Let me take a stab at this:
> thoroughly insufficient as proof of any long-term trend we can expect to continue in a non-linear system as complex as the global climate
Sometimes, non-linearity doesn't really matter. Human body is complex system, we don't fully understand it, but you wouldn't stand in front of a car that runs 100kph. Because then we just know what is going to happen, from perfectly linear Newton's law.
In the 1950s, Plass did some equilibrium calculations of climate sensitivity to CO2. I think he came with a value around 2 degrees. Modern consensus is 3. This all happened despite decades of research, involving all sorts of additional non-linearities. What does that tell you? He was right in the ballpark, and he actually had a decent model!
Just as we cannot say what injuries you end up with in the first example, we cannot easily say what will happen to the Earth. But we are very certain it will heat up (and roughly how much). Unless of course humans on the Earth react non-linearly and decide not to emit more CO2, just like person in front of a car can somewhat non-linearly jump to the side.