Carr et al describe numerical simulations of three major Greenland glaciers from the present day to 2100. For each glacier an ensemble of simulations is constructed by varying (a) the climate (via the surface mass balance) according to published future projections, and (b) the physics approximation of sliding at the ice bed. All simulations begin with ice thickness and velocity in line with present day observations. They conclude that their variation in sliding physics has a far lower impact on future sea level rise than their variation in SMB. I would argue that this is model sensitivity rather than uncertainty, without observational calibration of the evolution, but the same language of uncertainty crops up elsewhere

The experiments make sense as modelling studies, and the conclusions are justified (although differing from one paper, which is discussed) but an important sliding law has been omitted. This is the regularized Coulomb law (e.g Joughin 2019), which is related to the Schoof and Tsai rules that are discussed, but produces Coulomb-like sliding  over a wider region of a typical glacier. It might make little difference, but should be considered. It is particularly important because it agrees most with time-dependent satellite observations in some regions  (which is also worth noting in the introduction)

I understood the paper in general although there are a small number of typos/grammar errors.

Specific comments

Fig 3 is a good figure, but the labels are small.

Figures 5,6,7,8 repeat the same data shown in Fig 4. It is not obvious to me that they serve a purpose. At the same time, other useful figures, e.g plots of the model initial state vs observations are not included.

Please do not name a ‘Cornford’ sliding law (line 90, 130 and elsewhere). It appears in earlier work (which is cited in both the Cornford  and Asay-Davies papers mentioned), and Cornford certainly makes no claim to it . It will also be useful to say something about the physical meaning of each law when they are introduced

Eq 2: use the same conventions as eq 1 and eq 3 (\tau ^B_b to match \tau ^W -b)

Eq 7 : looks incorrect, as though the second term inside the integral has been copy-pasted from the first and then not edited in some way.

L202: ‘fully converged’ : really? That is unusual, if not impossible. Was there a stopping criterion?

References

Joughin I, Smith BE, Schoof CG. Regularized Coulomb Friction Laws for Ice Sheet Sliding: Application to Pine Island Glacier, Antarctica. Geophys Res Lett. 2019 May 16;46(9):4764-4771. doi: 10.1029/2019GL082526. Epub 2019 May 13. PMID: 31244498; PMCID: PMC6582595.

Please see comments in the attachment.