Wednesday, March 19, 2014

Who speaks for the facts?

Everyone else is speculating on what happened to Malaysia Airlines Flight 370, so we don't have to. The one thing that most people agree on is that the plane met a tragic end, and we pause for a moment to recognize the pain and sorrow this has caused so many people. But, the wide range of hypotheses is fascinating, and we wanted to briefly consider why there are so many and what this might say about science.

We don't need to rehash the possibilities, which range from simple accident to malicious intent on the part of one or more passengers, or even the pilots.  Indeed, you can generate your own conspiracy theory right here.  E.g.:

But a new hypothesis is intriguing and puts the rest in stark relief -- a piece over at Wired, "A Startlingly Simple Theory About the Missing Malaysia Airlines Jet", written by a pilot named Chris Goodfellow, is a reminder that perspective is everything. Well, informed perspective.

Goodfellow says:
There has been a lot of speculation about Malaysia Airlines Flight 370. Terrorism, hijacking, meteors. I cannot believe the analysis on CNN; it’s almost disturbing. I tend to look for a simpler explanation, and I find it with the 13,000-foot runway at Pulau Langkawi.
He goes on to say that pilots are trained to always know where the closest, safest airport is if something were to happen to the airplane they are flying.  He says the pilot of Flight 370 must have landed at Pulau Langkawi many times, and that he must have realized that there was an emergency on board, and turned toward that airport, knowing it was the closest place he would be able to land.
What I think happened is the flight crew was overcome by smoke and the plane continued on the heading, probably on George (autopilot), until it ran out of fuel or the fire destroyed the control surfaces and it crashed. You will find it along that route–looking elsewhere is pointless.
Goodfellow says this fits all the data, from the loss of transponders to the turning of the aircraft.  He is appalled that people are considering the pilot to have had malicious intent, when in all likelihood -- in fact Goodfellow is certain -- instead he acted heroically.  See his post for the detailed explanation.

Our point here is not to explain what happened, because we have no idea, but to take a look at how hypotheses are made, and how they are evaluated.  Goodfellow is an old and experienced pilot.  He is evaluating the data from inside the pilot's head, and it turns out that his is probably the most parsimonious explanation.  And, as everyone knows, a parsimonious explanation is always the best explanation.  Occam's razor.  And add to that that 'experts' are assumed to have judgment as to what explanation is most parsimonious.

Fossils from the Cretaceous, found in Lebanon; Wikimedia

Except that parsimony is an arbitrary measure of the quality of data assessment.  The basic idea, and it certainly has much merit, is that when a simple explanation will do, adding other variables and explanations is not called for by the data and as a general principle we avoid making explanations more complex than they need to be to account for our data.  Which is not to say that parsimonious explanations can't be the best, just that parsimony itself isn't why they are the best.

Everyone is working with the same data on this airplane mystery.  But terrorism experts see evidence of terrorism, mechanics see sure signs of engine trouble, and so forth.  It's the pilot who can imagine himself in the pilot's seat who, at least to us, makes the most sense.  Not because of parsimony -- do we have good reason to think that the flight tragedy was due to one thing only?  An electrical failure or fire, or a hijacking, or sudden gain in altitude that knocked everyone, pilot and passenger, out cold?  Normally, in this case, we would not think that there was, say, a hijacking and then an on-board fire.  That would be complex compared to just one of those--but how would we know that more than one event didn't account for things?  The hijackers could have jerked on the controls and sent the plane too high, which knocked everyone out, etc.

Also, of the separate hypotheses, is there really at this stage a good reason based on parsimony to choose one over the other?  Each author has his own idea, or advocacy.  The pilot's explanation seems to make the most sense because he's a pilot, and can evaluate the evidence as if he'd been flying a troubled plane.

And in science
Everyone saw the same world Charles Darwin saw in the early to mid 1800's, too, and only he and Alfred Wallace made lasting order from the data, suggesting that the diversity of life is the result of descent with modification from a common ancestor via natural selection, and the fact of evolution is confirmed every day in biology labs around the world.  Alternative explanations for the origins of the diversity that everyone saw around them were many; then as now fundamentalists saw the hand of God and, fossils were thought to be 'sports of nature', not evidence of extinct species, because God wouldn't have made species that went extinct.  A case can be made that Darwin lost his faith after the death of his daughter, Annie, and a theory that didn't require a divine being was one he could readily accept.

One argument against the God argument is that God would have had to have a separate reason for making insects have 6 legs, but not spiders or lobsters, and vertebrates four, but not fish or birds, who have fins and wings.  That is, God made a separate, unique decision about each creature.  That is more complex than saying evolution and common ancestry generically account for what we see. 

Now evolution is different for every creature, just as God's choices would have been, and 'God's choice' is as parsimonious as 'evolution', so the criteria for what is the empirically simplest explanation does involve some subjective judgment.  If evolution is a true generic explanation for life and its diversity, and all the evidence suggests that it is, we tend to seek the simplest evolutionary explanation.  Still, nothing guarantees that an ad hoc phenomenon like evolution has to go from A to Z by the shortest path!

Indeed, there are even those who question the 'obvious' physics principle that radiation like light travels in 'straight' lines (here, modern physics might say 'geodesic' lines in spacetime).  However, some physicists say that energy takes infinite histories all at once, and that the straight-line path is not 'the' path but the average path taken by the radiation.  That puts a twist, so to speak, on parsimony ideas.

And in genetics, everyone has access to the same data, when GWAS results are reported, or genome sequences uploaded to GenBank, yet some geneticists see simple answers and some see complexity. And evolutionary psychologists see adaptive explanations where others might see genetic drift. And many look for one explanation -- this gene, or that bad lifestyle -- even when it is blatantly obvious that genes and lifestyles have effects only in their overall context.  So what is the most 'parsimonious' explanation is debatable and we have no real rigorous way to adjudicate in many instances -- even if we would all agree that we don't invoke a gene plus the tooth fairy plus raging storms on Jupiter to account for a case of cystic fibrosis.

Observations don't speak for themselves, in science or forensics.  It is through the mouths of subjective observers that they speak, observers with more and less credibility.  Who we choose to believe is in turn affected by our own subjective decisions, not usually an evaluation of which argument is more parsimonious.  Parsimony can be one criterion, but it may not be 'the' true explanation, or even the best one.  Life, due to its meandering and highly probabilistic pathways, may be one of the least parsimonious phenomena that science tries to understand.


Manoj Samanta said...

How does the data - "plane continued to contact satellite for 7 hours" - fit the simple explanation?

Ken Weiss said...

Several explanations have been offered to make this possible. What is 'parsimonious' about any of them is unclear

Ken Weiss said...

Ah, yes, well, parsimony de-parsimonied: