well seems to be a belief I've seen expressed here, and elsewhere a few times (look at the OP of this thread: "I understand that we can't separate the preamp and power amp due to IP protection"...., or
here (I know there's others I've seen)). I keep asking about it because I wonder what that belief is based on - for me the wish is more related to wanting to isolate p.a. and have mix/matchable p.a. models, but, kinda the same thing since if one could isolate p.a., one could mimic an fx loop by having one amp block before fx with pre on/p.a. off and a 2nd one after with pre off/p.a. on (personally I'm not interested in this).
I can't find anything accurate or definitive referring to IP protection for the Amp block in the Wiki or the forums. Perhaps it was said somewhere else, or someone conflated a statement about an Amp-block-based plugin with the Amp block inside the modeler.
It'd be pretty hard to figure out anything about the architecture or code by dropping a block between the pre-amp and power-amp sections since that block would have to be inside the firmware already, and we can't do anything to peak inside unless Fractal gives us access to it. The sound is a stream of numbers at that point.
If there was an actual effects loop between the two sections to send/receive audio to an external effects loop, then there'd be access to the reconstituted audio signal coming from a DAC, but even that wouldn't expose what sort of code was written, only the effect of changing parameters in the code. Like listening to, or using an oscilloscope on a modern guitar amp and turning the knobs and flipping switches, without ever having had access to the schematic or foreknowledge of the circuit design or what components are inside, it'd be hard to understand what's happening internally; It'd be like black magic because there's a lot happening dynamically. Reverse engineering the internal code from how the waveform changes would be the same. When Cliff implemented Cygnus, it was a eureka moment when he noticed that the amp block's decay didn't sound exactly like a real amp. It was a tiny detail but he noticed the dynamic nature of the change and knew the code and what it was doing, and wanted to see if he could improve the code to model the real power-amp's behavior. IIRC, that led to changes in the pre-amp's code, and the aggregate difference is like the butterfly effect inside just that one block; a tiny input change can have a major effect on the output.
If someone wanted to know the algorithms, it seems like a much faster path would be to hook up an in-circuit emulator and capture the machine code then try to rebuild the source from that. And, by the time they'd get done, we'd be running on firmware that is much more advanced, again.
I think the reason for no effect-loop in the Cab block comes back down to oversampling and the impact on the CPU load. Whether it's a virtual loop or an external loop, the end result would be a hit on the CPU and the attending loss of quality, or added latency, and neither is acceptable because of Cliff's goals, and the continuous demands from the users for more available CPU and less latency.
It's all an interesting brain exercise but, as is, the system sounds like a pedal board into a great amp without an effects loop being recorded in a studio with post-effects, and anything else would be more icing on top of the existing icing on a cake. That's a pretty good thing already.