I'm a bit late on this thread but I don't agree with the basic premise that there is audio feedback in the LA-2A.
A feedback circuit is like most Vari-mu's in that a signal is taken from the output, rectified and applied to the input. You can never put a volume control in the middle of one of these because as you turned it down you would lose compression. Feed forward is when a sample is taken from the input instead of the output like the LA-2A, the output can then be controlled separately.
The LA-2A takes a sample of the input and amplifies it to drive an EL panel attached to a specific LDR (which is the vital key to the whole performance). In fact it drives 2 units as the other one controls the meter in a bridge circuit. The result of all this is that the input to the main amp is controlled. You can't control the output separately on a feedback design except by pads.
Look up some old pre-war compressor designs and quite a few were feedforward back then, it was not until Western Electric and Presto came up with their classic feedback design that the other designs went out of fashion.
Incidently, you do not have to rely on one source for the EL/LDR, that can be successfully modified to a lightfast tube containing a green LED and the original Allied Signal LDR fitted to oppose it. The drive amp has to be modified to use a 3W output TX to drive the LEDs. I made one of these 4 years ago and have used it for every recording since. The main advantage is that an LED lasts for at least 50,000 hours whereas EL panels have to be replaced more often because they wear out (get dark) after a while. They were designed for 50/60Hz so 600Hz will wear out 10 x faster (i.e."wear" is frequency dependent) . This is due to the migration of Cu atoms within the EL phosphor. LED's are just as fast as EL panels too, probably faster but it makes no difference because the "rise time" of the LDR is what controls the dynamics of the compression.
best
DaveP
Head of R&D
Phosphor Technology
Stevenage
UK