An interesting thought: historically, a major consideration for companies choosing what to use is how many engineers use it (because that will make hiring much easier/harder). Going forward, this becomes "how big of a corpus of this code is out there", because it impacts how proficient are likely to be at using it.