For the most part I'm with you on this. Developers complain that it's too hard to work with, and I think they want to be able to store an object structure without having to worry about how it's stored. However it's the type of mapping they often have to do when working with the data on a screen, so I'm not sure why it's that hard.

And if it's too time consuming, why not have one person dedicated to writing the mapping in a data layer?

I just don't buy that it's a ton of time.

It is a knowledge investment, and maybe that's an issue. We don't have a way to quickly and easily train someone how to design a db.