In June this year, the world's top high-performance computer was announced as Japan's K computer at the RIKEN Advanced Institute for Computational Science (AICS) in Kobe. With a maximum performance of 8.16 petaFLOPS, an impressive leap from the previous number 1, the Chinese Tianhe-1A system at the National Supercomputer Center in Tianjin, which had 2.57 petaFLOPS.
Before the reveal in June and November each year few people know which system will top the list, usually it is just the people who work at the institute with the most powerful computer. But in June this year, the website Top500 predict used the power of 'crowd sourcing' to get the accurate prediction of 8.17 petaFLOPS at Japan's AICS.
The website, launched in May 2011, currently has 250 active users and uses 'game theory' algorithms and human prediction to forecast what the next supercomputer will be. You can see it like stock market trading, according to Martin Kolb, co-founder of delphit, a business intelligence company in Mannheim, Germany that created the Top500 predict site to test their software algorithms.
Place your bets
The website gives users 'virtual money' as the currency to place their bets. You can join an existing bet, or place one of your own, on questions such as: How many CPU cores will the next top high-performance computer have? or Which country will have the next top high-performance computer?
There is a risk to reward ratio too. The more you bet against the crowd, the higher the risk, and the more chances you'll double or triple the amount of credits you receive. "You can bet on the future, just like the stock market, but unlike it you won't lose any real money if you lose your bet," said Kolb. Delphit offers prizes, such as the Apple iPAD 2, to the people who are the most accurate in their predictions.
This 'crowd sourcing' tool is a great way for Kolb and his team to use social behavior to make their prediction algorithms more accurate and to develop their business intelligence models. And it has already proven its worth, he said, as the first prediction about Japan's K Computer from their site was incredibly accurate.
Kolb hopes to use the data his website collects to help other companies in their risk management strategies. "As more predictions are made, our algorithms can be calibrated to be more precise. The overall goal is to help businesses become more efficient. The high-performance computer market is an ideal place to test our research as we aim to aggregate knowledge," said Kolb.
The next top 500 list will come out November 2011, so start making your bets for a chance to win a prize or just to test your knowledge of the future.