Theoretically, I'm technology agnostic when it comes to mass transit. Light rail, heavy rail, monorail, no-rail—I don't give a shit how many rails you have as long as it's grade separated. That's the secret to reliable transit: Not getting caught in traffic.
So yeah, theoretically, as Paul points out, true Bus Rapid Transit can work great. Until it doesn't. For the very flexibility that makes BRT cheaper to build also makes it cheaper to cut corners and eventually dismantle:
Wherever you build lanes that cars could use, car drivers will want to use them, and will exert political pressure to do so. Every BRT project that exists or is planned anywhere could be converted to a road for cars, without spending an additional dollar on construction.
This is not to say that BRT is useless. It certainly is not. BRT belongs in all our big cities, as one piece of a larger multimodal transit network. But the same flexibility and low cost that makes BRT attractive in many locations is simultaneously the reason it cannot be trusted to deliver on long-term promises in the same way as rail. It is easy to eliminate, and it has too great a history of being eliminated.
If you think about it, that's kinda the story behind the I-90 bridge's center HOV lanes. Originally built with the understanding that these lanes would be converted to transit, pro-car patriots like Kemper Freeman have fought and fought for BRT over the bridge instead of rail, knowing full well that this would leave the lanes open to cars. Angry drivers, stuck in bridge traffic while the center lanes remained virtually empty, would demand it.
But lay down tracks on the those lanes and what's done is done. (Well, almost.)
Technically, BRT could work. But politically, not so much. Politicians looking to do rapid transit on the cheap will inevitably cheapen BRT until it isn't really rapid at all, and any dedicated lanes will face a constant battle to fend off HOV and SOV encroachment.
So give me a rail or two, and let's be done with it.