February 6, 2015
By William A. Masters, Friedman School of Nutrition and Department of Economics, Tufts University
This post draws on research into pull mechanisms described here: http://sites.tufts.edu/willmasters/research/prizes
Pull mechanisms like AgResults succeed by drawing a target, and paying people to reach it. The most successful such contests attract people who might otherwise be doing other things, elicit their best efforts – and then identify not just who won the contest, but how they came to perform so well, so that winning strategies can be scaled up and sustained over time.
The widespread use of pull mechanisms has generated useful evidence about what kinds of contests attract the most effort and yield the most sustained impact. I have been studying contest design and potential pull mechanisms in agriculture for over a decade, motivated by a simple puzzle: contests established by philanthropic or government donors almost always offer one first prize and perhaps a few runner-up prizes, but they typically pay nothing to contestants who did less well. Why is that, when real-life markets typically reward many participants, not just the best?
What’s puzzling is that the everyday economy involves many winner-take-all choices, each of which can be seen as a contest among discrete options. But then people form companies and join markets in part to smooth things out. Successful economies make earnings proportional to success over many different contests, rather than just one, and economic policy aims to make the whole market grow. Would donor-funded contests achieve greater impacts if they imitated real life, and offered payments that are proportional to success rather than winner-take-all?
My initial work on contest design focuses on the historical record of pull mechanisms in agriculture, such as the 320,000 francs offered in 1874 by the French Ministry of Agriculture for a method to stop a devastating wine blight. The prize was never awarded, in part because the problem was actually solved by spreading an old technique: grafting French vines onto blight-resistant roots. For the prize contest to have rewarded that solution, it might have targeted the ultimate goal (prevention of yield loss), and divided the 320,000 francs among many winners according to their share of total acres covered.
Soon after my initial research on agricultural pull mechanisms was published in 2003 and 2005, I was invited to visit the X Prize Foundation in Los Angeles to discuss a possible new contest aimed at global poverty reduction. The founder of X Prizes, Peter Diamandis, had spurred extraordinary breakthroughs in space travel by organizing the Ansari X Prize for suborbital flight. He and many others hoped that their success with rocket science would translate to other domains, like global poverty.
I asked Diamandis what he thought of paying prizes proportionally to success. He laughed, and said: “You don’t know anything about prizes. It’s about the hero. It’s about the story. Without a winner, it’s just technology.”
Diamandis was right, of course. His X Prizes brilliantly capture the power of pull mechanisms, leveraging an initial $1m grant from the Ansari family into over $1b in other investment. He used the initial fund to buy an insurance policy paying $10m to anyone who met his space flight goal, which in turn attracted over $100m from contestants. The first to reach Diamandis’ target spent about $25m doing so, and then used the prestige from winning to earn many times that amount from Richard Branson’s Virgin group and then the aerospace giant Northrop Grumman.
But what if a donor is not so interested in having one winner, but wants to promote less charismatic kinds of widespread, incremental innovation? Back at the lab, with collaborators Tim Cason and Roman Sheremeta, I constructed a set of experiments asking ordinary people to compete for cash prizes. We manipulated the contest rules so as to measure which kinds of pull mechanisms attracted the most entrants and the most effort in pursuit of the sponsor’s goal. So far, we’ve found that some people really are motivated to be the lone winner, even if it costs them money -- but that winner-take-all contests discourage potentially valuable entrants who think they are unlikely to win, and on average don’t elicit any more effort than prizes paid in proportion to success.
The bottom line is that contest designers face a stark choice: they can follow Diamandis and the X Prize Foundation, constructing winner-take-all contests to identify an individual hero. Or they can imitate actual markets, and distribute funds more widely among many innovators in proportion to measured gains. In some markets, especially in agriculture where acreage and yields are relatively easy to measure, proportionality is possible. Agricultural innovations tend to be local and incremental, with small improvements building on each other. Rewarding many small efforts may be boring, but it’s what enables agricultural markets to feed the world – and could be pulled to do so in new directions by targeted prize funds such as AgResults.