There's nothing that about trains that directly addresses the "physics first principles" of getting a certain number of folks from point A to point B. That's a problem associated with the number of units to be moved, time, and distance.
Trains are one way to solve the equation. Associated with that are a number of secondary considerations: queuing issues, congestions at hubs, packaging, routing, etc...
But from a first principals standpoint, if the total # of units can arrive from point A to point B in less time using alternative methods then we aren't violating any physics. This will require applying different sets of solutions to those similar sets of secondary considerations.
Think about it in terms of something other than the traditional transportation "box". This is a lot like the move from circuit-switched communications infrastructure to packet switched.
Back in the day, communications lines were allocated on a circuit basis (i.e. the operator plugged your line in to the destination line, and that circuit remained "nailed up" for the duration of the conversation). This required large central offices, large numbers of dedicated trunks between dense communications centers, a specific set of batching and queuing methodology to avoid busy signals, etc...
The solutions to expansion and growth in this era was to build larger central offices, install more trunk lines, build more efficient switches, automate operations, etc... but fundamentally the solution remained the same: nail up a line between two parties. That underlying premise necessarily bounded what could be done with the system.
The move to packet-switched networks changed everything. By necessity of physics, a packet that was allowed to only "hold the line" to a destination for a fraction of a second was going to be able to transfer less information than an established circuit. There was still the need for centralized offices, central lines, etc... but what could be done with that infrastructure was a paradigm-change.
Because these individual packets could be smaller they could ultimately be made faster. Multiple packets could be multiplexed on the same lines. Queuing and switching became more flexible. Routing possibilities expanded dramatically. It became possible to upgrade existing lines to handle more traffic via control mechanisms not possible in circuit environments. Trunking became dynamic. Distributed paths became possible. Redundancy and congestion control became significantly more flexible. Advancements in information systems had dramatic impacts in system management.
Do all of the necessary components for a fully distributed "packetized" human transportation system exist today? No. But things like cheaper alternatives to high-speed rail such as hyperloop, advancements in underground boring, BFR-style trips, automated driving/piloting systems and the like, could very well be the initial sparks in developing that sort of infrastructure. This is the same way that the first DARPA IMP packet routers and the initial DARPANET network paved the way for the internet
is the analogy perfect? No. But my point is: never say never. It may not seem possible with what you know today, but try not to let it hamper your view of what may be possible tomorrow. Otherwise we should be talking about who makes the finer buggy whip.
