It's because Cornelius' algorithms are so clever and adaptive that they work over a wide range of settings and program material. It's amazing to watch AGCs into 20 dB of gain reduction, but without any artifacts you would associate with deep GR!
CRL was doing that 25 years ago, with their linearized multiband AGC.
You can drive 'em up to the eyeballs, and they still sound smooth. The G/R meter on the TVS-3001 goes up to 40 dB!
Driving Multiband AGC's harder vs. a wideband are two different animals. When you split audio into multiple bands, it is always easier to drive multiband AGC's harder as each band only has to deal with a small segment of the overall signal. Wideband has to do everything, and (IMO) has to deal with this wide range gracefully.
The wideband AGC in the 11 is just about as complicated as the entire multiband sections...combined! It took just as long to develop as the rest of the multiband sections. This was no trivial achievement!
CRL's (to me, anyway) have always been sort of a one trick pony. They work OK for certain programs / programming styles, but as a general processor for everything, they fall short. I never liked the "murky" sound they had....lack of transient detail, etc. But that's just me.
A CRL driven harder to me always sounded like a CRL driven harder. Yes, you probably could get 40 dB of gain reduction...but it didn't mean you REALLY could in any practical sense. It may sound OK that way on some material, but to run them that way as a general rule would be a different discussion all together.
Processing is subjective, and what I don't care for is someone else's pot of gold.
The wideband AGC in the 11 is one that I felt would be the perfect match to my multiband AGC designs. Much of my multiband AGC concepts pre-date DSP processing by quite a bit. My challenge back in the analog day was to develop a wideband AGC to go with it. To build the 11's wideband AGC in analog would require a chassis the size of an 8100, and then some!
The era of powerful DSP's have made my wideband AGC a possibility, and a reality. Pegging the meter out at 20 dB of GR, and having the audio sound no different than if the AGC were running at 5 dB of GR was the goal here. All while easily handling dynamic / highly transient material with ease was the icing on the cake.