Someone in an earlier Kotlin discussion (there has been quite a few just today) was praising how Kotlin solves the "1M$ mistake" having no null. He was also complaining that sadly is day-to-day job involves Go.

Unline Kotlin, Go does support len() (.length in Kotlin) on nil: https://play.golang.org/p/koqYbIdltw

In Kotlin, b.length can throw a NPE, only a b?.length won't (it will return zero or null, I have not understood that).

Together with the var/val debacle, it seems like it is designed to be very terse, which can lead to easier errors.

I don't think it solves the 1M$ mistake at all. And how could it, while maintaining full compatiblity with Java?