This is slightly misleading. What we've understood for a century is the thermal forcing due to CO2, which is a logarithmic function of the CO2 concentration (i

This is slightly misleading. What we've understood for a century is the thermal forcing due to CO2, which is a logarithmic function of the CO2 concentration (ie, forcing == O(log n) in CO2 concentration). What we don't understand (unless we decide GCMs are predictive and can be validated) is the impact of a given thermal forcing on planetary temperature. This is the "climate sensitivity." Unfortunately, this is the number we actually care about. If climate sensitivity is low (negative feedback), Earth is thermally buffered and we have to worry less. If climate sensitivity is high (positive feedback), the opposite. (Note also that Arrhenius, the discoverer of this effect, and pretty much everyone through the '60s, assumed that global warming was a positive side effect of burning fossil fuel, just from the preindustrial assumption that warm winters and high crop yields are good for humanity.)