"if I'm allowed to make stuff up whenever I want to make my theory fit the model, I can do at least as well as the Lambda CDM"
Go ahead - you're more than welcome to. Empty assertions don't show much but new cosmological models are welcomed. *I* welcome them, anyway; I've never liked Lambda CDM much and it's obviously a phenomenological model. But they have to be predictive, and founded on firm principles.
I didn't actually want to suggest you're an idiot because I think it's apparent you're not, but this type of post at the same time implies that *cosmologists* are idiots and brainwashed into a model that doesn't really make much sense. And in some cases that's actually true -- there are more and more cosmologists trained into cosmology rather than general relativity and it's a bit dangerous -- but on the whole I don't think many people *like* LCDM. There are too many unanswered questions in it, and everyone is looking to answer those. Just some people work more tightly within its framework than others.
"is there a point where you would ever consider reexamining the questions of the assumptions? Why haven't we reached that point yet?"
Oh, don't misunderstand me -- I *constantly* question and re-examine the assumptions. At some point, if you're genuinely interested, flip back through my posts on Slashdot; I've made my position I think fairly clearly. Boiling it down and putting it in bullet form it goes something like this:
* The "big bang theory", and Lambda CDM in particular, is an astonishingly successful theory, particularly when attached to an inflationary period in the early universe or something that mimics its observational results closely
* The successes of Lambda CDM -- such as the predicted abundances from Big Bang nucleosynthesis, the *prediction* of the angular power spectra of the CMB (temperature auto-correlation, temperature/E mode cross correlation, E mode polarisation auto-correlation and now the B mode polarisation auto-correlation) from a simple early primordial power spectrum, the direct mapping between the wavelength of the sound horizon at last scattering as seen on the CMB and that same wavelength imprinted on large-scale structure and *observed* as the baryon acoustic oscillations, and their ilk -- are far too numerous and significant to be ignored.
* Any alternative absolutely has to preserve these, and they're all extremely sensitive
* Lambda CDM is wrong. It is dead wrong. It is wrong in principle. It is questionable from a particle physics perspective, particularly where it comes to dark energy, but far more importantly, it cannot be justified with general relativity.
Lambda CDM rests on a few main assumptions:
* The universe is on average isotropic around the Earth. OK, fine, we can't argue that; the CMB is proof enough.
* Since the Earth is nowhere special, the universe is on average isotropic around every point: homogeneous. Well, this is debatable since the Earth *is* in a particular position, but on the whole this is probably at least approximately true.
* Gravity is best described on large scales (ie > mm) by a metric theory. This is currently practically unquestionable; metric-based theories of gravity are vastly more succesful than any alternative.
* Gravity is described by general relativity. OK, now we're entering questionable territory but GR remains our best example of a metric-based theory and is yet to be seriously challenged (though there are many, myself among them, who point out that the appearance of dark matter on galactic scales, and the addition of dark energy on cosmological scales, may very well imply that actually we cannot apply gravity on such scales or else that it simply doesn't act this way on large scales)
* GR can be applied directly on large scales. This is extraordinarily shaky. Actually, it's unjustifiable. We've got two main objections here: firstly, there is no reason to assume that gravity actually obeys GR on large scales. The direct evidence we have stretches up to, I don't know, 100AU or so. We're extrapolating that not just to a parsec but to a kiloparsec for galactic scales, a megaparsec for cluster scales, and a gigaparsec for cosmological scales. That has to be justified. But far more importantly, *in GR it is impossible to define an averaged field*. The word "on average" and "on large scales" is peppered through these assumptions. Try and define an average in GR and you'll burn three or four years of your life and end up with something brutally non-covariant and which, worse, only applies on globally hyperbolic spacetimes. Which more practically means it cannot apply whenever there is geodesic crossing -- whenever light rays can cross one another. But that's *precisely* what we have in GR. We have gravitational lenses out there. A single gravitational lens fucks up an averaging procedure, and the universe is riddled with them. For this sole reason, cosmology as an absolute, provable science is currently totally fucked.
Now what this does not mean is that cosmology is worthless or meaningless. The success of Lambda CDM + inflation demands that it be taken extremely seriously. Indeed, up to a redshift of call it 2 or 1 then I wouldn't even think of using an alternative. But at lower redshifts it gets increasingly shaky. We can recreate some aspects of Lambda CDM in pure dust spacetimes simply by swapping from the Robertson-Walker metric. We can prove that the idea of employing the RW metric at all is unjustifiable. That much we can do. What we can't currently do is suggest an alternative, so we're stuck for now with LCDM - because there *is nothing better* that can be founded as securely on reasonable physical bases. (And I even include dark energy in "reasonable physics", with a bit of a twinge.)
Sure, I question the assumptions constantly. So do plenty of others. GR is replaced constantly -- check for f(R) theories and Horava-Lifshitz gravity for two recent explorations in totally different ways to look at gravity, and which produce very different cosmologies that mimic LCDM closely. People consider inhomogeneous models -- non-RW metrics -- or sit there and try and average things more securely. People hack at the underpinnings constantly. But in terms of a standard model we can all look at, like it or not (and many do not), we can't beat LCDM.