The cognitive burden of developing in C++ anymore is excessive, but that appears impossible to recognize without completely stepping away from the language for a few years.
I loved the language; I even used it for web development for a couple of years and it was enjoyable to do so; but I did step away, and at this point I'd rather use Rust (a language I neither love nor find enjoyable).
Not sure the C++ committee have the capacity to "fix" this (whatever that may mean), nor whether the industry would let them, even if they were able to recognize the cognitive burden as a problem.
There is a certain masochism and macho attitude in C++, see the C++ frameworks of the 1990's like OWL, VCL and CSet++, where C++ was as expressive as you could expect from a .NET or Java framework a decade later, something that only lives on C++ Builder and Qt/QtCreator.
Both not really welcomed in most C++ circles, where naturally one codes close to C, with a thin layer abstraction wrapping OpenGL/Vulkan/DirectX in imGui for the ultimate performance, meanwhile the same authors use an Electron based application to write their code.
This is what made eventually move away into managed compiled languages, C++ isn't really the same as used to be, my next favourite programming language similar in spirit to Object Pascal, as in the last century of desktop GUI frameworks.
We're lucky that the LLM stuff came about after C++ was already in decline. Or could there be a resurgence of C++ if we can teach LLMs to tackle what humans can't?
I loved the language; I even used it for web development for a couple of years and it was enjoyable to do so; but I did step away, and at this point I'd rather use Rust (a language I neither love nor find enjoyable).
Not sure the C++ committee have the capacity to "fix" this (whatever that may mean), nor whether the industry would let them, even if they were able to recognize the cognitive burden as a problem.