A train is sort of like Shaquille O’Neal on ice skates. It can be extremely efficient, but it is not good at starting, stopping, or making sharp turns. And you really do not want it to crash into anything.
If Shaq were to be given those ice skates, you would want him to skate in one of two types of ways. One type would be to give Shaq an empty, straight rink: he could then skate quickly without having to stop or turn. The other would be to have Shaq skate extremely slowly: that way he could share a rink with others without risking a fatal crash, and it would also be much easier for him to start, stop, or turn without too spending too much of his energy or risking a crash.
In our transit system, we tend to value speed, so we opt for the former method over the latter. We spend lots of money to build subways, so that our trains can get their own empty rinks to skate on, with stops spread far apart (relative to busses/streetcars) and enough room for wide turns.
However while speed is likely to remain a top priority, it is also likely to become less important than it has been until now, relative to other concerns like cost, comfort, capacity, and cleanness. The need for speed reflected the fact that people were in a rush, because people had jobs and kids. Going forward, though, technological and demographic changes seem to indicate this will not be the case so much. Demographically, barring a dramatic change in immigration, Canada will very soon have many more pensioners and fewer children than it has today. (Look at our population pyramid and you will see this is the case). Technologically, meanwhile, it is becoming increasingly easy to get some work done while sitting on a train; plus, of course, there is the possibility that automation and/or outsourcing will cause unemployment and/or leisure time to rise.
In theory, the simplest way to increase the capacity of a train is to make it easy for people to sit on the train’s roof. This would also dovetail well with the idea of having the train travel extremely slowly, in order to maximize efficiency and safety in a busy, intersection-filled urban area without the need to construct a subway. The idea would be for the train to serve as a sort of portable café for people to work or relax in while they travel, complete with a rooftop patio. The seats on the roof could, perhaps, have optional retractable covers, to keep out wind, noise, and bad weather.
But this still leaves you with the challenge of how to make turns, if the road is not wide enough to allow a long train to do so. This could be addressed either by using light rail instead of heavy rail, or by having the train descend into a tunnel to make its turn. (Although in that case, everybody would have to get off the roof for the tunnel portion). Or by having rail lines with no sharp turns (for e.g. the Eglinton Crosstown, or a train using the extremely long Queen Street streetcar line).
Also in theory, the cheapest way to build a subway is to build a train with as narrow a diameter as posible. This would also dovetail well with the sitting-on-the-roof idea, as it is easier to get on and off a low roof than a high roof. (Though again, passengers would have to get off the roof before the train enters the subway tunnel). The problem here, though, is how to get passengers to and from their seats within a train with a narrow diameter and low roof.
The solution to this problem could (again, in theory) be autonomously-moving chairs, which move passengers in and out of the train. Such chairs would also dovetail well will with a train system that mixes an above-ground, extremely slow section with a typically fast subway section. The autonomous chairs could bunch together to fit more passengers in during the fast subway portion, but then spread apart to give people much more room to work or relax during the extremely slow above-ground section. Space could be freed up by perhaps having some chairs autonomous move onto the rooftop ‘patio’ during this section, and lock in to grooves within the rooftop.
For Toronto, given the high cost of tunneling downtown (because of downtown’s closer-to-surface bedrock, utility barriers, underground building barriers, and busy street traffic to disrupt during station construction), the extremely slow above ground portion of the train system could run across downtown Toronto, using existing streetcar tracks on Roncessvalles and on King (the turn from Roncessvalles to Kind is a very wide one, so this could be possible without needing a subway). It could then enter into a subway tunnel by King and Broadview, and head north as a typically fast subway (albeit with a narrow diameter) all the way to Sheppard and Don Mills. It would thereby run the entire “Relief Line” route, just in a much crazier (and just maybe, in a much more efficient and cheap) way than those who want to build a conventional subway here imagine.
Alternatively, you could have a stacked version of this system: in one direction it would be a subway at high speeds, in the other direction it would be a surface train at extremely slow speeds (but with trains bunched much more closely together), and with people sitting on the roof. An advantage of this system could be that you could use existing surface streetcar tracks to have two surface lanes going in the same direction, so that you are less likely to have a delay that would cause the subway tunnel lane to go under-utilized. Plus, by reversing direction midday, you could have the subway section run downtown in the morning and uptown in the evening.
Alternatively, you could have no subway section, and run the entire line as an extremely slow, cafe train. If you were to do this, you might also have the option of building trains (or streetcars, if you want to do sharp turns without a tunnel) with twice the width of normal trains (they could run west on Queen and east on King, for example). This would give you a much more spacious interior for a slow, comfortable ride, and a rooftop patio that is much wider too. A wider rooftop patio would be especially useful if you were to use trains of normal height rather than low-roof trains. If people are going to be sitting on roofs that are high up, they may want a wider roof so that they don’t have to be so close to the edge. Having a system of this kind, with no subway, also means that you don’t have to worry so much that the tunnel will run underutilized: this means you could have the extremely slow train line share its road lanes with cyclists or with cars.
It also means that, instead of rooftop seating, you could simply have double-decker trains. If there are height barriers preventing these — as in most urban settings there would be — you could use low-roof double-decker trains and autonomous chairs to get passengers in and out.
Better yet, we could just do this: https://nerdist.com/the-first-cat-cafe-train-is-carloads-of-cuteness/