That's a tough one. Python's dynamic typing means that you can write concise code as long as everything is of the correct type. But the consequence of this is that it is easy to accidentally express the wrong thing and only find out about it at runtime.

Pony's type system requires that you be a bit more explicit, but the tradeoff is that you get better guarantees about the correctness of your program. As your program grows in size these guarantees become more and more useful, or at least that's what I've found.

One part about Pony that can seem like an impediment to expressiveness in Pony is reference capabilities. Pony forces you to think about how objects can be accessed from different actors, which can sometimes make it hard to do things that seem like they would be simple, especially if you are new to the language. The payoff for this is that once you get the reference capabilities correct the compiler guarantees that your program will not have race conditions. In some ways, I'd say that if you're writing multi-threaded code then expressiveness that doesn't offer any safety guarantees is a false economy, because you'll probably end up rewriting the "expressive" code many times to get things right.

Python's a great language, and I don't mean to disparage it. I've used it quite a bit in my career, and for certain types of problems it is still the first tool that I reach for. In my mind, Pony excels in some areas where Python is somewhat weak, so they can live comfortably together in your programming toolbox.