These arguments are not about the ‘goodness’ or ‘badness’ or actors in general, they are about their use in software that is trying to acheive ‘continuous delivery’.

Actors have a signature ‘Any => Unit’. This means that they accept anything and are all about the sideeffects. They require living a framework which means that almost anything that is an actor cannot be tested in a unit test (using the definiton of unit tests used in this policy) They make it very hard to ‘invert the pyramid’

Actors are a very low level abstraction and are excellent for very very low level programming. If we do want to use them, hey should almost always be in a library and expose their behavior to the code we write through APIs that totally hide their nature as actors from the users of that library.

There are as far as I know very few meaningful things that can be done by an Actor that cannot be handled by the use of Kleislis which are a type safe, easy to test, easy to reason about alternative: so easy that we use them all the time and don’t need a huge library and framework to work with them.