The argument starts with the thesis that "you should use modern C++ everywhere" is generally meant to mean "you should use value semantics everywhere."
This is odd to me since my C++ experience is from the 1990s when "you should use modern C++ everywhere" often meant "use string instead of char *" and "use new instead of malloc". I see the emphasis on using value semantics as being a delta on top of the older meaning, where the old meaning still exist, but the speaker just naturally assumes them to be correct.
I don't have much experience with modern C++ but I thought the speaker's complaints about "this begin() and end() everywhere" has been (mostly?) resolved with range-based for loops added in C++11 and various improvements since then, and the complaint about string split has been resolved with std::ranges::views::split in C++20, like the example at https://en.cppreference.com/w/cpp/ranges/split_view .
Which makes it sound like those issues can be resolved by ... using modern C++ everywhere.
The author correctly highlights how some standard library components have faster equivalents elsewhere, concluding "so they provide the STL for you and it's a piece of crap, right? But you have to use it because it's the modern way to do thing."
Thing is, the author already pointed to Boost as a faster alternative for map, for example. Using modern C++ does not mean only using the STL.
The author seems to think that people using Boost are generally considered dinosaurs. Which doesn't fit my understanding.
I stopped listening about 5 minutes in when the author said the concepts behind the movement to use modern C++ are "virtue signalling". It's a culture war term that has no real useful meaning here, and a far more appropriate term is "shibboleth".
