[aetherson]

Okay, sure. But that's not anything analyzed by the article at the top of this thread. And further:

1. Presumably, if lateral acceleration proves to be preferable to vertical acceleration, you just turn off the banking.

2. You control the precise amount of vertical acceleration by the size of the bank, so if there's some amount of vertical acceleration that's desired and some that isn't, you only bank hard enough for the desired acceleration. You also to at least some degree control the jerk (change in acceleration) so you can smoothly slide into the acceleration change as you bank slowly harder.

3. Importantly, this is only increased-acceleration in local "down," not decreased acceleration in local down. I don't have any scientific data to back this up, but it's the drops (ie, reduced vertical acceleration compared to gravity) that seem to nauseate people on roller coasters, not the climbs.

But most importantly, I think that people were badly misinterpreting this article (not the original hyperloop paper) in thinking that its charts were about vertical rather than lateral and longitudinal acceleration.