I remember taking a badly written Python project which was difficult to work on because it took hours to run and could fail at the very end due to a syntax error or other trivial kind of mistake. I needed some way to make it less difficult to change so I could change it to something good. I tried applying type hints because I thought this had minor impact for potentially some gain. In this case it didn't and what helped was to take every tiny bit that was unit-testable and make a test and then to mock the part of the process that took all the time (it was an android build system so I mocked the android build). Every bit of refactoring-for-testability made it easier to do the next bit.
So to me I feel that writing tests has more value than any other activity - I'm ready to use types when it makes my IDE work better and when I think it stops me making a mistake I've made before but for my money tests deliver more and I will rather work on them than trying to build an elaborate type system that models the problem well enough to prevent one from doing invalid things under any circumstances. For me, the types/classes I create are there to make my program understandable and manageable as it gets bigger.
I find it useful to be able to choose where to put my effort. When I worked on Java long ago I found the type system of the JDK libraries horrendously overdesigned and hard to use - flexibility is sometimes bad if you don't gain from it but are forced to pay the price anyhow.