Yeah, I've caught a lot of flack for this in the past :/

Basicially, by evolution, I'm not referring to the overly-broad "descent with modification" - we've know that a child will look different from its parents for a good while now.  Nor do I mean by evolution even natural selection combined with random mutation (I acknowledge both of those as true).  By evolution, I mean the idea that, given natural selection and random mutation, a bird could eventually have a fish for a descendent.

I believe you have failed to understand the meaning of these words in some cases.

Observability. This does not mean that a process has to be observed as it occurs, if historical evidence of something exists observing that historical evidence is still observation. It is without a doubt that science is based purely on observed evidence - whether this evidence is observed as soon as the process has taken place or whether the evidence is observed millions of years after is inconsequential.

The fossil record would be considered historical evidence, yes?  In that notion, yes, I would naturally agree that the fossil record exists.  What does not, however is any hypothesis I come up with that includes it.  That is the unobservable part

Testability. The ability to make and test predictions does not necessarily require the process to be actively observable. In the case of evolution the predictions include things such as that more advanced organisms will be primarily in the newer layers and less advanced organisms in the lower layers. Or as Rubang pointed out the chromosone thing.

Evolutionary theory does not predict that more advanced organisms are found in newer layers; rather, it is observed that they are (no contesting there from me).  The theory does say that they are desendents of the lower life forms, and that is the untestability.

Repeatability. The same evidence of evolution has often been found in different places at different times. That is repetition.

This is the same case of the first two; there are things we find in the fossil record, yes, but those things are not the evolutionary theory.  Rather, the explanation of them is.

Falsifiability. Evolution is very falsifiable. Proving that genes do not mutate would demolish the theory of evolution.

I disagree; random mutation is a proposed mechanism of evolutionary theory.  Disproving it would not disprove the explanation of the fossil record.

Also science cannot come to a supernatural conclusion ever. If there are two contradicting sets of well observed observations (which interestingly enough sounds a lot like quantum and classical physics) scientists will work out to find out how and why they contradict. God is not falsifiable and as such can not ever be considered science.

By continuing to work out why they contradict, I suppose that would imply that it is believed one set of of observations must be false, which would contradict the basic principles of science (which is, it is assumed that, given enough observation, repetition etc, the theory is true).  As for God not being falsifiable; while that may be true, it is irrelevant as science must conclude it anyways.

Okay hold up

Hold up

Let's say I take a breed of dog and from these dogs over hundreds of accelerated generations I breed them into animals the size of deer, which bear no outward resemblance to their ancestors and require very different kinds of nutrition in order to function.

Is that evolution?

Likely not (although the different nutritional needs may imply something more than just little outward resemblence).

I do think that, given your scenario, you can get an animal to have considerably different looking descendents.  What I do not think is that random mutation of a gene can produce every other genetic blueprint (bird-to-fish, for example).  In other words (perhaps less accurate ones), I believe there is a "limit" to what random mutation can do.

Likely not (although the different nutritional needs may imply something more than just little outward resemblence).

I do think that, given your scenario, you can get an animal to have considerably different looking descendents.  What I do not think is that random mutation of a gene can produce every other genetic blueprint (bird-to-fish, for example).  In other words (perhaps less accurate ones), I believe there is a "limit" to what random mutation can do.

I don't reject evolution because I don't think it science, I reject it because I think it is wrong (much like my own young-earth creationism, except opposite, of course).  I also reject your description of it as the summation of observed phenomena, per some reasons I think I gave above.

Why is there a limit?