Fas42, assuming the goal for all equipment is as much accuracy as possible, I do lay the entire blame on speakers. Solid state electronics can be designed to be, for home audio purposes, perfectly neutral, in that their noise and distortions are inaudible, and their frequency responses are flat regardless of load.
Speakers, on the other hand, are usually "voiced" Sometimes this is because the designer wants to achieve a certain "sound", but they are also voiced to apply judgment of how much high frequency roll-off the designer wants to apply to make the speaker sound natural in a room with certain absorption assumptions. I suppose you could argue voicing in the bass might also be necessary, because of room variations.
The latest speakers, designed with the latest driver technology and sophisticated crossovers designed with the latest software, can produce awesomely flat frequency responses these days. The latest speaker designs seem to have frequency responses that remind me of curves we used to see for amps in the 1960's. (Distortion levels too.) But lots of speakers are still "voiced" to sound a certain way according to a designer's biases, and the best drivers and crossovers (and cabinets too) are actually very expensive, so a lot of high end speakers still have response curves that look like saddles (too much bass and too much treble), and compromised crossovers that produce anomalies at the crossover points.
The differences in solid state electronics are far smaller. With tube electronics, one can design them to be as neutral as SS types (VTL comes to mind), but things like SET amps will have nearly unpredictable interactions with a particular speaker.
