Ask HN: Rails, Django, or Phoenix?

OK, I'd like to get serious about web development to launch a few side projects that I hope will become my main gig. The three stacks I'm considering tripling down on are Rails, Django, or Phoenix. I have played with all three, but playing isn't creating more complex, production projects.

What I like about each:

- Rails: It's highly productive and I currently know Ruby the most. Plenty of jobs, if needed.

- Django: It's fairly productive and I know Python. Faster than Rails. Can use data science/ML in my web apps. Python's adoption is growing fast. An okay amount of jobs, if needed.

- Phoenix: It's fairly productive and much faster than Django and Rails. Functional programming and the Erlang ecosystem could lead me to write better code. While I know Elixir the least, I like the language and don't mind dropping OOP.

What I don't like about each:

- Rails: So much magic. Slowest of 'em all. Ruby adoption is going down, so I feel I'm investing in a depreciating asset and Ruby is mostly used for web development so I can't reuse the language that much in other areas.

- Django: the most explicit (by design) so I feel like I'm hand holding Django. It's slower to develop than Rails and less flexible. I feel like I'm investing in old technology, right or wrong, that hasn't progressed as quickly as Rails has. In a way it's the opposite of Rails' situation. The language is going places unlike Ruby, but I'm not sure the framework is.

- Phoenix: it's a niche, there are fewer libraries, few jobs if I lose my current one, and there is so much to learn at once to get both Elixir and BEAM/OTP.

Do you have any thoughts on this? What would you do?