It would clearly be an exageration to say that the third peak amplitude has been measured. Nevertheless, the Boomerang data and the last credible point in the 3-year WMAP data both have power that is clearly in excess of the no-CDM prediction. The most natural interpretation of this observation is forcing by a mass component that does not interact with photons, such as non-baryonic cold dark matter.
This appears to falsify the simple no-CDM model. Until now, this has served as a useful proxy for MOND. I'm sure some people will leap at this to infer that MOND itself is falisfied, but this would be an overstatement. As many of those same people rushed to remind me when the no-CDM prediction provided and accurate, a priori prediction of the second peak, MOND itself doesn't make a prediction here. Only the no-CDM proxy does, and we knew that had to fail at some point.
However, I do find this particular mode of failure rather hard to understand. Perhaps it can be explained in relativistic theories like TeVeS. This remains to be seen, but seems unlikely to be satisfactory.
I do not find the result satisfactory for CDM either. One wants a clear answer: either the third peak is high (CDM) or it is low (no CDM). Instead, the data split the difference. They come in well under the extrapolation of the LCDM fit to the WMAP first year data, which misses predicting the high L points by nearly as much as the simple no-CDM model. There is still plenty of flexibility in CDM models - they remain far from over-constrained - so it is quite possible to fit this. In this case, the most natural tweak is in the tilt (n = 0.95), as anticipated by Boomerang (n = 0.86!). It is worth noting that the small scale amplitude of the power spectrum has continued to move down (both smaller n and smaller sigma8) as demanded by rotation curves (McGaugh, Barker, & de Blok 2003 - indeed, the new WMAP data are now consistent with the limits on cosmological parameters we derived, albeit only in the extreme limit of zero stellar mass).
We have come to an unfortunate cusp. If one is already disposed to believe in CDM, then it is natural to take the WMAP data as further confirmation of the pre-existing belief. By extension, one may feel comfortable assuming that any difficulties faced by CDM in galaxies are details that will be resolved because we know the basic picture is right.
I don't think so. I have shown repeatedly that it is very hard to understand the systematic properties of rotation curves in terms of CDM. It is too much to say that those data falsify CDM, because CDM makes no falsifiable prediction for rotation curves just as MOND makes no falsifiable prediction for the CMB. If we are disposed to believe in MOND, we are equally well justified (and equally wrong) in feeling comfortable that any deviation from the no-CDM model in the CMB will be explained in the context of a relativistic theory like TeVeS, for the same reason: we know the basic picture is right.
The simple fact is that neither paradigm now gives a satisfactory explanation of all the data. It is very far from obvious that either can.