It always felt wrong to start a programming language design already having a type system in mind. I couldn't put my finger on it before, but now I think I understand why.

There is well known tradeoff between exploration and exploitation when it comes to solution search.

During *exploration* you try to open undiscovered areas. Something that might not be very viable at the beginning but in the long run could lead to outstanding results. During *exploitation* you try to find best solutions in selected area, relying on already accumulated assets.

Eventually exploration and exploitation would be in conflict, since resources are limited. You can resolve this conflict differently depending on the task you're trying to accomplish.

If you want to design a PL that is more convenient for humans, starting with a type system in mind would pull you away from exploration too early. There might be a brilliant solution that is incompatible with that type system.

You can always think how to limit flexibility and reduce space for errors later.