An abridged, reader-friendly version of this paper (and published here) is now available. Here's the article and opening section:
The past is the key to the future. Contrary to popular belief, climate models are not the principal basis for assessing human-made climate effects. Our most precise knowledge comes from Earth's paleoclimate, its ancient climate, and how it responded to past changes of climate forcings, including atmospheric composition. Our second essential source of information is provided by global observations today, especially satellite observations, which reveal how the climate system is responding to rapid human-made changes of atmospheric composition, especially atmospheric carbon dioxide (CO2). Models help us interpret past and present climate changes, and, in so far as they succeed in simulating past changes, they provide a tool to help evaluate the impacts of alternative policies that affect climate Paleoclimate data yield our best assessment of climate sensitivity, which is the eventual global temperature change in response to a specified climate forcing. A climate forcing is an imposed change of Earth's energy balance, as may be caused, for example, by a change of the sun's brightness or a human-made change of atmospheric CO2. For convenience scientists often consider a standard forcing, doubled atmospheric CO2, because that is a level of forcing that humans will impose this century if fossil fuel use continues unabated. We show from paleoclimate data that the eventual global warming due to doubled CO2 will be about 3°C (5.4°F) when only so-called fast feedbacks have responded to the forcing. Fast feedbacks are changes of quantities such as atmospheric water vapor and clouds, which change as climate changes, thus amplifying or diminishing climate change. Fast feedbacks come into play as global temperature changes, so their full effect is delayed several centuries by the thermal inertia of the ocean, which slows full climate response. However, about half of the fast-feedback climate response is expected to occur within a few decades. Climate response time is one of the important 'details' that climate models help to elucidate.
H/T: Andy Revkin's indispensable Twitter feed.
