I think we cannot currently determine which modern languages offers real extra value. I think there is a bias to exclude things where the flaws are known for things where the flaws are unknown.

My more general thesis is programming is in its infancy and nobody knows what they're doing. Bloodletting was used frequently and unnecessarily in medicine for 2000 years. I wouldn't be surprised if we do something similar with programming. We need the programming equivalent of double-blind studies.

If I'm right and nobody has a clue, maybe it doesn't really matter what language, so just choose whatever you feel like, flip a coin, or whatever. I always try to remind myself that code is just a tool to deliver value, not an end in itself - I'm really not experienced enough to be giving out this kind of advice, but, hey, it's the internet 😛