It's been a while since I've answered a question on a blog, but Adam Haverson asked me the following:

Hey man - it's been a long time! We weren't really close at school but you are one of the few people I know who posts thoughtfully on FB & I've enjoyed following you & your conversations with friends. Been meaning to ask you - what is your take on the Berkshire/Morgan/Amazon healthcare initiative? I saw they just appointed a CEO so it must be moving ahead.

Atul Gawande is certainly a respected figure in medical literati; I'm a little surprised by the pick, but there are few forward thinking medical doctors that can do things that the medical community may strongly disagree with - which may be a good reason to pick him. Other major medical systems like Geisinger, Cleveland Clinic, Mayo - that are physician led - were probably mooted, but they picked Gawande.

It's impossible to know precisely what Amazon/JPM/Berkshire will do - and I suspect they can'd know precisely either. Startups by nature change as they develop. That said, given the players: Logistics, finance, investment/insurance - they pull together very interesting domains for one of the thorniest problems in modern technology and business... how do you create a system that has efficient healthcare delivery?

While this is completely out of bounds for a full discussion here (where is a good forum for this?)
Healthcare in a nutshell

Potentially limitless spending
Huge range of cost/impact curves
Opaque and sometimes quixotic price signalling
Incomplete metrics at all levels of care.
Limited vertical integration
Most systems have payor-provider(multiple tiers)-admin tiers - patient (who may not be the decision maker).

Presently, Geisinger and Kaiser are the most well known vertically integrated healthcare systems in the US. By having insurance and provision, they have the potential to better understand where the money goes, and they have an incentive to spend less to get equivalent results - other systems have no such incentives. But, they lack complete data.

By having a complete vertical, including deep, big data analytics that takes into account logistical problems, the potential to have a much more "aware" system of care arises.

That's my bet on where the collaboration is trying to lead. Provide the insurance, the care, track the delivery, outcomes and feed it back into not only your actuarial assessment for premiums, but also the care delivery - the how, when and how much. This allows internal pricing signaling. Why pay for things that don't work?

The only problem with all of these lean-ish models (and our present system, TBH) is that it's inefficient at dealing with new innovative and effective things... That's something that I wish would be addressed more adroitly.