Prizes solve AI-based innovation – Vesuvius Challenge proves

Last week, I wrote about the Vesuvius Challenge, a $1 million competition with the goal of using artificial intelligence and machine learning to read 2,000-year-old papyrus scrolls. I’m still thinking about the way a bunch of students pulled off the seemingly impossible—and not just because using modern technology to crack an ancient mystery is undeniably amazing. It’s also because the success of this contest should inspire a lot more contests.

Prize contests encourage talented people to work across disciplines and bring unusual perspectives beyond their areas of expertise. They are especially valuable in areas where progress is hard and unlikely to happen without a push, but not so hard that it’s unattainable. The most effective prize contests are built around a simple premise: This is what we want to do, and we don’t care how it gets done or who does it.

But the real key to prize contests—and the reason there should be more of them—is that they pay only for success. That’s a more efficient model than predicting who is most likely to succeed and giving those people money even if they fail.

They can help with today’s issues and shape the future. And they have already uncovered secrets of the past.

The Vesuvius Challenge happens to be a perfect example of how prize contests should work.

The first thing it got right is that it had a clear, ambitious target: Read four legible passages of 140 characters in the unopened scrolls by the end of 2023. Instead of telling people to do something interesting with AI, people were told to use AI to do this specific interesting thing.

Netflix was similarly precise about benchmarks when the company famously offered $1 million in 2006 to the first algorithm that could beat its recommendation engine by 10%. The company declared a winner in 2009 and paid out the prize, but never used the code.

The Vesuvius Challenge also focused the attention of an entire community on a puzzle that had vexed humanity for centuries. It manufactured an urgent race against the clock by setting a hard deadline. And it gave smart young developers an incentive to compete and collaborate, dangling incremental prizes for those who published their code and shared their critical discoveries with others.

Some of them were driven by money. But many of them just wanted to win. And most of them kept working because they were too obsessed to think about anything else. Silicon Valley has a word for that: They got nerd-sniped.

Napoleon to NASA

Prize contests like the Vesuvius Challenge have existed for so long and resulted in so many unexpectedly brilliant ideas that it’s surprising there aren’t more of them by now.