Some simulations produce unrealistic mountain heights (> 10 km). The reference simulation uses a combination of erodibility, tectonic uplift, and rainfall parameters, which produces a reasonable mountain height. But when these parameters are modified, for example when the rain is divided by 3, the mountain can reach 13 km, which is obviously unrealistic on Earth. We can predict the total height that a mountain can reach in these simulations: this is the height of the mountain in case the rain is too little (or the uplift too fast) for no river to form. In this case erosion is only controlled by landslides that impose a slope limit of 30 degrees. Such a slope and a half mountain width of 20 km predicts a pyramid-shaped mountain with two sides 16 km high, which again is totally unrealistic. However, these simulations are useful on two levels. They provide a good visualization of how variables such as drainage density vary. These experiments are also useful because they should make it possible to question either the parameterization of the model (erodibility too low for example) or the model itself. For example, the model does not take into account glacial erosion, which can limit relief growth. Experts can read Brozovic et al. (1997). The elevation of mountain depends also on the rheology of the lithosphere, that has a limited capacity to support high mountains and collapses or flows when these mountains become too high (e.g. Tibet). Even worse, mountain heights may have nothing to do with climate or glacial erosion according to Dielforder et al. (2020). In any case, an unrealistic simulation result raises questions about what has not been taken into account in the model. That’s what models are for. In summary, we can say that the simulations predicting too high heights are wrong, that we can reject the set of parameters used or even the model which is too simplistic, but these simulations remain useful.