Standards for 'Normal Hearing'
You ask a good question, one that audiologists have been asking themselves for many years. The zero figure, as you imply, is simply an average of the thresholds obtained by "healthy young adults" under optimal conditions. The results themselves, however, fall in a "bell-shaped curve," in exactly the same way that other human dimensions do (think "average" height, weight, age for example).
The lower limits of what is considered normal gets influenced by how the information is going to be used. We know, for example, that children with even minimal hearing losses (e.g. within the "normal range" you outline) will, on the average, perform slightly poorer academically than other children, so we should watch them just a bit more closely. Young people should have better (lower) thresholds than an older person, although both can be perfectly normal for their age. For an adult, the best judgment I can offer is actual communication difficulty.