MH370: Triumph of the Weird

Well, here we are in cold-case land, scratching our heads. Some $150 million has been spent and no plane. Where does that leave us?

For one thing, with radically altered probabilities of what might have happened.

Imagine that we dial the clock back to August, 2015. You’re Warren Truss and this the story that your data is telling you:

Based on this default story, you’ve concluded that there’s a 97 percent chance that the aircraft hit the ocean within a 120,000 sq km box. What about the other 3 percent? Let’s imagine there’s a 2 percent chance that it’s somewhere else in the SIO, and a small but finite chance — let’s say 1 percent — that, for unkonwn and uncalculable reasons, the plane didn’t go into the SIO at all.

Time goes by. You search out all but the 1 percent of the search box that your sonar equipment can’t image (e.g. seabed crevasses), throw up your hands, and call it off.

So this is how things now stand: Of the orginal 100 percent, 96 percent has been scanned and ruled out. Here’s how remaining probabilities now stack up:

Of course, these are all very rough numbers. The point being, no matter how you slice it, the scenario that was once nearly a dead certainty (flying into the SIO search box) is now less than an even bet, and outcomes that once barely merited an asterisk are now not only possible but probable.

The most probable category, according to this rough calculation, would be scenarios of the second variety. But if the data is valid, how could the default story be wrong? How could the plane have wound up somewhere in the SIO outside the search box? To square that circle, you have to choose one of the assumptions above and bend it. For instance, one might imagine that the 18:40 BFO value was not caused by the plane flying south, but by a plane that descended–perhaps, say, for a descent into Car Nicobar–and then changed plans and flew instead to, say, Antarctica. Or maybe the plane didn’t fly straight and fast, but flew slowly in a curve toward the Cocos Islands, creating a pattern of ping rings that only happened to look similar to those generated by a plane flying fast and straight.

Such eventualities are so unlikely that, back when the search box was being drawn, it was easy to simply discard them. But now that the most reasonable options are off the table, this very geographically dispersed (and hence impractical to search) population of possible endpoints collectively adds up to “very likely.”

Then again, it’s also now significantly more plausible that the plane didn’t go south at all, or that if it did it wound up in the search box but then fell into an unscannable crevasse.

Whatever happened to MH370, it wasn’t the default story told by the data, but rather something that in the summer of 2015 would have been discarded as hopelessly implausible.