The ocean is a big place. Many people think that's reason enough to explain why the wreckage of MH370 wasn’t found in the seabed search.
But actually, the scientists who defined the search area thought they had a pretty accurate understanding of where the plane had flown to. In episode 4, we talked about why they were confident the plane couldn't have flown very far beyond the 7th arc. In today’s episode, we discuss how they managed to define the other dimension of the search box, its length. How far north or south could the plane have flown as it reached the end of its flight? They struggled to pin down a methodology, trying and discarding several approaches before settling on a method that allowed them to state, with mathematical precision, where the seabed search would find the plane.
That search, of course, failed. But math is math — if the calculations failed to yield the correct location of the plane, there must have been an extenuating factors that the scientists hadn't taken into account. A branch of statistics called Bayesian inference offers guidance on what to do next.
And now for a quick aside: as you may know, we’re running a Kickstarter campaign to fund an experiment that could resolve some of the key paradoxes in the mystery of MH370. We’ve only got a couple more weeks to reach our goal, so if you’d like to get some clarifying answers, the time to act is now. Click here to go to the Kickstarter page of The Finding MH370 Project.
Back to Bayesian statistics. The technique is enjoying something of a heyday right now, in part because of its crucial role in neural network learning. I first became aware of Bayesian methods because of its role in helping devine the seabed search area. But once you know about it, you see in everywhere; its influence is remarkably widespread and deep across the sciences. For a smart and entertaining explanation, I recommend the book that I held up during today’s episode, Everything is Predictable, by the British science writer Tom Chivers. I had the pleasure of interviewing Chivers for a New York magazine article about the software billionaire Mike Lynch, who drowned recently aboard his megayacht named — yes, really — Bayesian.
I also talk about the work of DSTG scientist Neil Gordon, who co-authored a paper that explained how Bayesian methods were applied in defining the seabed search area. That paper was eventually published in book form, which you can find here.