Reasoning starting from conclusions to lead to initial constraints is backwards reasoning. For example you don't talk about maintenance or productivity, and yet you end up making a choice without factoring this. Chances are, the choice in most codebases is made because of existing code and culture, not because of rational reasons.
But from the engineering side it's the only "everything" language. (There aren't any good GUI kits for C, and NVCC is C++)
Then how does C++ improve?
We are likely getting modules (and reflection) with the next iteration (C++20), which -- if it moves like the last two versions -- will be almost completed and already supported by GCC, VS and Clang in two years. Clang and VS2015 even support modules experimentally already.
Give me a tool that's: (i) as fast, (ii) as mature and well supported, (iii) as powerful as C++ and I will switch in a heartbeat. But until there is such an alternative it's futile to complain about the shortcomings of C++, because if you want the powerful, zero-cost abstractions, the mountains of support, and access to billions of existing lines of code, you pretty much have nowhere else to go.