We're commenting on an article pretty much saying we have models which estimate in the right directing, but don't go far enough. Now we can learn from the changes and improve the models.

If you're discounting a whole field of research, can you provide more reasoning for why it can't be improved from the current results?