This is purely speculation, but I wonder if this "root access" doesn't also come with downsides, e.g. more gene defects, which would help explain the short lifespan of most cephalopods.
IIRC it greatly limits the long-term flexibility, because most DNA changes will either break RNA edition or cascade into undesirable post-edition features.
It's like taking a program at v0.1 and then developing the entire thing via binary patches, at one point you've got so much binary patching going on any edition and recompilation of the source makes them not work.
