1. F=ma, Newton's second law, can be replicated easily. It's a fact.
2. If I kick this soccer ball right here, applying a force, it will accelerate away from me. It's a fact.
3. Look, it has accelerated away from me, it's a measurable fact!
...a short time later...
4. Oh, I was standing at the bottom of a hill, so the ball just rolled back to my feet.
Do you understand? We are discussing an insanely complex system, with feedback loops everywhere, and a scant (at best) 100 years of quality data on a planet 4 billion years old. There is nothing simple about this. The physics of the radiation balance isn't in question, but for all we know it may be dwarfed by negative feedbacks. They are forced to fallback on computer models for a reason.
Nobody talks about 99% of doctors and smoking, because we can empirically test this. Hundreds of millions of people have smoked, and we have observed the effects.
The smoking/cancer analogy always makes me smile, because it is actually a very good one. Let's turn it around. The physics of how smoking causes cancer is pretty solid. We know about toxins and carcinogens, etc. But let's say I hand you a single human being, and I ask you, "if he smokes for the next 40 years, will he get cancer?". We don't have a freaking clue. The human body - the feedback of the immune system defences, nutrition, genetics - are beyond our power to model. Science is hopelessly unable to answer that question. And ponder this for a moment - we have 6 billion potential test subjects to work with! We have one earth. This is the state of climate science. The biosphere/atmosphere/oceans/etc. are every bit as complex as a single human body, and we have only one patient to study, and data for maybe 0.0000025 of the patient's lifetime.
But, I hear you say, "shouldn't we be cautious then? Shouldn't we quit just in case?". And fair enough, we probably should be cautious. But the counterpoint to that is that fossil fuel usage isn't a nicotine hit. It doesn't deliver us a few moments of pleasure in return for the risk of cancer. It fuels our entire industrial civilization. So the risk/reward calculation looks quite a bit different, doesn't it?
In one case, we have the potential risk confirmed by the empirical evidence of a test set of hundreds of million of smokers, and the potential reward being a nicotine hit. In the other, we have no way of testing the theorized risk outside of models, and the reward is the energy to power our entire civilization.
So yeah, it is an analogy that always makes me laugh a little.