Sean and b.l.z: There is a very clear distinction between active amplifier and passive LCR networks.
First, the active circuit has much more non-linearity than the passive network, leading to different types of distortion. You get more mixing (IM distortion) with active components, and those are not directly related to the originating tones.
Secondly, most analog active crossovers are implemented with high-gain opamps, which means that the designer had to apply high-levels of feedback in order to achieve low gains (usually unity at the passband). There is a specific sonic signature to that type of circuit. Cheap implementations may have slow loop delay so the delay-induced distortion may be very crude. The high-speed opamps are better in that regard, but still - "no free lunch"...
Digital crossovers have digital issues and analog issues. Cheap implementations have all the "good" traits of digital audio, in terms of converter non-linearity, clock jitter and sometimes crosstalk of digital into the analog circuitry, which raises the noise floor. Just listen at high gain to low level passages, where the noise floor is most apparent.
Bottom line is: there are issues...
Of course passive crossovers are not a magic solution, but they may be the least of all evil. If the drivers are well matched (some brands modify drivers or develop their own) then the passive crossover can be simple and accurate.
Last thing to add about multi-amping. Splitting audio signals is a nasty thing to do. It's true that a multi-driver speaker does that too, but in that case it's a necessity, unless you accept the limitations of a single-driver speaker system. I believe most people don't.
With the passive approach we assume that the speaker designer has done a decent job and the response would be acceptable with most commercial amps. That's not always true, but at least there is a good chance this will happen.
With multi-amping, we not only split but also run different amps and cables for each "band". Each path is a different audio chain, with differences by design (like solid-state for bass and tubes for mids and treble) or due to production tollerances. So not only do you have to measure drivers and match the crossover, but you also have to consider the full chain.
It's all doable. In fact, with digital crossovers it can be measured and the filters re-calculated in real-time, like some systems adjust equalization for room modes. It's a convenient way to do things, if everything can be matched and the above issues resolved.
Don't misunderstand me. I'll agree that for some price points the active approach is a good solution. I'm not sure that's true for achieving the ultimate sound. All this is my own very subjective take on amplification.
