How one man figured out corn.
Computers can predict a lot of things: the weather, baseball statistics, traffic... and now, crop patterns. (While predicting crop patterns may not seem like a big deal to you, tell that to your ancestors who lived off the land two hundred years ago.)
Descartes Labs was founded by Mark Johnson to compete with the USDA in figuring out how healthy corn crops are. While the USDA does an okay job, it uses up a lot of resources and man hours to do so, whereas Johnson leads a tiny operation of twenty employees and some satellites. Corn is pretty important, as most midwesterners will tell you; it's used for a lot more than just food (think: the stuff you use to make your car run) and we need to be able to accurately predict how much corn there will be in any given harvest year. According to Johnson, his predictions have beaten out the USDA's in accuracy every single time, with just a 2.5% margin of error.
Johnson's secret? Nanosatellites. They're tiny satellites that allow start ups to map the planet without relying on government resources. Of course, there are a lot of ethical questions to be asked about the role of private companies in competing with government agencies, and the amount of jobs eradicated because of machine learning, but perhaps this time the US government can take a page out of Johnson's book and step up their game so we know exactly how much corn there'll be on the Thanksgiving dinner table this year.
(h/t The Verge)