Could someone please help me to understand stick-free stability a bit better? (I'm thinking mainly about pitch stability and horizontal tail sizing but I realise it applies to other control surfaces.)

Have I got this right:

Stick free stability is only different to stick-fixed stability in reversible systems. Thus in non-reversible systems stick free = stick fixed = no difference = one can have a smaller overall tail size than with a reversible control system (since stab and elevator count together).

In a reversible system, stick free is a hands-off scenario where the elevator will float (weathervane) with the relative wind. The elevator will therefore play no part in stability, stick free, and therefore when sizing the tail one discounts the elevator and sizes just the fixed stabiliser (or else one ends up with a tailplane that is too small and has consequent reduction of stability, which, particularly in turbulence, then requires constant control input to make up the difference).

If the fixed stabiliser is too small but overall sizing correct then you have correct stick fixed stability but reduced stick free stability. You can't use trim tabs or trim devices such as springs to make up the difference. For better stick free stability you just have to increase the size of the fixed stabiliser or make the control system non-reversible.

If the above is correct then I'm having trouble understanding how an all-flying tail or stabilator provides any stick free stability at all in a reversible system. Have I got this right:

The stabilator is hinged at 20-25% mac. It therefore provides little to no control feedback and is also free to weathervane. Stick force is usually created using an anti-servo tab. This tab is attached to a movable point and thereby also acts as a trim.

But since any tab is going to be subject to the same relative wind as the stabilator, what stops the stabilator weathervaning? In other words, what gives the stabilator its stick-free stability?

I must be missing something very obvious because if an elevator can float and therefore has to be removed from the stability equations for tail sizing then surely a stabilator must behave in the same way and would provide virtually no stick free stability for the aircraft (since the whole surface is floating, not just an elevator). Yet that must be wrong since there are many light aircraft using reversible control systems with stabilators and which fly along quite happily with excellent hands-off stability.

What am I missing? Thanks!