Suppose we have ten sick people. Dr. Quack persuades them to take his Patent Medicine. Five of them die and five of them recover. Dr. Quack says, "My medicine saved five lives!" But Dr. Nay says, "Nonsense, your medicine killed five people." Both claims are 100% consistent with the facts, and yet they are contradictory. Consistency with the facts is a necessary quality for a scientific theory, but it is not sufficient. If we allow consistency with observation to be sufficient proof of a theory, we are practicing pseudoscience. The Food and Drug Administration (FDA) rejects both Dr. Quack and Dr. Nay's claims, saying the Patent Medicine has absolutely no effect upon the recovery of the patients until experiment has proved otherwise.
The FDA has adopted the null hypothesis. The null hypothesis is the foundation of scientific reasoning. We determine the null hypothesis with Occam's Razor, by which we cut off any unnecessary parts of our hypothesis until we arrive at the simplest possible explanation of our observations. And the simplest possible theory about the relationship between one thing and another is that there is no relationship at all. The only way to disprove the null hypothesis, according to scientific method, is with observations. Compelling arguments and sensible speculation are insufficient, nor is consensus among scientists, nor the authority of experts.
When it comes to the climate, our initial null hypothesis is that the climate does not vary at all from one year to the next, nor does its carbon dioxide concentration. We will have to disprove this with observations before we can begin to discuss how humans might cause climate change. So let us look at what observations are available to disprove this hypothesis. We observe valleys carved by glaciers in warm climates. We find fossils of tropical plants in cold places. We pull out ice cores and they suggest the Earth's temperature has varied dramatically in the past few hundred thousand years, along with its atmospheric carbon dioxide concentration (6-16°C and 200-300 ppm when measured with a 1000-year running average). These observations and many others disprove our null hypothesis. The climate does vary. It varies dramatically and naturally, whithout any human influence. This is our new null hypothesis. We call it the theory of natural variation.
We can use the infra-red absorption spectrum of carbon dioxide to argue that doubling the carbon dioxide in the atmosphere will warm the planet (by roughly 1.5°C if we ignore clouds, and 0.9°C if we account for clouds, according to our own simulations). Before we can hope to show that human carbon dioxide is affecting the climate, however, we must show that human carbon dioxide emissions affect the atmospheric carbon dioxide concentration, and on this subject our theory of natural variation states that human emissions have no effect. So we must disprove the theory of natural variation. Natural emission and absorption of carbon dioxide were already in equilibrium before man started burning fossil fuels. Our annual emissions are only 4% of the natural emissions (8 Pg/yr from burning fossil fules compared to 200 Pg/yr of natural emission). In the simplest chemical equilibriums, absorption is proportional to concentration, so our 4% increase in emission will, to the first approximation, cause atmospheric CO2 to increase by 4%, or 10 ppm. Such an increase is so slight that it's not clear how we could distinguish it from the larger natural variations.
Even if we could prove that the recent increase from 330-400 ppm atmospheric carbon dioxide was due to our burning of fossil fuels, we would still have to prove that increasing carbon dioxide concentration causes the world to warm up. We may have a compelling reason to suspect that is has this effect, but we cannot abandon the theory of natural variation until it has been disproved, and we need observations that contradict the theory to disprove it. We cannot disprove the theory with compelling arguments alone. Even if we accept that increasing carbon dioxide traps heat somewhere in the atmosphere, this does not neccessarily mean that the climate will, as a whole, warm up. It may seem obvious that lighting a fire in my fireplace will warm my house, but the reality is that my children's bedrooms get cold when I light a fire. That's because the thermostat is near the fireplace, so when I light the fire the radiators turn off. The climate is a complex system. It could be that it contains similar surprises, where we observe the opposite effect to the one we expected. In fact, looking at the ice cores, it appears that atmospheric carbon dioxide increases occurr one thousand years after increases in temperature, which could mean that carbon dioxide somehow stops the world from getting any warmer at the end of an ice age, by a process that we don't understand.
So far as we can tell, scientific method, when applied to our observations of the Earth's climate, arrives at the assumption that the climate varies dramatically and naturally, and that human carbon dioxide emissions have no effect upon it. We look forward to seeing this assumption disproved by observations of nature, but so far we have been disappointed.