Here's some more from wikipedia writings on non-standard cosmologies

Redshift, AGN, and Quasars
In the meantime, there are other issues that some non-standard cosmologists insist must also be considered. A good example is the observations made since the 1960s by the astronomer Halton Arp, which offer an alternative to the standard interpretation of quasar formation, redshift and Hubble's Law.

Arp has observed a handful of correlations between quasars (and more recently, X-ray sources from Chandra data) and AGN (Active Galactic Nuclei) which he claims demonstrates that quasar redshifts are not entirely due to the expansion of the universe, but contain a local, or non-cosmological, component. Arp claims that clusters of quasars have been observed around many galaxies (examples include NGC 3516 (http://www.haltonarp.com/?Page=Images&Image=4) and NGC 5985 (http://www.haltonarp.com/?Page=Images&Image=5) as well as M51, NGC 7603, NGC 3370, NGC 4319, NGC 4235, NGC 4258) which all have some properties in common:

The active galaxy always has a lower redshift than any of its associated quasars.
The quasars tend to lie within a narrow conical zone centered about the minor (rotational) axis of the associated active galaxy.
Schematically, the quasars' redshifts are inversely proportional to their angular distances from the AGN, i.e. as apparent distance from the AGN increases, the redshift of the quasars decrease.
Some of the quasars occur as pairs on either side of an AGN, particularly the X-ray sources appearing in the Chandra data.
Some astrophysicists believe that gravitational lensing might responsible for some examples of quasars in the immediate vicinity of AGN, but Arp and others argue that gravitational lensing cannot account for the quasars' tendency to align along the host galaxies minor axis.

These observations indicate to Arp that a relationship may exist between quasars (or at least a certain type of quasar) and AGN. Arp claims that these quasars originate as very high redshift objects ejected from the nuclei of active galaxies, and gradually lose their non-cosmological redshift component as they evolve into galaxies.

The biggest problem with this analysis is that today there are tens of thousands of quasars with known redshifts discovered by various sky surveys. The vast majority of these quasars are not correlated in any way with nearby AGN. Indeed, with improved observing techniques, a number of host galaxies have been observed around quasars which indicates that those quasars at least really are at cosmological distances and are not the kind of objects Arp proposes. Arp's analysis, according to most scientists, suffers from being based on small number statistics and hunting for peculiar coincidences and odd associations. In a vast universe such as our own, peculiarities and oddities are bound to appear if one looks in enough places. Unbiased samples of sources, taken from numerous galaxy surveys of the sky show none of the proposed 'irregularities' nor any statistically significant correlations that Arp suggests exist.

In fact, the question of whether quasars are cosmological or not was an active controversy in the late 1960s and early 1970s, but by the late 1970s most astronomers had considered the issue settled. The main argument against cosmological distances for quasars was that the energy required was far too high to be explainable by nuclear fusion, but this objection was removed by the proposal of gravity powered accretion disks.

In addition, it is not clear what mechanism would be responsible for such high initial redshifts, or indeed its gradual dissipation over time as the quasar evolves. It is also unclear why objects ejected from a galaxy should never seem to produce a blue shift. Moreover it is unclear how nearby quasars would explain some features in the spectrum of quasars which the standard model easily explains. In the standard cosmology, the clouds of neutral hydrogen between the quasar and the earth at different red shifts spikes between the quasar redshift and the rest frequency of Lyman alpha in a feature known as the Lyman-alpha forest. Moreover, in extreme quasars one can observe the absorbion of neutral hydrogen which has not yet been reionized in a feature known as the Gunn-Peterson trough. Most cosmologists see this missing theoretical work as sufficient reason to ignore the observations as either chance or error. Arp himself proposes Narlikar's variable mass hypothesis, which contains alternative explanations of various observed cosmological features, but it remains, at best, incomplete.

A consequence of Arp's proposed AGN-origin of quasars would be that quasars would be much closer, much larger, and much less luminous than currently supposed and their heavy element composition would no longer require primaeval Population III stars. Such a theory would predict that the heavy element composition of quasars would be similar to the associated AGN, though observed metal lines in quasars are notoriously weaker than AGN. Variable luminosity and absorption phenomena such as the Lyman-alpha forest would both be explained by as yet theoretically undeveloped "local means".

A further anomaly comes from the magnitude-redshift relation first discovered by Hubble. Plotting absolute galactic magnitudes against their redshift produces a clear linear relation, which in 1929 led Hubble to propose an expanding universe and Fritz Zwicky to propose the tired light hypothesis. However, quasars were discovered much later, and the same plot done using quasar data produces a much more diffuse scatter with no such clear linear relation. However, since the absolute magnitudes can only be calibrated using a size constraints from variability and an Eddington luminosity limit, it is likely that quasars are exhibbiting differing absolute luminosities that cannot neccessarily be derived from such simplistic first principles. Arp, Burbidge, and others maintain that the scatter in these plots further supports the idea that quasars have a non-cosmological component to their redshift, but nearly everyone else in the field accepts that quasars have variable luminosity.