Posted a draft of my new article on arxiv.org. Here is the link

arXiv:1308.5626 http://arxiv.org/abs/1308.5626

The statistical approach to preconditioning is somewhat weird to the preconditionists however I've got something that works (actually nails it down!)