I'm going to post something I wrote elsewhere. This addresses some of the discussion here, but I'm to lazy to completely rewrite it. Apologies for arriving late:
First thing to stress, the paper is not using standard cosmology. Secondly, if you open the paper you will see that it contains exactly zero JWST data. Some have got the impression that this paper implies JWST results are in conflict with standard cosmology, but it's not that at all.
So what is their calculation based on? The answer to that is many assumptions pulled out of thin air, with little regard for observations or physics. Their basic assumption is that the massive elliptical galaxies we see today in the local universe all formed immediately in the early universe, redshift 15 to 20. So even higher than any confirmed JWST galaxy. In their model, the take an simplistic model of galaxy formation (monolithic collapse), which is sometimes invoked by MOND people. Note that in standard cosmology you have hierarchical structure formation where small galaxies formed first, they are essentially assuming top down growth, it's not standard at all. They also assume that these galaxies are miraculously totally enshrouded by dust. The dust is then conveniently destroyed. They then get a "background" which is comparable to the CMB. The problem is it's all based on these untested assumptions.
The paper never asks if these assumptions actually correspond to reality. For that, we can turn to observations. Firstly, it violates JWST observations. The most distant confirmed galaxy has a mass of 10^8 solar masses in stars. These galaxies would be 10^11.5 and above. There are no high redshift candidates that are anywhere near this massive. In their model, these become passive (quenched) galaxies at lower redshift. The number density of such objects measured by JWST is at least a factor of 100 lower than they require, even ignoring the fact that there is this huge discrepancy in mass. They don't consult JWST luminosity functions, or number densities, they don't ask if what they assume is real. It would also require the number of massive galaxies to be constant at all redshift, whereas observations show a steep decrease. Based on JWST alone, it's ruled out.
The paper doesn't bother to calculate what spectrum these galaxies would emit. One heading within the paper says that dusty galaxy spectra resemble blackbody, which is true, but they're also clearly wrong. In real dusty galaxies you get absorption and re-emission by the dust (radiative transfer), giving rise to a modified blackbody (greybody). The authors have confused dust temperatures with real blackbodies. My image is the far infrared SED of a local ULIRG Arp220. The red line is an assumed blackbody, the black dotted line is a modified BB. If the CMB were significantly contaminated by dusty star forming galaxies, it would mess up the spectrum. There are extremely tight limits on the spectrum from COBE FIRAS.
The paper says observers should look for this contamination. But they already have. Something which is (incredibly) not even mentioned is the Cosmic Infrared Background. The CIB is the cumulative effect of dusty star forming galaxies over cosmic time. The CIB is not like the CMB in that with enough resolution it can be separated into individual galaxies, it's also not a blackbody. The CIB has about 3-4% the energy compared to the CMB. In a press release Kroupa makes the suggestion that maybe this light is the whole CMB. Which is just wrong. In the paper they calculate there are just 6 source galaxies per Planck resolution element. So higher resolution telescopes like SPT, ACT, LMT and ALMA would easily resolve this background into individual galaxies. Which doesn't happen. If it were true ALMA would not be able to use the CMB to measure clusters via the SZ effect, because it would resolve it into one or two bright galaxies.
The paper is just a simple calculation based on many assumptions that are never justified. None of it is motivated by observations. It can also be rejected just with the data that exists today. It's missing cosmological context. In galaxy formation simulations it takes a long time for fluctuations to grow until galaxies can collapse. They aren't using simulations to calculate the full physics.