At the corner of Church and Market (above), you have transit choices. To get to Downtown San Francisco, you can take the J Church light rail line (the silver train), which ducks into a tunnel 2 blocks later and runs in a subway under Market Street. You could also take the F Market and Wharves streetcar (the orange train), which runs down Market on the surface. The J will get you to the heart of the financial district in 11 minutes. The F will get you to the same spot in 20 minutes.

There’s another set of transit choices at Balboa Park, 6 miles south. This is the end of the line for the J, 35 minutes from Downtown. But noone would make that whole trip on the J, because the BART heavy rail trains that stop here make the same trip in 15 minutes.

The BART line, in turn, ends at Milbrae, 13 stations south of Downtown It shares that station with Caltrain, the commuter train that comes in from San Jose and Gilroy. When its downtown extension is complete, Catrain will makes it from Milbrae to San Francisco in less than 20 minutes with only one stop; BART takes 33 minutes.

This might seem like a gratuitous duplication of transit. Market Street is the only street in the United States with three different rail transit systems along it — fast, slow, and medium. One might think that, given a faster and a slower transit line on the same street, people would ride the faster one. But ridership proves otherwise. All four run standing room only during rush hour: the F carries 19,000 trips a day, the J carries 18,000, BART carries 59,000 within San Francisco, and Caltrain carries 33,000.

What we’re seeing here is a common pattern in mature transit systems. I’ll call it hop, skip, and jump. For local trips, you need to provide a system with a lot of stops. But for longer trips, that gets too slow. So you need to provide another system with fewer stops, and probably another system with even fewer. Then you connect the systems.

Here’s how it works on Market Street: The F stops 15 times between Church and Embarcadero; the J in its subway stops only 4 times. If you want to get from one spot in Downtown San Francisco to another, the shorter walk to the F makes up for a trip that’s maybe 5 minutes slower. But if you’re riding the J in from Noe Valley, a trip that’s ten minutes slower each way makes a big difference.

There’s no way a “one service fits all” system would serve all these needs. If you were to replace the J with BART, a lot of people in the gaps between stations wouldn’t be within walking distance of transit anymore. If you replace BART with the J, then nobody would want to put up with the slow ride all the way from the airport.

It’s important that all these systems make it to Downtown San Francisco, the most important employment center hereabouts. It would be possible to stop Caltrain at Milbrae and ask everyone to transfer to BART. But that means a longer, less convenient trip, and fewer people would chose transit.

But it’s also important to connect the systems at their outer ends: if you live at Balboa Park and work in San Jose, at the other end of the Caltrain line, you shouldn’t need to travel north to Downtown in order to travel south. What looks like duplication is actually a series of different transit service serving different needs.

So, if you’re going to build multiple systems, which comes first?

There will always be political pressure to build the express system first, because fast is sexy. But doing that means only a few people will be able to walk to a station, and a system that’s completely dependent on local bus feeders likely won’t attract more people than the bus system it replaced. One could, of course, rely entirely on park-and-rides. But that still serves only the jobs that are right next to the few stations, and it tends to encourage more low-density development, the most expensive kind of urban form to serve with transit.

So you build the local transit first. You put high quality transit within walking distance of as many people as possible. That builds ridership, and that ridership then justifies the express service. And the local service makes that express service more useful to everyone who uses the system. An asphalt analogy: we have local streets, we have frontage roads, and we have freeways. All are useful; together they make the complete system. We build the local streets first, then the frontage roads, then the freeways.

Look at any mature transit system in the world, and you’ll see hop-skip-jump. In Boston and London, commuter rail and heavy rail run parallel to each other, with the heavy rail stopping more often. In Chicago and New York, you have heavy rail express lines running alongside heavy rail local lines. In Frankfurt and Toronto, streetcars run parallel to subways. Cities that have tried to do everything with one system — a popular idea in the 1970s and 1980s — are finding the limitations of that approach. That’s why Atlanta is considering a streetcar directly above the MATRA subway, and why Oakland is putting BRT alongside BART heavy rail. In transit, one size does not fit all. Am effective transit system is really multiple systems, serving multiple roles, all linked together: hop. skip. jump.