Comment by scott_s

I thought the article's comparison of go routines to goto was fair, in the context made: when calling a function in a language that allows go-routine like things (including pthread_create), one can't know if background tasks will be spawned. This is similar in the free-form time of unrestricted goto that one could not know if control flow really would return to the point lexically after the function call. His proposed abstraction does allow one to look after a nursery block and know whether or not background tasks are still running.