Zero feedback and class A,which is better?

I expect digging around the archives could lead you to a respectable body of opinion, especially if you widen your search to include other ML models which have similar minimum impedance levels.

The argument of zero negative feedback and Class A are also in the archives (with lots of contribution from the aforementioned Ralph, a respected designer who also makes for an interesting read when he gets involved in a thread). Long story short, I expect a great deal of the result of using one, both, or neither attribute successfully is about implementation, not about inherent characteristics.

And with regard to the 5 watts, unless you listen to thrash metal or bass drums constantly, the vast majority (in terms of minutes listened) of what you listen to will probably not exceed a 5 watt pull, it's just that some of it will, and when it does, it will exceed that by a long way, so yes, you need more wattage than that.

I hadn't thought about matching OTLs with auto-formers as I would have thought the autoformers negate the OTL-ness but the answer is probably that I was not thinking creatively enough. Sounds like a great idea.

Ralph,
Is the reason why SS amps sound "bright" on ESLs (because of the constant power demand) because ESLs see low impedance at high frequencies, and the SS amps double down in power at the low impedances, making SS amps, on an ESL-specific frequency response curve, relatively "bass-shy"?