05-15-2016, 12:00 PM
A Very Newbie Contribution
As someone who has closely followed efforts to find MH370 from the beginning, it is annoying to read that Martin Dolan continues to blow off his and ATSB’s collective negligence in a horribly botched search for the plane and its victims.
To be clear, no one knows where the plane is: I don’t know where it is; Martin Dolan doesn’t know where it is; the Pope doesn’t know where it is; PAIN doesn’t know where it is. But I know that a random kid working his first job at a fast-food drive-up window would have made consistently better decisions than Martin Dolan made. And on top of it, Dolan has the chutzpah to suggest he has a “diminishing level of confidence”. Such a pretentious shrug of the shoulders!
In a fair world Martin Dolan would be criminally charged for letting the same thing happen in MH370 that happened in Pel-Air. Did he learn anything at all from the Pel-Air fiasco? Yes! He learned how to be even more vague, evasive, and unaccountable. On a visceral level, Martin Dolan is offensive: 1) he deliberately hired a company that intended to use cheap, outdated, nearly blind sidescan sonar towfish; 2) he deliberately ignored SAR applicants with proven track records in the Air France Flight 447 recovery (because he preferred to save a few bucks); and, worst of all 3) after all ‘Ts’ were crossed and contracts signed, Martin Dolan never looked back; never considered that his strategy halfway to Penguin Land was hopelessly flawed. Instead, he pulled strings to have CSIRO and other Australia General Fund “troughers” contribute ridiculous assessments of the “correctness” of his poorly informed decisions.
A Look Back
ATSB’s original best guess for MH370’s endpoint was at -38.0S, 88.5E. At the time, that was as good a guess as any. I certainly supported it, and set out to monitor search efforts via satellite for as long as it took.
But I am also statistically oriented by training and occupation, and I frequently revisit assumptions I attempt to work within. If I notice that original assumptions are not supported by what I’ve learned since framing them, I stop and review everything. It’s usually called Quality Control, and no modern agency or business succeeds without it. Japan learned that lesson quickly after World War II, thanks in part to W. Edwards Deming. But it was a lesson United States industry failed to learn before it lost dominance in auto markets worldwide, and it is a lesson ATSB and Martin Dolan have never learned.
From Bad to Worse
Unknown to most of the world, Prime Minister Tony Abbott’s term was rapidly drawing to a close in August 2015 when the first piece of debris from MH370 turned up on Reunion Island, thousands of kilometers west of the “7th Arc”. It is impossible to know if anything would have changed if Mr. Abbott had not been embroiled in a fight for political survival. What we know is that it gave Martin Dolan a free hand to do as he wished with the search itself at the single most critical juncture since the plane vanished.
And Martin Dolan’s instincts were to cover his behind. He quickly came up with a new drift model… invented overnight… to replace the one he had been using for six months. The original drift model predicted aircraft debris was likely to wash up on Indonesia’s Java and Sumatra islands if the plane came down on or near the 7th Arc in the Indian Ocean.
But with recovery of the flaperon thousands of kilometers in the wrong direction, Dolan instantly recognized he had a problem. With a free hand from the diminishing presence of Mr. Abbott, Dolan quickly convinced CSIRO to come up with a joke of an analysis that asked the drift question in a way that required the answer Dolan wanted. (NOAA made a special run for me with the same data, and it points unequivocally to the Wharton Basin area.)
Next, Dolan took steps to make sure he wouldn’t be pressured into moving the search at all. He found someone who came up with a Bayesian model to “prove” ATSB was searching the correct part of the southern Indian Ocean. It is a 27 page farce (128 pages for the full PDF). It was a farce the moment it was published, but it is now a verified farce. The only parts of MH370 that have been found have been found by people who are not being paid to do stupid things or claim they are knowledgeable of Bayesian Statistics.
(What most people don’t realize is that “Bayesian” is researcher-speak for “I have absolutely no idea what I’m doing, but I’ll bet you’re impressed I know ‘Bayesian’ is a word.”)
How Do We Know ATSB is Wrong?
By December 2014, less than six months after sidescan sonar began, it was statistically apparent that the search was in the wrong part of the southern Indian Ocean. In fact, it wasn’t at all clear that the plane came down anywhere in the Indian Ocean until the flaperon appeared in late July 2015. (Probability assessments never “prove” something did or didn’t happen; they simply quantify the likelihood of an event; usually less than 100% certainty.)
To the best of my knowledge, ATSB never made a professional, testable statement about where the plane would be found. It picked -38.0S, 88.5E, but I’ve never seen a descriptive analysis of how ATSB believed it got there. Just one more omission in a long string of them.
In any event, it follows that if the plane came down at the coordinates ATSB has been using for two years, it had to have hit the 6th Arc at a predictable point some 100 km northwest. One only has to work it back to find that point, and then work it forward and calculate a couple of statistics to know exactly how much seafloor has to be scanned to either find it or go somewhere else.
Of course, ATSB didn’t do all that formal professional stuff. Probably has no idea what ‘margin of error’ is or how to calculate it. They also subscribe to Mr. Miyagi’s “scan on, scan off” guide. The one that just keeps going until it seems like it’s time to quit. If Toyota made cars that way, they’d have gone out of business long ago.
The endpoint, by the way, was based on a set of assumptions that had the plane turning mostly south somewhere over the south Andaman Sea and flying … probably on autopilot … until fuel was exhausted. Various contributing speculation framed it as a murder / suicide by the pilot, an electrical or cargo fire, or as just a horrible tragedy those on board could not bring under control.
In any event, it turns out that the margin of error for ATSB’s terminal location is not very large: about 15 kilometers in radius for 3-Sigma certainty, which is fairly standard for “government work”; 30 kilometers for 6-Sigma certainty, which is an industrial standard; and about 50 kilometers for what is known as 10-Sigma certainty. The 10-Sigma variety is so rare and so expensive to test, it is never used. But, ATSB has so over-scanned the area it predicted the plane came down in that it is an event that could only happen once in many billions of years. It isn’t there!
As someone who has closely followed efforts to find MH370 from the beginning, it is annoying to read that Martin Dolan continues to blow off his and ATSB’s collective negligence in a horribly botched search for the plane and its victims.
To be clear, no one knows where the plane is: I don’t know where it is; Martin Dolan doesn’t know where it is; the Pope doesn’t know where it is; PAIN doesn’t know where it is. But I know that a random kid working his first job at a fast-food drive-up window would have made consistently better decisions than Martin Dolan made. And on top of it, Dolan has the chutzpah to suggest he has a “diminishing level of confidence”. Such a pretentious shrug of the shoulders!
In a fair world Martin Dolan would be criminally charged for letting the same thing happen in MH370 that happened in Pel-Air. Did he learn anything at all from the Pel-Air fiasco? Yes! He learned how to be even more vague, evasive, and unaccountable. On a visceral level, Martin Dolan is offensive: 1) he deliberately hired a company that intended to use cheap, outdated, nearly blind sidescan sonar towfish; 2) he deliberately ignored SAR applicants with proven track records in the Air France Flight 447 recovery (because he preferred to save a few bucks); and, worst of all 3) after all ‘Ts’ were crossed and contracts signed, Martin Dolan never looked back; never considered that his strategy halfway to Penguin Land was hopelessly flawed. Instead, he pulled strings to have CSIRO and other Australia General Fund “troughers” contribute ridiculous assessments of the “correctness” of his poorly informed decisions.
A Look Back
ATSB’s original best guess for MH370’s endpoint was at -38.0S, 88.5E. At the time, that was as good a guess as any. I certainly supported it, and set out to monitor search efforts via satellite for as long as it took.
But I am also statistically oriented by training and occupation, and I frequently revisit assumptions I attempt to work within. If I notice that original assumptions are not supported by what I’ve learned since framing them, I stop and review everything. It’s usually called Quality Control, and no modern agency or business succeeds without it. Japan learned that lesson quickly after World War II, thanks in part to W. Edwards Deming. But it was a lesson United States industry failed to learn before it lost dominance in auto markets worldwide, and it is a lesson ATSB and Martin Dolan have never learned.
From Bad to Worse
Unknown to most of the world, Prime Minister Tony Abbott’s term was rapidly drawing to a close in August 2015 when the first piece of debris from MH370 turned up on Reunion Island, thousands of kilometers west of the “7th Arc”. It is impossible to know if anything would have changed if Mr. Abbott had not been embroiled in a fight for political survival. What we know is that it gave Martin Dolan a free hand to do as he wished with the search itself at the single most critical juncture since the plane vanished.
And Martin Dolan’s instincts were to cover his behind. He quickly came up with a new drift model… invented overnight… to replace the one he had been using for six months. The original drift model predicted aircraft debris was likely to wash up on Indonesia’s Java and Sumatra islands if the plane came down on or near the 7th Arc in the Indian Ocean.
But with recovery of the flaperon thousands of kilometers in the wrong direction, Dolan instantly recognized he had a problem. With a free hand from the diminishing presence of Mr. Abbott, Dolan quickly convinced CSIRO to come up with a joke of an analysis that asked the drift question in a way that required the answer Dolan wanted. (NOAA made a special run for me with the same data, and it points unequivocally to the Wharton Basin area.)
Next, Dolan took steps to make sure he wouldn’t be pressured into moving the search at all. He found someone who came up with a Bayesian model to “prove” ATSB was searching the correct part of the southern Indian Ocean. It is a 27 page farce (128 pages for the full PDF). It was a farce the moment it was published, but it is now a verified farce. The only parts of MH370 that have been found have been found by people who are not being paid to do stupid things or claim they are knowledgeable of Bayesian Statistics.
(What most people don’t realize is that “Bayesian” is researcher-speak for “I have absolutely no idea what I’m doing, but I’ll bet you’re impressed I know ‘Bayesian’ is a word.”)
How Do We Know ATSB is Wrong?
By December 2014, less than six months after sidescan sonar began, it was statistically apparent that the search was in the wrong part of the southern Indian Ocean. In fact, it wasn’t at all clear that the plane came down anywhere in the Indian Ocean until the flaperon appeared in late July 2015. (Probability assessments never “prove” something did or didn’t happen; they simply quantify the likelihood of an event; usually less than 100% certainty.)
To the best of my knowledge, ATSB never made a professional, testable statement about where the plane would be found. It picked -38.0S, 88.5E, but I’ve never seen a descriptive analysis of how ATSB believed it got there. Just one more omission in a long string of them.
In any event, it follows that if the plane came down at the coordinates ATSB has been using for two years, it had to have hit the 6th Arc at a predictable point some 100 km northwest. One only has to work it back to find that point, and then work it forward and calculate a couple of statistics to know exactly how much seafloor has to be scanned to either find it or go somewhere else.
Of course, ATSB didn’t do all that formal professional stuff. Probably has no idea what ‘margin of error’ is or how to calculate it. They also subscribe to Mr. Miyagi’s “scan on, scan off” guide. The one that just keeps going until it seems like it’s time to quit. If Toyota made cars that way, they’d have gone out of business long ago.
The endpoint, by the way, was based on a set of assumptions that had the plane turning mostly south somewhere over the south Andaman Sea and flying … probably on autopilot … until fuel was exhausted. Various contributing speculation framed it as a murder / suicide by the pilot, an electrical or cargo fire, or as just a horrible tragedy those on board could not bring under control.
In any event, it turns out that the margin of error for ATSB’s terminal location is not very large: about 15 kilometers in radius for 3-Sigma certainty, which is fairly standard for “government work”; 30 kilometers for 6-Sigma certainty, which is an industrial standard; and about 50 kilometers for what is known as 10-Sigma certainty. The 10-Sigma variety is so rare and so expensive to test, it is never used. But, ATSB has so over-scanned the area it predicted the plane came down in that it is an event that could only happen once in many billions of years. It isn’t there!