It occurred to me that this thread was in need of some expert attention, and so I am glad to see that marcus stepped in. I don't claim to be an expert, but I am able to comment on a couple of the ideas that were raised in this thread:
1. The "Tired Light" Hypothesis
This was proposed as an
alternative explanation for the observed redshifts of distant objects (i.e. instead of cosmological expansion). It was not really meant to be something that went along with expansion as sophiecentaur suggested in post #2. In any case, to answer mquirce's question in post #10, the tired light hypothesis has been thoroughly ruled out. There are a couple of fairly obvious reasons why it can't work, including:
i) Scattering at this level would require that distant objects be blurred, with the blurring getting worse the farther out you looked. We don't see this.
ii) The scattering and/or absorption would have to be frequency dependent. We don't observe this. The amount of redshift is independent of the original wavelength (in the rest frame of the emitter). The entire spectrum of the emitter gets shifted by the same amount
iii) More generally, the tired light hypothesis just doesn't explain a broad range of observations as well as big bang cosmology does (e.g. the CMB, primoridal abundances of light elements, large scale structure, Type Ia supernova data etc. More details here:
http://en.wikipedia.org/wiki/Tired_light#Criticisms).
That's not to say that there isn't any absorption or scattering of photons along the line of sight at all. It just isn't the dominant effect and can't adequately explain the observed cosmological redshifts.
2. Event Horizons
It may be true that light from some objects is redshifted to the point that we can no longer see it, but that is not the sole answer to the question of why we may never see the light from some distant galaxies. Depending on what cosmological model you use, there may very well be a distance beyond which light will
never reach you, at any time in the future. We call this distance the radius of the event horizon, since you as an observer can never have information about events that take place beyond it. This is not the same thing as the horizon that cosmologists speak about more frequently, which is the particle horizon: the maximum distance that light can have traveled so far (since the beginning of the universe). The particle horizon sets the size of the region of everything that you can see
now, but it expands to encompass more and more volume as time goes on. The particle horizon scale is therefore the radius of our
currently observable universe (which is how you may also have heard of it referred to).
Getting back to event horizons -- again, whether or not one exists depends on your cosmological model (i.e. the parameters of the universe you think you live in). For some models, there is no finite event horizon radius. It turns out to be infinite, meaning that if you could just wait long enough (i.e. forever) then you would eventually see everything. However, I seem to recall (and I will double check this) that our standard lambda-CDM cosmological model does have a finite event horizon. Due to the accelerating expansion, there is a distance beyond which you will never be able to see. As for how this can be reconciled with the argument the OP made about how we should always see photons traveling towards us at c, I suspect it has something to do with general relativity that he/she wasn't taking into account (sorry I don't know enough to explain further). I should point out that the OP failed to take into account that the universe is expanding, so that the distance you would infer from the light travel time does not correspond to the proper distance to the object (neither at the time when the light was emitted, nor at the time when it was received).