There seems to be a bit of confusion over likely future warming; lets try and clear it up a bit. Last week I published a piece suggesting current climate policies put us on track for ~3C warming by 2100, with a wide uncertainty range (1.9 to 4.4C): thebreakthrough.org/issues/energy/… 1/8
Uncertainty in future warming mainly comes from two things: uncertainty in future emissions and uncertainty in the sensitivity of the climate to increased CO2 and other greenhouse gases. 2/8
We tried to control for the former, extending IEA scenarios through 2100 to estimate warming implied by current policies. Even here we looked at two different cases (current policies and stated policies) and two different post-2040 assumptions (constant and linearly increasing) 3 
Using the CMIP5 multimodel mean (which has a climate sensitivity of just above 3C), this would imply warming of 2.9C to 3.4C for current policies and 2.7C to 3C for stated policies (e.g. including Paris commitments not yet reflected in policies). 4/8
However, climate sensitivity is also quite uncertain; it means that our best guess 3C warming by 2100 could be off in either direction by around 1C. So combining emission and sensitivity uncertainties gives a range of 1.9C to 4.4C. 5/8
While its possible that current policies could lead to >4C warming if climate sensitivity is on the high end of model estimates, that same would would have close to 6C warming under RCP8.5. There still is a large difference between 2100 current policy and RCP8.5 outcomes: 6/8
That said, its important to emphasize that current policies continuing is neither the best or worse possible outcome. The IEA could end up being conservative on clean energy deployment rates. Alternatively, a rise of populism and isolationism could spark a return to coal. 7/8
Future emissions are inherently difficult to predict, so we tend to look at a wide range of possible outcomes. The worst of these outcomes – where emissions increase dramatically driven by a 500% increase in coal use – appears much less likely today than a decade ago. 8/8
