Kummer & Dessler seem to have performed their basic calculations improperly.

They claim that the AR5 forcing time series are referenced to the late 19th century, and therefore deduct from the temperature time series the 1880-1900 average temperature. But the AR5 forcing time series are, as stated in Table 8.6 (which they cite), referenced to 1750. They should therefore have likewise deducted the 1880-1900 average forcing from the forcing time series.

Upon making this correction, I calculate that Kummer & Dessler's basic estimate of climate sensitivity should be (taking the median) 1.5 C, not 2.3 C.

I dispute their/Shindell's contention that aerosols and ozone have an efficacy substantially above one. But even taking their high case of an efficacy for them of 1.5 would only increase the climate sensitivity median estimate to 1.75 C, not to 3.5 C as they calculate.

This appears to be an unscientific attempt to undermine the robust energy budget evidence for climate sensitivity being low. It is appalling that the paper has been accepted for publication by GRL.

Nic's remarks were picked up in a conversation at Troy Masters' blog with Troy observing of the problem:

I fear this would have a significant impact, since the large volcanic activity from 1880-1899 (averaged ~ -0.6 W/m^2 according to GISS) means the forcing differential from 1880-1899 would actually be much larer than the forcing difference since 1750, despite the increase in anthropogenic forcings from 1750 to 1880. This would obviously reduce the ECS estimate substantially (to 1.5 C in the E=1.0 case according to his calculations, though I have not done these myself).

The point was also picked up in the current discussion of climate sensitivity at Climate Dialogue, where Dessler agreed that the baseline was incorrect, but noting rather astonishingly that the journal had agreed to let him change the proofs of the paper:

I appreciate your comments. Your statement about the referencing period of the forcing is correct and that will be corrected in the galleys.

But the rest of the comments werr even more surprising. As noted in the various discussions, the obvious way to make the baseline for temperature and forcing consistent is to use an 1880-1900 baseline for both. This would make the climate sensitivity lower. But Dessler had other ideas:

Assuming that the climate in the late 19th century is warmer than that in the mid 18th century (probable since radiative forcing is +0.3 W/m2 in the late 19th century), then referencing both time series to 1750 will increase the calculated climate sensitivity (I can explain why if it’s not clear). Thus, it does not affect our conclusion that incorporating efficacy has a significant effect on the inferred climate sensitivity.

So in the new calculations, temperature and forcings are calculated against a consistent baseline of 1750, 100 years before the start of the instrumental records, presumably by means of using proxy reconstructions. And of course because of the decision to allow the changes to be made to the proofs, this choice is not subject to peer review.

Some ensuing discussion, including Dessler's defence of his new position, can be seen here (keep reading).