dy² − dz² = ds².¹⁰

The logic runs from the Hubble law.  Write the Hubble law as R = τv, which is just R = v/H₀ rearranged.  Carmeli sets that beside the equation for light, R = ct, and notes the formal resemblance.¹⁰  He then assumes that the combination x² + y² + z² − τ²v² is the same when measured at any cosmic time, exactly as x² + y² + z² − c²t² is the same in any inertial frame.¹⁰  Out of that assumption falls a transformation that looks like a Lorentz transformation with t/τ where v/c normally sits.

The consequence that gives the theory its name is the analogue of the speed limit.  In ordinary special relativity nothing exceeds c.  In Carmeli's version nothing exceeds τ:  no cosmic time interval can be longer than the Hubble time.  The universe has a maximum age built into the geometry.  Hartnett read the same structure off the supernova data and described it as showing "that the Hubble expansion of the universe is speed-limited."⁴

Two features of this deserve emphasis before anything is claimed for it.

Carmeli's cosmological special relativity explicitly assumes gravity is negligible.  His own abstract says so:  "in this theory it is assumed that gravitation is negligible."¹⁰  It is a theory of the empty case, and cosmological general relativity is the attempt to put gravity back.

The Hubble law is an input, not an output.  Hartnett states this plainly in the introduction to his supernova paper:  "The Hubble law is assumed as a fundamental axiom for the universe and the galaxies are distributed accordingly."⁴  He repeated it in 2014 when answering a critic:  "the metric upon which the model is built implicitly assumes the Hubble law as an axiom."¹²¹²

That second point is the deepest structural objection to the whole programme, and it does not require any equations to see.  The redshift-distance relation is the central observational fact of extragalactic astronomy.  Standard cosmology derives it from the field equations plus a homogeneous matter distribution.  Carmeli's theory writes it into the metric at the start.  So the theory cannot count agreement with the Hubble law as a success, because agreement was assumed.  What it can legitimately claim is everything downstream of that assumption:  the shape of the magnitude-redshift curve at high redshift, the rotation curves of galaxies, and so on.

The fifth dimension nobody could interpret

The velocity coordinate is where readers get stuck, and Hartnett got stuck there too.

Asked in 2015 what a velocity dimension is, he answered:

"What is a velocity dimension?  I do not know.  Essentially Carmeli treated it as redshift and used the mathematical formalism of Einstein's field equations.  I followed the same approach.  But I have never been happy with his new dimension."²²

He had made the same concession in print two years earlier, in a reply to a letter in Journal of Creation:  "I do not know what expansion of space, nor what accelerating space, really means, let alone a velocity dimension describing it."¹²

There is a defensible position underneath the discomfort.  Every cosmological model is a mathematical structure fitted to a short baseline of observations, and Hartnett's answer to the charge of inventing an unobservable dimension was to ask whether dark energy, inflation, and metric expansion are any better off.¹²  That is a fair debating point and it is also the point at which the argument stops being about evidence.

The stronger version of the objection, which Hartnett accepted, is that a five-dimensional metric has an extra function's worth of freedom in it, so fitting a rotation curve with one is less impressive than fitting it with four dimensions and no free functions.  Randy Speir put it this way in a 2012 letter published in Journal of Creation in 2013:  "it is not at all surprising that a bonus dimension might well give the appearance of solving the problem of galaxy rotation curves."¹³¹³  Hartnett's reply is that he did not expect it to work and was surprised when it did.¹²  Both things can be true.  A surprised author is still an author with a spare dimension.

What Hartnett claimed the theory bought

Three claims, in the order he made them.

Galaxy rotation curves without dark matter

Stars and gas in the outer disks of spiral galaxies orbit faster than the visible mass can explain.  The standard answer is a halo of unseen matter.  The best-known alternative is Mordehai Milgrom's MOND, which changes Newton's law below a characteristic acceleration a₀ ≈ 1.2 × 10⁻¹⁰ m/s² fitted to the data.

Carmeli's theory produces a MOND-like low-acceleration regime without fitting anything, because a characteristic acceleration falls straight out of the constants:  a₀ = c/τ = cH₀.¹¹  Hartnett worked the metric through for spiral galaxies and got the Tully-Fisher relation, the observed proportionality between a galaxy's luminous mass and the fourth power of its rotation speed.  His paper says the agreement with Milgrom holds "in the low acceleration limit," and that the numerical constant comes out at (2/3)a₀ ≈ 5.2 × 10⁻¹⁰ m s⁻², depending on the precise value of τ.¹

That is a genuine theoretical achievement of the same kind MOND claims:  one number, not fitted, reproducing a scaling law across many galaxies.  It is also about four times Milgrom's empirical value, which Hartnett describes as "slightly different."¹  A factor of four is not slight.

Hartnett then fitted the rotation curve of NGC 6503 and reported the result honestly, including the part that goes against him.  His derived total galaxy mass came out at 2.0 × 10⁹ M☉ against a published luminosity mass of 4.8 × 10⁹ M☉.  His own comment:  "It is underestimated, which is the reverse of the dark matter problem."¹  A theory built to remove missing mass produced missing light instead.  He printed it anyway.

He extended the same treatment to elliptical and spheroidal galaxies in 2008.¹⁴

The supernova magnitudes without dark energy

The 1998 discovery of accelerating expansion rests on Type Ia supernovae being fainter at high redshift than a decelerating universe predicts.  Standard cosmology accounts for that with a cosmological constant, about 70 per cent of the energy budget.

Hartnett fitted the same magnitude-redshift data with Carmeli's theory and no dark matter at all.  His best fit gave a matter density Ωm ≈ 0.021, which he pointed out "falls well within the measured values of the baryonic matter density."⁴⁴  For 185 data points the speed-limited form of his relation gave χ² = 19.94, or χ²/N = 0.108.⁴  The paper went into Foundations of Physics in 2006.⁴

This is the claim a skeptic should take most seriously, because it is the one with a number attached.  It says a theory with no dark sector fits the supernova Hubble diagram at a matter density equal to the measured baryon density.  If true and unique, that would be a serious result.

The problem is uniqueness, and the best evidence on that point comes from outside creationism entirely.  See below.

Starlight during creation week

The third claim is the one this article's parent series is about.  In Starlight, Time and the New Physics (2007), Hartnett proposed that God switched on an enormous acceleration of the cosmic expansion at the start of Day Four and switched it off at the end.¹⁵  During that one ordinary Earth day, clocks in the distant cosmos ran vastly faster than clocks here, so light had time to cross the universe.

The numbers, as he restated them in 2014:

"Everywhere cosmic clocks 'tick' almost a 10 trillion times faster than Earth clocks for the period of Day 4 of Creation week.  Light then can travel the 8 million light-year distance he speaks of in about half an hour as measured by Earth clocks.  See Fig A6.2 in Appendix 6.  Light from anywhere in a 13 billion-light-year radius could reach Earth in less than 24 hours as measured by Earth clocks."¹²

The mechanism is that space expands everywhere except inside the solar system.  Earth clocks therefore keep ordinary time while everything outside runs fast.  "Part of the mechanism is that one can calculate an apparent one-way speed of light which becomes enormous by the same magnitudes mentioned here.  But it is not a change in the local speed of light (c), which determines all normal particle and gravitational physics."¹²

Two things about the book are easy to miss and both matter.

The creation model is not Carmeli's, and Hartnett says so.  Appendices 1 to 5 apply Carmeli's published theory to observations.  Appendix 6 is Hartnett's own five-dimensional extension, which Carmeli never wrote:  "Carmeli only constructed a 4 dimensional (4D) space-velocity model for the Universe.  He never found a full 5D model for the whole Universe, which might be used to construct a history close to the Creation epoch.  I theorised on a new model, call it the Hartnett-Carmeli model if you like, which is 5D."¹⁶¹⁶

Hartnett never claimed the book proved anything.  His own words in the same post:  "I do not claim it is the answer but if God did something like I describe it could provide us with an understanding of it.  It is a possible solution and I admit there are several potential areas where the solution may be found."¹⁶  Anyone quoting Starlight, Time and the New Physics as a settled young-earth answer is claiming more for it than its author ever did.

The bounded universe that made no difference

This is the most useful result in the file, and it is Hartnett's own.

Every young-earth cosmology that uses gravitational or cosmological time dilation needs the universe to have a centre and an edge, with the Earth near the centre.  Without a centre there is no preferred location for clocks to run slow.  Hartnett stated the requirement in his 2005 TJ paper:

"The observations that place the earth near the centre of the universe are consistent with God's focus on mankind.  The Bible implies that there are a finite number of stars, which suggests that the universe is finite.  A creationist cosmology requires a finite universe that is most likely spherically symmetric about our galaxy."¹⁷¹⁷

So he went and solved it.  He took Carmeli's field equations, put our galaxy at the centre of a finite, bounded, spherically symmetric matter distribution, and found the solution.  The abstract of the paper reports what he found:

"The solution of Einstein's field equations in Cosmological General Relativity (CGR), where the Galaxy is at the center of a finite yet bounded spherically symmetrical isotropic gravitational field, is identical with the unbounded solution."⁶⁶

He restated it in 2013 when the paper appeared in the International Journal of Theoretical Physics:  "The new solution though is identical with the unbounded solution."¹⁸¹⁸  And he drew the honest consequence himself:  the supernova data "cannot distinguish between unbounded models and those with finite bounded radii of at least cτ."⁶

Read that slowly.  A young-earth physicist set out to show the universe has a centre and an edge, succeeded mathematically, and reported that the result is observationally identical to the model without a centre or an edge.  The boundary buys nothing.

The same thing happened to the other flagship model, and by a different route.  In 1998 Samuel Conner and Don Page examined Russell Humphreys' white-hole cosmology, which also requires a bounded universe with Earth near the centre, and concluded that "physical clocks located on Earth and on distant galaxies behave identically in bounded and unbounded Universes which have identical interior matter distributions."¹⁹¹⁹  That was a critique from outside the model.  Hartnett's is a report from inside it.  The two agree.  See the white hole cosmology article for the Conner and Page case in full.

Two independent bounded young-earth cosmologies, one checked by critics and one checked by its own author, both come out observationally indistinguishable from an unbounded universe.  That is a convergent result and it is worth more than either paper alone.

Hartnett's response to this was philosophical, and he was open about it.  His conclusion in 2013:  "Only philosophically can this solution be rejected."¹⁸  He supported it with a line from George Ellis that circulates widely in creationist writing, taken from a 1995 Scientific American profile:  "I can construct you a spherically symmetrical universe with Earth at its center, and you cannot disprove it based on observations… you can only exclude it on philosophical grounds."²⁰²⁰

Ellis says the same thing in his own peer-reviewed survey of the philosophy of cosmology, and there it can be checked.  Section 4.2.2:  "We can easily construct exact spherically symmetric universe models, as indicated by these observations.  In general they will be spatially inhomogeneous, with our Galaxy located at or near the centre;  this is currently a philosophically unpopular proposal, but is certainly possible."²¹²¹  He also states the general result as a numbered thesis:  "Establishing a Robertson-Walker geometry for the universe relies on plausible philosophical assumptions.  The deduction of spatial homogeneity follows not directly from astronomical data, but because we add to the observations a philosophical principle that is plausible but untestable."²¹

The very next sentence in Ellis is the one creationist citations leave out, and it should be printed.  "The purpose of the above analysis is not to seriously support the view that the universe is spherically symmetric and inhomogeneous, as is allowed by the observations, but rather to show clearly the nature of the best observationally-based argument by which we can (quite reasonably) justify the assumption of spatial homogeneity."²¹  Ellis is mapping the logical structure of an assumption, not endorsing a geocentric matter distribution.  He also gives the argument that does most of the work against one:  the Ehlers-Geren-Sachs theorem says exact microwave-background isotropy for every freely moving observer forces a Robertson-Walker geometry, so the Earth-centred model survives only by denying that other observers see what we see.²¹

Hartnett is still right that the bounded solution cannot be excluded by any measurement now in hand, and that is a two-edged point.  If the bounded solution cannot be excluded by observation, then it also cannot be supported by observation.  A model that no measurement can distinguish from the standard one has stopped competing with it.  Hartnett later drew this lesson himself and made it a theme:  cosmology is underdetermined, meaning many different models fit the same data, so the choice among them is made on other grounds.²²²²  That is an honest position.  It is not a young-earth argument.  It is a reason for thinking that no cosmological argument, in either direction, decides much.

The criticism from outside creationism

The published critical literature on Carmeli's cosmology is thin.  Most physicists ignored it rather than refuting it, which Hartnett attributed to its being "a significant departure from the standard model."²  That is probably a fair description of what happened, and it is not by itself a criticism of the theory.  What criticism exists is worth setting out precisely, because it divides cleanly into objections to Carmeli's physics and objections to the fit.

Macdonald 2006: a published comment that finds an elementary error

Alan Macdonald, a mathematician at Luther College in Iowa, published a comment in Foundations of Physics Letters on a paper Carmeli wrote with Hartnett and Firmin Oliveira.²³  That paper derived a relation between the cosmic time when light was emitted and its redshift, then modified it to include gravity by swapping the age of the universe t₀ for the Hubble time H₀⁻¹, a step the authors justified because it "seems to be the proper thing to do."²³

Macdonald's objection is short enough to state in one line:

"We can see immediately that Eq. (2) is not an exact time-redshift relation.  For it does not give t(0) = t₀, which is obviously true for every cosmological model.  (Eq. (1) does pass this test.)"²³

A formula for the age of the universe at redshift zero must return the present age of the universe.  Carmeli, Hartnett and Oliveira's did not.  Macdonald also computed the size of the disagreement with the standard flat ΛCDM model:  "they differ by over half a billion years for 1 ≤ z ≤ 4.5."²³

This is a narrow objection.  It hits one formula in one paper, not the whole theory.  But it is the only published, peer-reviewed, non-creationist technical criticism of a Carmeli cosmology paper this project has been able to find, and the authors had written that their equation was "valid for all redshift values."²³  No reply from Carmeli, Hartnett or Oliveira has been located.

Öztaş and Smith 2014: the fit is real and it proves nothing

The best test of the supernova claim comes from two authors with no stake in creationism.  Ahmet Öztaş of Hacettepe University and Michael Smith of 4S Fuel Research derived and evaluated a set of cosmological models against the 580-supernova Union2.1 compilation, including Carmeli's.²⁴  Their title tells you where their sympathies lie:  the results present "underwhelming support for the standard model."  They are not defending ΛCDM.

They fitted the Carmeli distance relation Hartnett had published, and it fitted.  So did several others.  Their conclusion is the important sentence:

"This problem is reflected by the inability of even 581 data pairs to allow discrimination between several models examined here, even though these models present drastically different physical realities."²⁴

That is the correct verdict on Hartnett's headline claim, and it is not hostile.  His fit to the supernova Hubble diagram without dark energy is real.  It is also not evidence for his theory, because the supernova magnitudes alone cannot tell his theory apart from several incompatible rivals.  Supernovae are one leg of the modern cosmological case.  The others are the microwave background power spectrum, baryon acoustic oscillations, big bang nucleosynthesis, and the growth of structure.  Carmeli's theory was never fitted to any of them.

The objection with no citation attached, because it needs none

The Hubble law is an axiom of the theory.⁴  So the theory cannot explain why redshift is proportional to distance;  it starts from that.  Standard cosmology derives the relation from the field equations plus homogeneity, and gets, as a by-product, a prediction for how the relation bends at high redshift.  Carmeli's theory has to be told the relation and then fits the bending with the free constants it has left.  That is a weaker kind of success and it should be labelled as one.

The criticism from inside creationism

One published critique of the Hartnett model exists in the creationist literature, and it went badly for the critic.

Randy Speir, writing from Barry Setterfield's Genesis Science Research site, sent a letter to Journal of Creation in June 2012 raising four objections.  It was published with Hartnett's reply in April 2013.¹³  Speir later posted a longer version on Setterfield's site, and Hartnett answered it line by line on his blog in January 2014.¹²

Speir's four objections and what happened to them:

The fourth objection is where the exchange turns.  Speir proposed replacing Hartnett's differential equation dr² = H₀⁻²dv² with one containing a constant term for the initial 8-million-light-year radius, and concluded that the model could not deliver distant starlight after all.  Hartnett's reply is blunt:  "This is a differential equation which involves no constant… This proves he does not understand the model, nor, it seems, basic mathematics."¹²

Hartnett was right on the mathematics.  He was also conceding, in the same exchange, that his fifth dimension may be a fudge factor and that he does not know what the microwave background is.  Speir's letter was the only sustained attempt anyone in the movement made to check the model, and it missed.

One more objection deserves recording because it comes from a young-earth astronomer who has actually read all the models.  Danny Faulkner's 2021 survey of light-travel-time solutions describes Hartnett's work accurately and neutrally, then makes an observation that applies to the whole class:  "many supporters of solutions of the light travel time problem based upon general relativity don't truly understand these solutions."²⁵²⁵  That is the condition in which Hartnett's model was being promoted, and Hartnett came to complain about it himself.

The retractions

Hartnett withdrew this work in public at least five times over nine years.  The dates and the reasons are different each time, and the reasons get deeper as the sequence goes on.  Set them out in order.

7 February 2015: the underlying theory is "fundamentally flawed"

The first and most technical withdrawal is buried in a blog post that answers a former colleague's questions about cosmology.  It gets no headline and it is the most damaging document in the sequence, because it is about Carmeli's physics rather than Hartnett's creation model.

On the velocity dimension:

"What is a velocity dimension?  I do not know.  Essentially Carmeli treated it as redshift and used the mathematical formalism of Einstein's field equations.  I followed the same approach.  But I have never been happy with his new dimension.  And his Cosmological Special Relativity theory (which is where he began, and is the title of his first book on this) I believe is fundamentally flawed.  It has problems which I could not see how to overcome.  But the spacevelocity theory applied to large scale structure of the universe seems to work quite well.  I say this to highlight that his theory has unsolved problems (more than I have mentioned here) and it is not a magic bullet."²²

On editing Carmeli's last book for World Scientific after Carmeli's death:

"Working with his personal assistant (with a PhD in cosmology) we found several sections by him, new material never before published, some of which did not make sense.  So we had to delete those sections, because he could not correct them.  So we were acting as not only an editor would but also as referees."²

On the gravitational-lensing chapter of that book, which Hartnett had hoped would let Carmeli's theory take on cluster dark matter:

"On closer examination, with a graduate student, I soon realized that Carmeli had made a mistake and the theory was inconsistent, hence invalid for the purposes he claimed.  So all that to say, Carmeli does not offer any new insights there."²

He gives a fourth technical complaint in the 2016 post:  Carmeli's derivation of the five-dimensional galaxy dynamics rests on "a simplifying assumption on a very-difficult-to-solve integral, which I do not believe is valid," and Hartnett says he raised it with Carmeli in person.¹⁶

So the position by early 2015 was: cosmological special relativity fundamentally flawed, the lensing extension mathematically invalid, parts of the final book incoherent enough to delete, and the key galaxy-dynamics derivation resting on a step Hartnett thought was wrong.  What survived was the empirical fit of the four-dimensional space-velocity model to large-scale structure, which he said "seems to work quite well."

19 April 2019: the ASC model is better

The second statement retracts the creation model.

"By August 2015, when I gave the lecture, I had come to the conclusion that Jason Lisle's Anisotropic Synchrony Convention (ASC) model was better than my own Carmeli-Hartnett model."²⁶²⁶

The reasons:

"In the years leading up to 2015 I had discovered several serious inconsistencies with Carmeli's cosmology.  Besides those, there did not exist a viable 5D (space-time-velocity) version that could be used to give a robust description of the early universe on a biblical timeline."²⁶

And the conclusion:

"On this website you'll find articles on my use of Carmeli's cosmology.  They will all remain, but note, that I no longer have much faith in such an approach.  So that brings me up to the present.  I now believe that Lisle's ASC model is the best solution by a long shot."²⁶

Notice the second failure, which is separate from the inconsistencies.  His model needed a five-dimensional solution covering space, time and velocity together, for the universe as a whole, near the creation epoch.  Carmeli never produced one.  Neither did Hartnett.  Appendix 6 of the book was a linearised sketch of what such a solution might look like, not the solution.

11 March 2021: it has too many problems, and it is still being promoted

The third statement is in a post about why Hartnett stopped writing for Creation Ministries International.  The post's main subject is a dispute over vaccines and editorial policy that has nothing to do with cosmology, and readers should know that before following the link.  The cosmological paragraph stands on its own:

"My own cosmological model, which I developed using Carmeli's cosmology, has too many problems and I have since abandoned it.  Even so it is still promoted as a viable model, even in their premier publication, see chapter 5 of the Creation Answers Book.  But Jason Lisle's model is not mentioned at all."²⁷²⁷

That complaint was accurate and it has stayed accurate.  The Creation Ministries International page listing arguments creationists should avoid still says, in the same paragraph where it retires the decaying speed of light, that "CMI currently believes that both Dr Russell Humphreys' and Dr John Hartnett's cosmologies (both involving relativistic time dilation) provide viable solutions for the distant starlight problem."²⁸²⁸  One of those two authors has retracted his cosmology in writing.  The other has moved to a different model, which Danny Faulkner reported in 2021:  "Many supporters of the white hole cosmology don't know that Humphreys has since abandoned his model."²⁵

23 June 2021: "I have moved on, so should you"

The fourth statement is the shortest and the most quoted.  Hartnett went back to his 2016 post defending the model and appended an update:

"Important update June 23, 2021:  I have given up on this model.  I have now abandoned this model.  There are too many inconsistencies and besides we never found a full 5D (space, time and velocity) solution for the whole universe.  I have moved on, so should you."¹⁶¹⁶

The last four words are addressed to his own readers.

6 December 2024: the universe may not be expanding at all

The fifth statement adds a reason that none of the others gave, and it is the one that closes the door.

"By the way I gave up on Carmeli's cosmology some years ago for a few reasons.  I came to the view that the universe is not expanding but Carmeli's cosmology describes only an expanding universe."²²²²

This is a bigger move than it looks.  Carmeli's fifth coordinate is the recession velocity of a galaxy in the Hubble flow.  If there is no Hubble flow, the coordinate has nothing to measure.  A physicist who concludes that space does not expand has not merely lost confidence in Carmeli's theory;  he has removed the thing the theory is about.

Hartnett had been arguing against metric expansion since at least 2011, in a two-part Journal of Creation review, and again in Answers Research Journal in 2014.²⁹³⁰  The 2024 post is where he states that the earlier conclusion is what finished the Carmelian model.

The trajectory on cosmic acceleration, laid out by date

One thread runs through the whole story and it is worth pulling out separately, because it shows how far the position moved.

2005.  Hartnett's TJ paper treats accelerated expansion as a triumph of the theory he is using.  His own sentence, on page 78:  "With this model, Carmeli was able to successfully predict an accelerating expanding universe, two years before it was measured from high-z supernova redshift observations in 1997–1998."¹⁷  In the same paper he describes Carmeli's equations as giving a three-stage expansion history:  deceleration, a moment of coasting, then acceleration.¹⁷

2006.  The Foundations of Physics paper fits the high-redshift supernova magnitudes in the Carmeli framework and reports the fit as successful.⁴  The whole point of the exercise is that the accelerating universe can be reproduced without dark energy.

2011.  In Journal of Creation he publishes the case against cosmic expansion, concluding that "the simple non-expanding Euclidean universe fits most data with the least number of assumptions."²⁹²⁹  He was still publishing Carmelian expanding-universe papers at the time.

2013.  The bounded Carmelian solution appears in the International Journal of Theoretical Physics, and it is an expanding-universe model.⁵

2014.  "Expansion of Space:  A Dark Science" in Answers Research Journal, arguing against metric expansion, published in the same journal where Danny Faulkner argues the opposite.³⁰

2015.  Creation magazine, on dark energy:  "Supposedly, dark energy is a confirmed fact.  But does the evidence confirm that the universal expansion is accelerating?"³¹³¹  Ten years after crediting Carmeli with predicting acceleration, he is asking whether acceleration happens.  What follows in that article is Richard Lieu's argument that cosmology cannot be an experimental science, not an examination of the evidence for acceleration.  Baryon acoustic oscillations, the integrated Sachs-Wolfe effect and cluster abundance evolution are not discussed.  The dark energy article works through what he left out.

2024.  "I came to the view that the universe is not expanding but Carmeli's cosmology describes only an expanding universe."²²

2025.  He still cites his own Carmeli supernova fit, but for a different purpose.  It is now an example of underdetermination, offered to show that a non-dark-matter fit to the same data exists, not a model he holds:  "One such example of that is the work I did myself on the Carmeli cosmology that showed excellent fit to the same high-z SNe data without any DM… Both of these cosmologies are expanding universe cosmologies."³²³²

There is a thirteen-year window, from 2011 to 2024, in which Hartnett was publishing arguments against cosmic expansion while his own cosmology required it.  He does not appear to have addressed the tension in print until the 2024 post.  A skeptic is entitled to notice that, and so is a creationist.

Where Hartnett stands now

He is still a young-earth creationist.  He did not leave the position;  he left the model, and he says which model replaced it.

His answer since 2019 has been Jason Lisle's anisotropic synchrony convention, which holds that the one-way speed of light toward an observer is a matter of convention, and that Scripture uses the convention in which it is infinite.  On that convention there is no light-travel-time problem at all, because there is no travel time.  Hartnett has restated this repeatedly and recently.  On 25 June 2026:

"My current thinking is that the ASC model, as initially promoted by Dr. Jason Lisle, is the best solution.  Under the assumption of the Anisotropic Synchrony Convention (ASC) we observe the whole universe in real time because that convention assumes the incoming speed of light is infinite towards the observer.  Hence there is no travel time and no problem."³³³³

He is explicit that the price of this is created maturity:  "The ASC model necessarily assumes what we would call mature creation.  That is the galaxies in the universe were created essentially the same condition as we currently see them now."³³  See the anisotropic synchrony convention article for what that convention does and does not buy.

He also holds two positions that put him at odds with most of the young-earth cosmology community.  He thinks space does not expand, which contradicts Humphreys' white hole, Faulkner's models, and his own former one.  And he thinks cosmology is underdetermined in a way that makes any cosmological argument weak, in either direction.  His 2011 Journal of Creation conclusion is the clearest statement of it:  "it would be a mistake to base one's theology on any particular cosmology."²⁹

That sentence, from the author of the most technically developed young-earth cosmology ever published, is the most useful thing in this article.

What the model would have needed to succeed

Set aside the retractions for a moment and ask what it would have taken for the Carmelian creation cosmology to become a serious scientific claim rather than a possibility.  Four things, none of which arrived.

A five-dimensional solution.  The creation scenario needs space, time and velocity treated together, near the creation epoch, for the whole universe.  Carmeli produced a four-dimensional space-velocity model and never got further.  Hartnett's Appendix 6 is a linearised extension, and he called the result "only a sketch of what I hoped would crystallise."¹⁶  Twelve years of work did not close the gap, and Hartnett names this failure in three of his five withdrawals.

A physical account of the velocity dimension.  Not a mathematical role, a physical meaning.  Hartnett asked for one and never found one.  Carmeli treated it as redshift.  A coordinate whose interpretation is "the thing we measure that we are trying to explain" is doing bookkeeping, not physics.

A prediction that ΛCDM does not make.  The supernova fit is not one, because Öztaş and Smith showed several incompatible models fit the same 580 points.²⁴  The bounded solution is not one, because Hartnett showed it is identical to the unbounded solution.⁶  The rotation-curve result is the closest thing to a prediction in the whole programme, and it is a shared prediction with MOND at a different value of the constant.

A treatment of the microwave background.  Hartnett conceded he did not have one:  "I do not know its true source."¹²  The power spectrum of the microwave background is the single most information-rich dataset in cosmology, and no young-earth model has ever been fitted to it.  Andrew Repp made the same point in the Creation Research Society Quarterly in 2022.³⁴³⁴

Judged against that list, the fair verdict is not that the Carmelian cosmology was refuted.  It is that it was never finished, and its author was the first to say so.

Objections and responses

"You are quoting a creationist against creationism.  That is dishonest."  Every negative verdict here is Hartnett's own, published under his own name, on his own website, and this article says so at every step.  The alternative is worse:  keep citing Starlight, Time and the New Physics as a live answer when its author has told his readers five times to stop.  A movement that lets its own retractions be quietly forgotten is the one behaving dishonestly.

"He abandoned it for personal or political reasons, not scientific ones."  The 2021 post about Creation Ministries International is indeed mostly about an editorial and vaccine dispute, and this article flags that.  But the technical reasons are in the 2015 and 2016 posts, which predate the dispute and concern the mathematics:  an invalid simplification inside an integral, an inconsistent lensing derivation, sections of Carmeli's last book that had to be deleted because they made no sense, and the missing five-dimensional solution.  Those are physics reasons, given first, by the person who did the work.

"Carmeli's theory is still viable even if Hartnett's creation model is not."  Possibly, and that distinction is worth keeping.  Carmeli's four-dimensional space-velocity model still fits the supernova data, and Öztaş and Smith reproduced that fit independently in 2014.²⁴  What has not survived is any claim that this counts as evidence.  The same data fit several incompatible models, the Hubble law is put in by hand, the theory has never been fitted to the microwave background or to nucleosynthesis, and the person who developed it furthest says the special-relativistic foundation is "fundamentally flawed."²

"Mainstream physicists ignored Carmeli because it threatened the standard model."  Hartnett said something close to this, and it is partly true:  his own account is that the theory was ignored because it was "a significant departure from the standard model."²  But a theory that is ignored is not thereby vindicated.  Carmeli's papers were on arXiv, in Foundations of Physics, and in the International Journal of Theoretical Physics.  When someone did read them carefully, the result was Macdonald's comment, which found an elementary error in the time-redshift relation and got no published reply.²³

"The bounded solution proves the universe could have a centre, which is what creationists have always said."  It proves that a centred universe is not excluded by the supernova data.  That is worth something and it is not worth much, because the same result says a centred universe is not supported by them either.  George Ellis, whom creationists quote for the first half of this, states the general point and then declines the conclusion:  he is showing "the nature of the best observationally-based argument by which we can (quite reasonably) justify the assumption of spatial homogeneity," not defending an Earth-centred cosmos.²¹

"So does distant starlight refute a young universe or not?"  This article does not settle that, and no honest treatment of one model could.  What it settles is that the Carmelian answer is not available.  The survey of all eight published answers works through the rest.  The short version:  created light in transit implies a sky full of events that never happened, the decaying speed of light was refuted by a creationist's reanalysis, and the anisotropic synchrony convention survives but changes no observation by its author's own account.

"Hartnett is a crank.  Look at what else he writes."  His blog since about 2020 carries material on vaccines, finance and geopolitics that has nothing to do with physics and that this project does not endorse or rely on.  None of that touches the physics.  His experimental work on cryogenic oscillators stands, his Astrophysics and Space Science paper killing the quasar-periodicity claim stands, and his retraction of his own cosmology is more careful and better documented than most retractions in any field.  Judging the physics by the blog would be the same error as judging it by the ministry that promotes it.

What would change the verdict

A published reply to Macdonald.  The comment appeared in Foundations of Physics Letters in November 2006.  No response by Carmeli, Hartnett or Oliveira has been located.  If one exists, it belongs in this article.

A full five-dimensional Carmelian solution.  Hartnett names its absence as a reason in three of his five withdrawals.  If someone produces one, the creation scenario becomes testable again for the first time since 2007.

A Carmelian fit to the microwave background power spectrum.  Nobody has attempted it.  Repp said so in the Creation Research Society Quarterly in 2022 and it remains true.³⁴

Direct measurement of the MOND acceleration constant.  Carmeli's theory predicts a₀ = cH₀ with no freedom.  MOND fits a₀ = 1.2 × 10⁻¹⁰ m s⁻² to rotation curves.³⁵³⁵  Those differ by a factor of several.  A sharper measurement of the low-acceleration scale across many galaxies is a live test of Carmeli's version specifically, and nobody in the creationist literature has run it.

Evidence that space does not expand.  Hartnett's current position rests on it, and so does Jason Lisle's Doppler model.  The Tolman surface-brightness test and the time dilation of distant supernova light curves are the standard measurements on this question, and the redshift article works them through.

What this case shows

Three of the young-earth cosmologies with real technical content have now been abandoned by the people best placed to judge them.

Barry Setterfield's decaying speed of light was refuted inside the movement by Gerald Aardsma, then head of the Institute for Creation Research's astro/geophysics department, whose weighted refit of the same 163 historical measurements gave a decay rate of 0.0000140 ± 0.0000596 km/s/year, consistent with zero.³⁶³⁶  Russell Humphreys built a second cosmology and then a third;  Danny Faulkner reports that "Humphreys has since abandoned his model," and Jason Lisle reports the same from a conversation with him.²⁵³⁷  John Hartnett withdrew his own model five times in writing.

None of that is evidence against a young universe.  It is evidence about the state of a research programme, and it points in one direction:  the technical young-earth cosmologies do not survive contact with their own authors' second thoughts.

The honest thing to say to a skeptical reader is that the light-travel-time problem is unsolved, that the movement's own survey literature says so, and that the most careful physicist who ever worked on it now holds a position that makes no cosmological claim at all.  Hartnett's sentence from 2011 is the right place to end:  "it would be a mistake to base one's theology on any particular cosmology."²⁹

Sources