Monday, August 26, 2013
Challenges designed to spur innovative uses of data are springing up frequently. These are contests, sponsored by a mix of government agencies, industry, foundations, a variety of not-for-profit groups, or even individuals. They offer prize money or other incentives for people or teams to come up with solutions to a wide range of problems. In addition to grand prizes, they often offer many smaller prizes or networking opportunities. The latest such challenge to come to my attention was announced August 19 by the Knight Foundation: $2 million for answers to the question "how can we harnass data and information for the health of communities?" Companion prizes, of up to $200,000, are also being offered by the Robert Wood Johnson Foundation and the California Healthcare Foundation.
Such challenges are also a favorite of the Obama administration. From promoting Obamacare among younger Americans (over 100 prizes of up to $30,000)--now entered by Karl Rove's Crossroads group--to arms control and identification of sewer overflows, the federal government has gone in for challenges big time. Check out challenge.gov to see the impressive list. Use of information and technological innovation feature prominently in the challenges, but there is also a challenge for "innovative communications strategies to target individuals who experience high levels of involuntary breaks ("churn") in health insurance coverage" (from SAMHSA), a challenge to design posters to educate kids about concussions (from CDC), a challenge to develop a robot that can retrieve samples (from NASA), and a challenge to use technology for atrocity prevention (from USAID and Humanity United). All in all, some 285 challenges sponsored by the federal government are currently active, although for some the submission period has closed.
These challenges are entertaining, call on crowdsourcing for knowledge production, find new sources of expertise way beyond the Beltway or even US borders, encourage private sector groups rather than government to bear costs and risks of development (or failure), and may bring novel and highly useful ideas to light. So what's not to like? I may be just grumpy today, but I have some serious worries about the rush to challenges as a way to solve persistent or apparently intractable problems.
Challenges may be more hype than achievement, more heat than ultimate light. They may emphasize the quick and clever--the nifty over the difficult or profound. They may substitute the excitement of awarding and winning a prize for making real progress on a problem. Most troubling to me, however, is the challenge strategy's potential to skew what government finds interesting and what it is willing to do. Many challenges have private partners in industry, appear likely to result in for-profit products, or set aside values that may be more difficult to quantify or instantiate.
Take the HHS Datapalooza, for example. Now entering its fifth year, the Datapalooza is an annual celebration of innovations designed to make use of health data available from a wide variety of sources, including government health data. "Data liberation" is the watchword, with periodic but limited references to data protection, security and privacy. A look at the 2013 agenda reveals a planning committee representing start-ups and venture capital. It also reveals a $500,000 prize awarded by Heritage Provider Network, a managed care organization originally located in Southern California but now expanding in markets in Arizona and New York and serving many Medicare Advantage patients. The prize was for a model to predict hospitalizations accurately and in advance--so that they could be avoided. The winning team, powerdot, didn't reach the benchmark needed to win the full $3m prize. So . . . Heritage is continuing the competition, making more (and apparently no longer deidentified) data available to a select set of leading competitors in the original competition in order to improve the accuracy of the modeling. (A description of deidentification methods for the data made available to all entrants in the original competition is available here.) There are of course real advantages in developing a good predictive model--for patients in avoiding hospitalizations, and for Heritage in saving money in patient care. This is potentially a "win win"--as Mark Wagar, the executive awarding the prize stated, "it's not just about the money; it's personal." But "it's not just about the money" is telling: the risk of these challenges is that they are about the money, and that the money will come to dominate personal or other values unless we are careful.
Solutions, if my concerns are well-founded? Trying to turn back the disruptive clock and fight the appeal of challenges is probably futile--although perhaps some of the initial enthusiasm may wane. One solution is to join in--after all, challenges are infectious and potentially innovative--encouraging more challenges aimed at different problems--say, challenges for privacy or security protection alongside challenges for data liberation and use. Or, challenges for improving patient understanding of their health conditions and informed consent to strategies for managing them--as some of the challenges aimed at patients with diabetes illustrate. Another solution is to watch very carefully what challenges are offered, who funds them, who wins them, and what is ultimately achieved by them.