While I agree with most of the post, it's just the tip of the iceberg of the discussion. Stuff like interfaces, traits, generics, concepts, inference, duck-typing, implicit conversions, etc. all fuzzyfy the notion of "Strong Typing" in various ways, some more than others.

The real discussion is not about types vs no-types, it's about the role and place for each of these bending of the rule and their associated tradeoff in different contexts. Op even takes inference for granted towards the end of the post, but make no mistake, that's a form of weak(er) typing.

For example, writing template code in C++ without frequent use of type inference (aka. auto) can be an absolute nightmare (I suspect that extends to generics programming at large, but C++ is where I have the most experience here), and the legibility gains from making use of it for iterators is broadly acknowledged as being easily worth the obfuscation for scope-limited variables. But there's also a strong case being made to avoid its use entirely in most other contexts.