McIntosh is simply making some different design decisions in using autoformers. Comparing an autoformer amp to a Krell, for example, one notices a distinct difference in how they're rated. The Krell tries, and usually succeeds, in having the ability to output a consistent maximum voltage level regardless of the impedance of the load. Because the voltage limit is fixed the maximum output current varies. So the Krell puts out X volts into 8/4/2/1 ohms, and that might correspond to 300/600/1200/2400 watts, for example, depending on the load.
The autoformer amps have a different design goal, output a consistent maximum current regardless of the load's impedance, using the autoformer to "match" the load to the amp's output stage. So a McIntosh amp might be rated at 1200 watts into 8, 4, and 2 ohms.
Which is "better"? Well, that depends. Assuming your maximum listening levels cause a current draw well within the limits of the amplifiers, I can't believe there will be much of an audible difference in sound attributable to the autoformers. They don't add distortion, but they *can* affect the system's frequency response.
The autoformer adds some resistance in the output path, so the so-called damping factor, or the output impedance of the amplifier, is probably going to be a lot higher than a direct-coupled amp, like the Krell. It's all that wire in the autoformer; it adds resistance. How is this audible? It may not be, but if you have speakers that present a very low impedance load you could have a roll-off or dip at those frequencies where that happens, I suppose. Will the roll-off be audible? I don't know, but some speakers do present very low impedance at some frequencies (early Legacy Focus, Apogees, and Wilson Watts come to mind), so if you own one of those I'd stay away from autoformer amps, but that's not intended to be some deep pronouncement.
Autoformer designs allow a manufacturer to spend more on producing high current into higher impedance loads, so assuming there was some real relationship between cost and price (there really isn't for products like this) for X dollars you could get more watts into 8 ohms than with the non-autoformer design. Take a look at the MC1.2 for example. 1200 watts/ch into 8 ohms. What does Krell offer that puts out 1200 watts into 8 ohms, and how much does it cost? Of course, the MC1.2 will still put out 1200 watts into 4 ohms, while the Krell is putting out twice the watts into 4 ohms, but since the MC1.2 starts out so much higher in power, you don't really see the two design approaches differ until very low impedances, when even a low-end Krell amp is putting out over 2000 watts, assuming you have a 220v/30amp power line running to the silly thing.
So are autoformer amps better or worse than direct-coupled amps? For most speaker loads you'll never hear a difference, but I admit to strongly preferring the direct-coupled approach, assuming you can afford a good enough amp that it does indeed act like a good voltage source, like a Krell. I think a very low output impedance is indicative of a better design, but I wouldn't bet five cents on my ability to tell the difference in a double-blind test. It's just my opinion, but I'm the buyer so I get to have one that counts!
