Your definition of "prove" is too narrow. From my online WordWeb dictionary:
Verb: prove (proved,proven) 1. Be shown or be found to be 2. Establish the validity of something, as by an example, explanation or experiment 3. Provide evidence for 4. Prove formally; demonstrate by a mathematical, formal proof 5. Put to the test, as for its quality, or give experimental use to ...etc...
Many senses of this word relate to subjecting something to a test, rather than proof in the mathematical sense.
Now the long term stability of CO2 vs. methane is based on simple chemistry, and water rains. As to why CO2 would increase the greenhouse effect see: http://en.wikipedia.org/wiki/Absorption_spectrum and its overall concentration in the atmosphere.
PS: Looking back in history the accumulation effect of CO2 works in reverse so other man made impacts could have easily dominated it.
Like most complex scientific theories there are several steps of logic involved. And while a few groups agree on which specific steps break the chain a slightly larger but still small group of people can agree that it's broken.
All modern climate "models" are horrible hacks. All contain one or more empirically determined "fudge factors". All gloss over numerous critical important elements of climate dynamics (such as clouds) because they are too difficult to model currently, or we lack understanding of their mechanisms.
The idea that anyone anywhere on Earth has some pristine from-first-principles climate model that you can input data into, turn some crank and spit out data on the climate of the future with any degree of certainty is just plain wrong. Nobody has an accurate model of climate. The AGW proponents have been claiming that they have created accurate climate models (to a degree), but the evidence to back that up is extremely thin. And, what's more, we've discovered recently that the evidence is even thinner than we thought it was.
This is why people object so much to the AGW doomsayers, because the evidence is very much lacking. At best, at best, there's a tiny window of a few decades where some models of climate show a similar trend in global average temperature to the actual trend. And on this tiniest thread of "evidence" so the entire catastrophic AGW theory and all of its proscriptives on human behavior and industry hangs.
Not that I ever said that you don't need to subject either to tests. You absolutely do. But if you think you can "prove" a scientific theory, I think it is you who is showing a profound misunderstanding of the scientific method.
You can never prove a scientific theory, you can only disprove it. You can confirm a theory a million times but that does not imply that it will work the next time. It does however give us some amount of confidence that it will.
My point is that as far as scientific statements go, that AGW happens is like saying "I think the ball will fall", while to say that it doesn't happen is to say "I think the ball will fall, but imperceptibly slowly.". Both should be tested, but given that one agrees with known physics and one does not, I think the latter should be viewed with more suspicion.
But maybe that's just me.