The problem with buses is that they're typically slow compared to even subway/metro rail (let alone inter-city). That doesn't mean they're not the right fit in a wide variety of scenarios (for example, they're about perfect for small/medium cities that can't afford to invest a whole lot in intra-city rail, and they're the only viable option for smaller cities/towns that aren't already stops on a larger inter-city rail line), but they're not necessarily the right fit for a large city with more ground to cover (even with dedicated bus lanes, buses can only - safely - drive so fast, and that speed is typically less than rail). The phrase "bus rapid transit" is kinda nonsensical/oxymoronic; there's not really much that's "rapid" about buses.
Ridesharing is slightly more promising, since the smaller vehicles can drive a bit faster and don't have to stick to specific routes. Unfortunately, because they're smaller, they don't carry as many people, so they're less suited for heavy load (i.e. lots of passengers trying to go from the same place to the same place).
Rail is probably the most efficient, but also the least flexible and the most expensive (the expense can be mitigated somewhat if you're proactive about building tram rails into your roads, but not all roads warrant that, and by the time they do it's likely already too late; plus, that doesn't do much for older cities).
The ideal would be something like the following:
- High-speed rail between cities
- Metro rail between major parts of cities (and/or between cities and suburbs/satellites)
- Full buses on fixed (or at least firm) routes to transit centers in neighborhoods/suburbs/satellites that don't have metro connections of their own (connecting to nearby metro stations, the high-speed rail hub, and/or other bus stations)
- Minibuses on more flexible routes for intra-(neighborhood/suburb/satellite) transit (between neighborhood minibus stops, to nearby macrobus/metro/rail stations, or to specific minibus stops outside that particular "zone" in rare cases where a point-to-point really is faster and enough people are trying to do it)
All this would be controlled by a rider walking up to some sort of kiosk (or using a smartphone app), picking a destination, and getting a route calculated on-the-fly (for kiosk users: probably printed, or if we wanna be real slick, dispensed on an e-ink display that'll update if the route changes).