Really awesome detection of 21cm absorption in the early universe (z=17). I’m very impressed that this measurement can be made. Surprising result - to most cosmologists - is how strong it is - about 2.5x stronger than expected in LCDM.
This does not surprise me. It is exactly the kind of test I proposed 20 years ago based on the cosmic baryon fraction. If LCDM is correct, fb=0.16. If dark matter is a chimera indicative of some modified gravity, then really fb=1. xxx.lanl.gov/abs/astro-ph/9…
This new result is nice because it provides a clean test. The rest of the physics is basically the same (radiative transfer) in either case. The baryon fraction is the one parameter that predicts a big difference. The strength of the signal depends on the 1/sqrt(fb).
In LCDM the predicted expected signal is -0.22K at z=17. If the universe is all baryons, the signal is 2.5x stronger, i.e., -0.55K. The claimed detection is -0.5K.
Looking at it, the signal looks closer to -0.55K. Bang on the all-baryon expectation. 
Of course, fb=1 doesn’t occur as a piece of parameter space to check if you’ve already convinced yourself that the discrepancies we see HAVE to be due to dark matter and CAN’T indicate a modification of gravity. I get that. I struggle with that all the time.
I tried to look at it from both sides in my review for CJP arxiv.org/abs/1404.7525 There are positive points to be made for both perspectives. If you thought his issue was settled, it is probably because you weren’t paying attention. Hard to change your mind from that perspective
Anyway, fb=1 is the most natural explanation for this observation. If we insist on retaining dark matter, we have to use some indirect lever to change the spin temperature of the gas. So we add non-standard physics to save the non-standard physics of dark matter.
Apparently one thing that can be made to work is have the baryons scatter off the dark matter in the early universe so as to change the spin temperature. Because baryon-dark matter scattering has been so prolific in the laboratory experiments designed to look for exactly this.
That pushes the DM particle properties into unexpected regimes of low mass. Some of these can be probed experimentally, but it is hard. I seem to remember a theoretical minimum around 2GeV. That’d help limit the field! But no doubt it can be evaded.
So: other predictions for a baryonic universe (all made around 20 years ago): baryonic oscillations should be frozen in the plasma, so the acoustic power spectrum of the 21cm absorption should have lots of sharp features and look very different from CDM.
Structure grows fast and becomes non-linear early. L* galaxies are forming already by z=10 (Sanders 1998). The cosmic web is similarly in place unexpectedly early. As are big clusters. Everything happens quickly, and the voids get swept clear.
I outlined my predictions a long time ago; see astroweb.case.edu/ssm/mond/LSSin… if you want to Case up some references.
Mostly I’m pleased that the universe remains an interesting place - still full of surprises!
Mostly I’m pleased that the universe remains an interesting place - still full of surprises!
The model that *fit* the second peak predicted a third peak that was larger than the second. This also was not observed: the third leak is about the same amplitude as the second. The tilt is the primary knob that fixes this, and also brings he fit back towards BBN (but not entire
If you don’t believe me, DDO the following exercise yourself: take the WMAP3 best fit model, and then plot the WMAP5 data. Then use 5 to predict 7 and so on. The data match great - where already fit. Not so much beyond that. Close of course, but not snack on.
That means it is worthwhile to keep observing to higher multiplies: we do learn new things. But we can’t accurately predict them. Once you get that right, it is hard to miss the polarization.
I predicted the 2nd peak correctly a priori. The same model fails to correctly anticipate the third peak. This is often portrayed as a falsification of MOND, but this is an exaggeration - all it does is falsify the ansatz I made to represent MOND, not MOND itself.
I knew the ansatz had to brew down somewhere. That somewhere is apparently l=600. I admit I do not understand this. At least I’m not committing the logical fallacy of the straw-man argument by asserting that my ansatz is exactly MOND. Yet otherwise intelligent people do just this
So no, I do not agree that CMB fits, brilliant as they are, count as an a priori prediction. An example of that would be the velocity dispersion of Crater 2 and the dwarf satellites of Andromeda. I can look at those and tell you what their velocity dispersion will be - in advance
But only if I use MOND. I can’t do it with LCDM. I can always explain it afterwards (“there is X mass in dark matter”) but the prediction of MOND here is unique while that if LCDM is not. Near as I can tell, it is impossible to use LCDM to make this prediction (strictly defined).
So I think we are very far up the proverbial creek. Neither paddle can be made to work everywhere all the time. And as a community we’re paddling hard in the wrong direction.
