The company offered a $1 million prize to improve its algorithm. Why not do that for risk adjustment?
Netflix ran a world-wide contest to improve Cinematch, its proprietary algorithm for predicting how users would rate films they’d never seen. The grand prize was $1 million. Within a year, over 2,000 separate teams from 150 countries had submitted more than 13,000 algorithms. Eventually, the winning team, which included researchers from AT&T Labs, Yahoo and an Austrian consulting firm, improved the algorithm by more than 10%. For a tiny cost, Netflix got a huge amount of computer-science research that even its highly skilled employees could not perform.
.. Medicare should do the same: create a contest open to anyone in the world who can beat its current risk-adjustment model. To ensure fairness and encourage competition, administering it should be outsourced to the X Prize Foundation or a similar group. The winner should be able to use objective patient data to account for at least 45% of the spending variation caused by disease. Medicare could award a $10 million grand prize and several million for second and third place. It also should be required to adopt one of the top methods. (There may be practical reasons not to pick the winner.)