Kaggle Solves Big Data Problems With Contests — And Now Has Big Funders and $11M on Board
What if the Netflix Prize model — solving hard problems about big data sets using contests — could be applied to all sorts of other things? A remarkable start-up called Kaggle is doing exactly that, and already seems to be making it work.
Kaggle has facilitated breakthroughs in NASA’s analysis of dark matter, improved Allstate’s actuarial methods, predicted many of the top finishers of the Eurovision Song Contest, and is currently hosting a $3 million prize to devise ways to reduce unnecessary hospitalizations.
The company, which started last year in Northern Australia and has now relocated to San Francisco, has been endorsed by some of the tech industry’s best-respected brains via a new $11 million round of funding that was apparently highly competitive.
Led by Index Ventures and Khosla Ventures, the Series A round included PayPal co-founder Max Levchin, Google Chief Economist Hal Varian and Factual CEO Gil Elbaz, as well as SV Angel, Yuri Milner and the Stanford Management company.
Kaggle founder and CEO Anthony Goldbloom, a former statistician for the Australian treasury, says his company addresses “a serious market failure.” That is: Companies have data and can’t analyze it as well as they’d like, and academia is desperate for real-world data sets.
Kaggle’s short history is full of charming anecdotes about its democratic approach to yielding better and unanticipated results. Goldbloom wrote in a recent blog post, “In past Kaggle competitions, breakthroughs in astronomy have been made by glaciologists, chess rating systems have been beaten by non-players, and bioinformatics problems have been solved by SEO specialists.”
Goldbloom hired Kaggle’s most successful competitor, Jeremy Howard, and made him president and chief scientist.
Today, Kaggle’s 17,000 participating data scientists — which happen to include former Netflix Prize winners — participate mostly for the challenge and the chance to prove themselves. But Goldbloom and Howard want to make it worth their while.
“In five years’ time, I want to do 10,000 competitions per year,” Goldbloom said. “I’m hoping competitors can earn a full-time income from Kaggle.”
On that front, Kaggle will now start paying small groups of data scientists — selected based on their past performance on Kaggle — to analyze sensitive data sets for which companies require NDAs.