This post is motivated by recent finding of Rachel Bean, who found, various WMAP, 2MASS, SDSS, COSMOS data concerning the Sachs-Wolfe, galaxy distributions, weak lensing shear field, and the cosmic expansion history doesn't fit general theory of relativity (GR for short). The reactions of Sean Carroll and/or Lubos Motl are careful, as someone may expect : "well, this could be challenging - but probably irrelevant, because GR has proved itself so many times, but the science should care about such details, mumbojumbo..."
Jeez - but how GR was derived before eighty years? This theory puts an equivalence between curvature of space and spatial distribution of energy of gravitational potential, as borrowed from Newton's theory (because we really have no better source for function of gravitational potential with distance, then the forty years old gravitational law). So, if we know the mass of object, we can compute the spatial distribution of potential energy, so we can compute the spatial distribution of space-time curvature - end of story (of GR). Or not?
Not at all, because from the very same theory follows, energy density is equivalent to mass density by E=mc^2 formula - so we are facing new distribution of matter in space, which should lead into another distribution of space-time curvature and energy of gravitational potential curvature, which leads to another distribution of matter, and so on - recursively. Such implicit character of GR was never mentioned in classical field theory of GR and corresponding textbooks - so it's nothing strange, it violates all observations available by now. But it's still prediction of GR postulates and it fits well with fractal implicit character of Universe and AWT - it just requires to derive Einstein's field equations more consequently and thoroughly.
Wow, this could be really breakthrough in physics and challenging task for new Einstein - or not? Of course not - and here we come to real problem of contemporary science - because such approach is fifty years old already and its even used in dark matter theory, in fact. Such modification would lead into quantization of gravity and longly awaited quantum gravity - the only problem for formally thinking physmatics is, it brings a quantum chaos into ordered world of formal relativity too, as there is (nearly) infinite number of ways, how to derive it - and all ways are still only approximations of real situation. The names like Cartan, Evans, Heim, Yilmaz, J. Bekenstein or Rudi V. Nieuwenhove are all dealing with this approach in less or more straightforward form - but this cannot change thinking of incompetent, though loudly blogging people, who invested two or more years of their life into learning of GR derivations, until they become "productive" with it (as measured by number of articles published) - so now they simply have no time and/or mental capacity to understand something new, to extrapolate the less.
Of course it's not just a problem of few desoriented bloggers, but inertia of whole mainstream community, the size of which prohibits introduction of new ideas and which has chosen formal approach to classical theories as a salary generator for their safe life. In this way, every new idea or derivation is simply forgotten, until it's revealed again in another, slightly different connection, when everyone appears surprised, how is it possible, GR isn't working properly?