Originally Posted by
MSPeconomist
Someone earlier in this thread stated that 0.5% ( 2 out of 350, so there's been some rounding) of this aircraft model have crashed, but what percentage of all the Concords that were ever build have crashed? Yet it was viewed as safe until the CDG runway incident happened and possibly should have still been considered safe. Or we can look at the space shuttle for another example. One crashed (cold weather O ring failure at launch), but that one is a big fraction of the number that were build and also a relatively big fraction of the total number of space shuttle flights that were ever attempted.
It's hard to conclude much (if anything) from a few bad random draws from a distribution where the bad draws occur with extremely low frequency. For example, think about Hurricane Katrina and the "once in a hundred years" claim.
And if we look at the loss of Challenger(and later Colombia) shuttles? The space program suspended operation, after a single incident, until they could properly investigate, determine a cause, and implement a fix. Granted these incidents as you mention are a much larger percentage of their fleet, but still it only took one event for them to put the program on hold while they investigated further.
I'll admit going the NASA route and grounding an entire model every time a single crash occurs isn't realistic for commercial air travel, you'll wipe out giant portions of airlines fleets for months at a time while an investigation occurs. However it is also concerning that two of the same model aircraft have crashed, in the same stage of flight, and based on initial reports suggest similar circumstances should raise enough eye brows. Can it be random chance? Sure, but how many other models of airplanes have we had have two crashes in a short period of time? While being a relatively new model? In a day-and-age where historically air travel has been the safest ever, shouldn't that raise even more concern that these crashes are occurring? Or has the past 20 years just been an anomaly, after all we are dealing with an overall small event rate across the millions of flights that have taken place? Perhaps the expected crash rate is actually higher than what we have been seeing over the past couple decades?
Point being, we can make any number of statements regarding probability. At what point, with the rare occurrence of airplane crashes, do we say that is enough to raise concern for a specific model? NASA's approach is one event is too many. That is obviously unrealistic for commercial airlines. One event is a data point, two events is a line, three is a trend. When you're dealing with human lives, do you want to get to a trend? That answer is going to be different to everyone, I know the medical field doesn't like negative trends; and the medical field is supposed to be learning from the airline industry.