This paper is an example of some very clever people creating a potentially useful tool and then proceeding to enthusiastically apply it in completely the wrong way.

It starts by criticising the performance of GCMs and then proposes what it describes as an irreducibly simple climate model. It walks through the authors’ preferred values for each model parameter, most of which are within the limits proposed in AR4 and/or AR5. The model output is then compared to various IPCC projections dating back to 1988. The paper finishes by concluding that the IPCC has over-stated the likely magnitude of anthropogenic global warming.

A fundamental problem I have with the paper is that it has bundled two separate topics together. The first is the model itself, and the second is the selection of parameters used to generate the output.

I am neither an expert mathematician nor a physicist, but at first sight the model derivation makes sense. As a simple tool for estimating possible temperature increases in response to different climate scenarios it could be very useful and time-saving. It should be noted that it is a straightforward linear model, and the output is an increase in temperature over a user-defined timespan. It considers only the response to anthropogenic CO_{2} and ignores all other climatic processes. Although the paper claims the model is irreducibly simple, no evidence is offered that this is true.

So far so good, and if the authors had more or less stopped here (with a bit of basic validation) it would have made for an interesting paper that could perhaps have been published anywhere. The problems come when the authors start selecting numbers to feed into the model.

The authors’ parameterisation pretty much follows IPCC preferred values, except for the value they select for the net sum of all feedbacks (in section 8.3.2). Because Vostok ice cores suggest that for the last 800,000 years mean surface temperatures have varied by only ±3K, the authors propose a range of values that are mainly net negative. This is in sharp difference to the IPCC’s range of values, which are all firmly positive. The problem I have is that even if the authors’ assertion is valid (and I think it probably is), they simply assert it is “plausible”, illustrate it with an analogy drawn from electronic engineering, and move on. But because this decision is so central to the results I would expect it to be justified far more thoroughly, arguably even requiring a separate paper. The assumption that such thermostasis will persist into the future in all possible CO_{2} scenarios is certainly open to question.

The next problem I have is with Figure 6, which compares the model output with various IPCC projections. This purports to show that the model is a closer match to observations than anything the IPCC has published. To be frank, this figure is shockingly bad. The graph shows modelled decadal warming rates carried forward to 2050, compared to “the observed temperature trend of recent decades”. Two different decadal rates seem to have been selected (63 and 17 years), with no justification given for either. And, on what planet do we expect a linear extrapolation of recent temperatures to be necessarily indicative of future temperatures over the next 35 years? In any case, observed temperatures include the net effect of all climatic processes, not just the response to anthropogenic CO_{2}, which is all that the authors have modelled. So how can the two possibly be directly comparable? Finally, even if it is true that the simple model closely matches observations, unless the net total of non-anthropogenic forcings is zero this can only be accidental.

Figure 6 is obviously beset with a number of fundamental problems and is exactly the sort of rubbish that sceptics get upset about when their opponents try it. I suspect it was created as an attempt to include a single attention-grabbing visual that would easily demonstrate the paper’s conclusions to non-scientists. Irrespective of other criticisms of the paper, it should never have been published with Figure 6 included in anything like its current form.

Moving on, the Discussion section of the paper introduces examples where increases in the complexity of some GCMs did not provide any improvement in performance, but this is essentially just two isolated anecdotes. To draw any wider conclusions a full study would be required to use as actual evidence. Again, this would make for another interesting but separate paper (albeit one that it is probably impossible to write). The authors do, however, quite fairly point out that their model is not intended to replace GCMs, nor is it necessarily more accurate.

Further into the Discussion, the authors make the claim that “The simple model confirms the hot running and exposes several of the reasons for it.” It is here I think that we reach the biggest conceptual problem with the paper. The proposed model itself confirms nothing. It is, as noted already, a simple linear mathematical function. It is the parameter values selected by the authors, that they then feed into the model, which create the variance to the GCMs. I’m not sure exactly what the authors intended, but the conclusions of the paper seem to suggest that somehow it offers a mathematical rebuttal of the GCMs, not one based on parameter value selection. I think it would be easy to skim-read the paper and come away with completely the wrong impression. The paper should more accurately say ‘The simple model **can be used to **confirm the hot running…’, but only if you agree with the authors’ selected values. In fact, the paper more or less assumes its own conclusion.

A more minor niggle is that in the Conclusions the paper states that “…the reviewed journals of economics generally report that the cost of mitigation today would be likely to exceed these low costs of adaptation to projected global warming, perhaps by as much as one or two orders of magnitude.” Given how much controversy all estimates of mitigation costs generate, I would have expected to see at least a few references to back up this assertion, but there are none.

My final criticism is that to date the authors have not released the data used to create the paper, for example the number crunching carried out to generate the various tables and figures.

In summary the paper boils down not to a criticism of GCMs but of the IPCC’s estimated value for net feedback. Given this, I believe the paper is misguided in what it tries to achieve and in what it concludes. It would be a constructive next step to propose a coherent, physical basis for the authors’ preferred net feedback value.

There will be a more comprehensive review at Climate Etc. March 1. There is no number crunching code necessary. All the calculations are just arithmetic, summarized in the tables. There is a ‘mistake’ that will be criticized in the comprehensive review. The argument about ice ages is weak, and follows from the ‘mistake’. Note also that the papers feedbacks are radiative forcings, NOT Bode feedbacks f. That is a source of considerable potential confusion which gets sorted out in the comprehensive review. Regards.

Rud, I understand the calculations for the tables and figures are very simple but they should still be released.