I enjoyed reading the revised paper, which is much improved in terms of the GMD concept of what a model experiment paper should be. The requirement is not that the experiment must be perfect, but that the decisions made should be fully explained in terms of both scientific justification and practical considerations.

You responded to every comment apart from one, which was actually the first made by Gavin Schmidt. Therefore, please provide a response to:
“DECK runs: These are a suitable ’entry card’ into the process, and the requirements for new DECK entries for variations in physics, resolution, interactivity etc. is necessary (one run that is missing is perhaps a slab ocean equilibrium 2xCO2 run for coherence with previous estimates of the ECS). However, there are some implications of the DECK/Historical approach that need to be addressed. Specifically, because this is a relatively low barrier to entry, more models and model versions will very likely be archived. Thus instead of 60 individual model configurations as were available in CMIP3, there will likely be far more DECK entries over the lifetime of the CMIP6 program. I think this will be a good thing scientifically, but people should be ready for this.”

Historical Simulation:
For simulations that are relevant for model evaluation against observations it is important that the simulations be as realistic as possible. The wider community, for example those using the CMIP ensemble as part of climate impact assessments, surely have the expectation that the historical simulations will be the best current estimates of the climate history, and that the spread in the ensemble can be considered a reasonable estimate of our uncertainty. Otherwise CMIP fails to address its first question, “How does the Earth system respond to forcing?” and instead is considering the less interesting question of, “How do different models of the Earth system respond to certain forcings?” Thus, I find your dismissal of Gavin Schmidt’s suggestion that some models may be more advanced than others and thus able to include new or different forcings to be inadequate. I’m not convinced that you have stepped out of the MIP box sufficiently to consider the wider perception of these runs. Each model should surely be able to submit their best historical run as their prime entry on ESGF. Isn’t “conflated uncertainty” exactly what the wider community requires for this particular ensemble?

Documentation:
That brings me to the topic of documentation. This is alluded to a number of times in the manuscript, but nowhere (apologies if I have missed it) do I see a description of how the documentation is to be done. This must be addressed. Perhaps it is to be addressed further in one of the other papers in the special issue, but the basic plan needs to be outlined here and that other paper referred to. Speaking personally, as a user of CMIP5, the level of documentation was inadequate, and I often email people to find out details about their runs, a pursuit which is not always successful, as the answer in some cases is already lost in the mists of time. Is there to be a documentation file of a certain format permanently attached to each data file? If this topic is not to be addressed elsewhere, can you include an example documentation file as supplementary information?

RFMIP-lite:
All three reviewers are deeply concerned about the radiative forcing and my impression is that they would prefer to see the RFMIP-lite fixed SST experiment included in the DECK. Personally, I think your argument is reasonable (I interpret it as saying that the DECK runs are useful even without RFMIP-lite), but I do not see why these three reviewers would all be unfairly biased the same way, and so I am concerned that your decision to exclude the experiment is actually strongly against the consensus of the community, and that this decision should be carefully considered once more before you finally commit to excluding it.

The 1%CO2 run and “effective” Climate Sensitivity:
Given your answers to Gavin Schmidt’s questions about these topics, it is apparent that CMIP6 has made the decision to abandon the estimation of equilibrium climate sensitivity. Gavin cites his own paper on GISS, but many have shown that the “Gregory method” based on a short 4xCO2 run is not a very good approximation to equilibrium climate sensitivity (see Feedbacks, climate sensitivity and the limits of linear models, Reto Knutti, Maria A. A. Rugenstein, DOI: 10.1098/rsta.2015.0146 and references therein). This combined with the lack of a stabilisation run after 140 years of the 1% CO2 leaves no means of calculating equilibrium climate sensitivity. This needs to be stated clearly in the paper as a positive decision, as otherwise it may later appear to be an accidental omission. Just a thought, but would replacing the abrupt 4xCO2 with an abrupt 2xCO2 be useful - the 150 years is short, but it would at least be on the way to converging towards the desired value?

Abstract:
There is a comment here that surprised me, and I didn’t find further explanation in the main body of the manuscript, “The participation in the CMIP6-Endorsed MIPs will be at the discretion of the modelling groups, and will depend on their scientific interests and priorities.” I wasn’t aware that the MIPs are policed. Why and how would a group be excluded from joining?

Section 3.2 last sentence.
“..to be evaluated fairly against those simpler models without representation of carbon cycle processes.”
I am not sure it is reasonable to suggest simplicity is relevant to any of these models, and models that do not have carbon cycle processes may be more complex in other ways. Could just delete the word “simpler”?

Section 4.2
“A number of MIPs are developments and/or continuation of long standing science themes within CMIP.” I do not think this is accurate. At least some of these MIPs were founded pretty much separately from CMIP, and were only included in CMIP for CMIP5. Maybe just delete the “within CMIP”. Otherwise it reads as a land grab that fails to give credit to the industry and ingenuity of those who founded the other MIPs.

The Special Issue:
In many places the content of other manuscripts in the special issue is alluded to. GMD has many editors, and their focus is on assessing whether the papers pass the GMD peer review criteria, rather than making sure they include everything that you have promised here. It is therefore important that the authors of this paper make comments in the online review of the other papers in the special issue. Not hard, I think - as you only need check 3 or 4 each, and you’ll be done!

Summary:
I’d thought the paper had ended rather well at “decade or more”. Then I turned the page and found another paragraph which didn’t seem to add anything more. Maybe the final paragraph isn’t required, or perhaps it could be made more compelling?

Data availability:
Did you check whether actual “DOI”s are to be used in the case of ESGF? If you are not using official DOIs, then a different term for the ESGF digital identifiers should be coined.

Finally:::::
“DECK and CMIP6 historical simulation”
I don’t understand this.
Let’s start with DECK - Diagnostic, Evaluation and Characterisation of Klima. What is Klima? I looked it up on Wikipedia and found that it means “climate” in Czech, Danish, Norwegian (2 dialects), Serbo-Croatian, and Slovene, but that it means “air-conditioning” in Polish and Turkish. You need to provide the translation in order to avoid confusion as it is unfair to expect people, especially those who speak English as a second language, to understand other relatively obscure European languages. There are, incidentally, many stupider acronyms than DECC. So… I think I’ve now understood DECK; but I’m still confused. Why is the historical run outside the DECK? I can’t see where any scientific or practical distinction is made between the DECK and the historical run. They are both equally required as entry cards. Looking at the definition of DECK again - Diagnostic, Evaluation and Characterisation of Climate, it seems to me that the historical runs also belong inside the DECK. Please explain. Are modelling groups actually secretly permitted to exclude the historical run?

Thanks to the authors for all their additional work on the manuscript. It is now appropriate to accept the paper for GMD. The paper fulfils the criteria for an experiment description in GMD, which is that the experiment is fully explained and justified in practical and scientific terms.

From the perspective of documentation of model development, it is a real advance that modelling groups will be required to provide documentation for their runs, at the level such that differences between model versions can be identified.

Note that the grammar is not perfect in the manuscript. Hopefully the copy-editing will improve this aspect.

The peer review process for this particular paper highlighted a potential problem for experiment description papers at GMD, although hopefully this is an extreme example. Although the descriptions of the experiments are greatly improved through the peer review process, the reviewers also made a small number of suggestions of changes to the experimental protocols, but none of these were implemented. Such suggested changes might be considered the most fundamental of the reviewers' concerns for a model experiment description manuscript, so it is a little difficult to accept the paper when none of these concerns have been acted upon. In this case I have decided to accept the paper anyway, as I understand that the authors do not feel to have the authority to change the protocols at this stage as the protocol is already agreed by the community. I will merely note here that, because of the open review process, a GMDD peer review is, in my opinion, itself a community effort - the discussion manuscript has been viewed or downloaded at GMDD, 2163 times to date, with quite a wide range of comments made during the open interactive review.

Finally, in the previous iteration I questioned the classification and nomenclature surrounding the "entry card" runs. I can now see the authors' point of view, and have gained perspective on my own view. Hopefully I'm in a minority of one here, but I still think that it is unnecessarily complicated and confusing. In English and in this context, DECK is quite a clever acronym, as it brings to mind a deck of cards, in this case, presumably, the "entry cards" for the federation of CMIP6 approved MIPs. That one of the entry cards is not included in the DECK of entry cards is confusing. It is stated in the manuscript that the contents of the DECK can change in future so I do not see why the DECK is not simply comprised of the entry cards. Don't say you weren't warned - I have already seen people using "DECK" as an informal short-hand for the entry card runs!