Alternatively, many people see value in focusing on what they develop and not have to bother studying the fine details of the underlying platforms they use. As someone who live deep down in detail and assist others using tools in the whole range from IDEs to cli, I have no disrespect for engineers who won't bother spending their time on knowing the subtitlities of the systems where their code will run.
>Bad analogy. Producing it would be closer to developing the blueprint.
Software tools are far from blueprints that are done once, they require constant maintenance to be compatible with changes in other tools and environments, bug and security fixing as well as implementing new features that users request.
Software development is extremely expensive, libre software is free only because someone is paying the cost of production and prefer to distribute it for free. Probably most of the open source software today is paid for by big companies, and their aim is usually to gain something from the investment. Docker wasn't developed as a manifestation of free speech, nor was Kubernetes born under GNU's roof. If not for the piles of money Google and Red Hat spent on it, Kubernetes couldn't be anything resembling the amazing beast that it is.
Docker was developed because a cloud provider (Dotcloud) wanted a better way to package their own and their customers' software. As it turned out, Docker was succesful while Dotcloud failed spectacularly. So Docker became the main product.. and now that failed too, as of a few months ago.