Regular US railroads generally are 1:100 slopes (0.6°) or less. Grades steeper than 2.2:100 (1.2°) are rare. Clearly there is a way to efficiently connect popul

Regular US railroads generally are 1:100 slopes (0.6°) or less. Grades steeper than 2.2:100 (1.2°) are rare. Clearly there is a way to efficiently connect population centers which is not steep. I think the issue is vertical curvature rather than slope. Traversing a slope at speed is just going to get you a constant vertical component which you will not feel, and not terribly fast, <20kph, about half as fast as a fast elevator. Changing slope is going to get you a vertical acceleration which could fill your tiny capsule with more vomit than one would like. The question is then, can you blend the slopes to keep that under control? Going maximum up grade to maximum down grade at full speed will be something like a 40kph velocity change, at 1G that takes 1.1 seconds, during which time you will drive 360 meters. So, blending over 360 meters gives you full weightless vomit motion or a 2x weight. The blend from full up to full down will probably need to be over a kilometer to keep things clean¹. ␄ ¹ The internet is not being forthcoming about real world elevator accelerations. It is too polluted by imaginary physics problems. So I unilaterally decided 1/3g would be ok.