One of the problems I find with "Testing to Destruction" is that unless you are only firing each weapon once, it doesn't paint a very good picture. Once you get into elevated material stress levels, each subsequent test is weakening it through crack nucleation.
I'd encourage anyone who wants to do something like this to Wikipedia "fatigue (material)". The subject matter basically deals with cylcling something, usually below yield limit for plastic deformation, until it breaks. Some ferrous metals and titanium have an endurance limit. Under this threshold, you can in theory cycle something forever and it will not break. All other metals including these metals above that limit slowly degrade as tiny cracks in the material form until they ultimately fail. Corrosion also plays a big factor.
Where firearms fall in the limit, I am not sure (above or below the endurance limit), but I have a feeling it is above it. The important thing to note is that the relationship is typically represented on a logarithmic scale. If you look up the Wikipedia article, you'll see a graph for brittle aluminum with the stress levels on the y-axis and the number of cycles on the x-axis. Look at 150 and 300 MPa for example. Say we have a fictious rifle made of this material, and the stress in a rifle were 150 MPa with normal loads, it will survive roughly 10,000 rounds before failure (1 E+4). Now say you double the pressure which should roughly double the stress in the rifle. If we fire 300MPa rounds, the rifle will last about 10 shots (1E+1). Notice doubling the pressure did not half the life of the rifle. Your rifle's life will now be 0.1% of what it would have been. This is just an example, and the actual stress levels would make a difference, if it were more like 100MPa and 200Mpa, you'd have about 1,000,000 and 1,000 respectively. That's still 0.1% of its normal life, but 1,000 isn't so bad. My gut feeling is that an actual firearm's life to critical failure would probably be on the order of E+5 (100,000s) or high E+4s, but that's just a guess.