When you think about transit technologies, how do you categorize them? And why?
Have a look at this first table, which sorts services according to the exclusivity of their right of way. The terms Class A, B, and C are from Vukan Vuchic, describing the basic categories of "what can get in the way" of a transit service.
Is this table two rows, each divided into three columns? Or three columns, each with two rows? Which distinction is more fundamental, and which is secondary?
Right-of-Way Class vs Rail-Bus Distinction
Exclusive right-of-way and separated from cross traffic
Exclusive right-of-way, NOT separated from cross traffic.
Mixed with traffic, including mixed with pedestrians.
Most rail rapid transit, using “third rail” power sources. Most classic “subway” and “metro” systems.
Most “light rail” in surface operations. Parts of some European and Australian tram networks.
Most North American streetcars. Many European and Australian trams.
Separated busways: (Brisbane, Ottawa, Bogotá, and segments in Los Angeles, Pittsburgh) Freeway bus/HOT lanes.
At-grade busways: Los Angeles Orange Line, Western Sydney busways, etc.
Buses in mixed traffic.
Well, if your objective is to get where you're going fast and reliably, the Right-of-Way Class tells you a lot about a services's potential to do that, while the rail-bus distinction, in isolation, tells you nothing. The fact is, both rail and bus technologies are capable of the complete spectrum of possibilities. Both can average 6 mph (10 km/h) in Class C situations, and both can run Class A at 60 mph (100 km/h) or more.
RIght-of-way isn't the only thing that matters for getting you where you're going. There's also stop spacing, with its inevitable tradeoff between speed and local access.
Stop Spacing vs Rail-Bus Distinction
(faster = fewer stops)
(slower = more stops)
(one long nonstop segment)
“Subway”, “Metro”, some commuter rail.
Tram / Streetcar
Some commuter rail.
Bus Rapid Transit,
Commuter express bus (often on freeway)
… and of course there are other essential distinctions like frequency, which are also entirely separable from rail and bus technologies.
UPDATE: Please note, yet again, that contrary to early comments I am NOT claiming that these are the only distinctions that matter. As I laid out in some detail here, there are several distinctions that matter. In fact, one of the reasons that people cling so hard to the rail-bus distinction is that the other crucial distinctions are a little more complicated and require some thought, and it's hard to think about this stuff in the political space where decisions get made.
Rail services do tend to be presented in ways that "package" the various crucial dimensions of usefulness. Typical metro systems, for example, are guaranteed to be frequent, with rapid stop spacing, and Class A right of way, because all three are intrinsic to the metro technology, so there's a psychological "packaging" effect when you see a metro map; you can be confident that this means a certain level of service.
I think these tables are interesting because now and then I meet someone who divides the world rigidly into rail and bus, often aligning these categories with a rigid class distinction (William Lind, say) or simply claiming that rail does beautiful things and buses don't. In that view, the different columns of these tables are secondary and interchangeable, while the rows express something absolute.
Patrick Condon, for example, proposes that instead of building one rapid transit line (Class A, rapid stops) we could just build lots of streetcars (mostly Class C, local stops). That can make sense if you judge technologies entirely on their influence on urban form, and prefer the kind of form that seems to arise from streetcars. But it will be just incoherent to a transit planner who's been trained to help people get places, and wonders if he's being told that nobody cares about that anymore. Because if you do care about personal mobility — people getting where they're going, now, today — you have to care about the columns.
I hope to leave this topic for a while, but I do think it's worth coming back to tables like this to ask yourself: Do I tend to divide the world according to the rows first, or the columns? If so, why? Is my way of slicing this table something I've discovered about the world, or something my mind is imposing on it?