Earlier, I wrote about the attractive potential of starting a ride-share of 'caro publico' between the Salt Lake Central Station and the U. While it looks like I was right in predicting the demand for quick passage between the station and the U, I was wrong about the mechanism. Rather than an on-demand service to the hospital, there is a enough demand for a shuttle.
The 'directness' seems to be the important part. It may be useful to start thinking about high-capacity transit in the same way we think about air-lines--as rapid transit between hubs, rather than as service along a route. Certainly, it seems like a waste of time not to pick up passengers along the route, but it's a matter of aggregate delay. Stopping a shuttle with 20 people on it 6 times to pick up six passengers (even if the stop is only 10 seconds) is still 20+ minutes of delay.
But the power of a shuttle is that it can do just a little more than a plane (which is hub to hub). You can also have some stops 'after' the end of the 'direct' route, so you can distribute passengers around an area. It's a stopping pattern I'm seeing a lot in light rail---a long 'direct' section, and then a large number of stops downtown.
Direct: x--------------------------x
Direct Plus: x-----------------------------x--x--x--x
Local: x--x--x--x--x--x--x--x--x--x--x--x--x--x--x
Rapid: x----x----x----x----x----x----x----x----x
Too many stop: x-x-x-x-x-x-x-x-x--x-xx-x-x-x-x
(Presume each dash is 660'---a SLC block.)
I increasingly believe that a great deal of the value of fixed-guideway transit is coming from stop spacing. Expensive stations means fewer stations, which means fewer stops,which means better speeds, which means people are willing to walk further (or wait longer).