This articulate post by Martin Smith at Concirrus says nothing new.  And that's why it's a big deal.

Why, still, do we need to write posts about this topic.  Most of the technology and data Martin talks about is readily available now - and has been for some years - but why are we not seeing the evidence of people doing what he says?

Much like the ever-heralded end to this cycle of the soft marine market, the promise of data-driven marine risk selection appears to always be just over the horizon.