I think the reason is more that the goals of the C committee align better with the goals of C programmers (which is a language that's a tool and not a playground for language designers).
Unlike C++, C also remained a simple language (which is definitely a side effect of the "conservatism" of the C committee).
That's what he means; like the puddle which marvelously fits its own hole, anyone for whom C in all of its aspects is not very close to optimal has long ago moved to another language.
WG14 has only "worked" fairly well insofar as it did absolutely nothing but file the serial numbers off of things standardized in C++ and release them. The biggest C-original features (type generic macros, annex K) are all gigantic boondoggles. Even many of the features that were lifted from C++ were poorly thought out (e.g. compatibility between <stdatomic.h> and <atomic>).
The biggest C-original feature since C89 is easily designated-initialization in C99, and WG21 couldn't even get that simple thing right when they tried to copy it into C++20 two decades later (even though Clang had a working implementation in C++ mode since forever which just had to be copied by the standard).