I think it would be more constructive to propose a constrained subset, get rid of legacy, propose default behaviours for parts that are undefined and so on. Stuff that does not belong to this subset can still be used within an unconstrained clause.

If I read e.g. "Why build Carbon?" on https://github.com/carbon-language/carbon-lang I dont find any concrete reasons or proves that show why C++ should be abandoned for something completely new but somehow very similar. There are so many question around the actual language definition like who maintains it, how is the standardization process. Who will adapt to it or to something else. ... This seems all so utopic.

I have been using C++ for many years with fantastic success. It allows me to create all I want/need from embedded to high performance projects. That you may shot you in the foot is in fact not bad. It compels you to be careful and think deeply about the design and only use what you completely understand.

With other languages your program might not crash but probably will do wrong things like eat up memory or access still references outdated data. You will find error much later in the process when it costs multiple magnitudes more.