In article ,
(Derek Lyons) writes:
| Dave Ludlow wrote:
| It would be quite logical to risk 4 volunteer crew with a (say) 10 to
| 30% risk of death in an attempt to save seven more with a (say) 80%
| probability of death.
|
| The problem isn't the crew, they are easily replaced. Atlantis
| however is near irreplaceable.

The same argument applies with slightly different numbers. Get the
crew down first, and run another mission with a foam repair kit.
If all goes well, you have saved both shuttles.

| I don't know how the *overall* probabilities would have worked out but
| the chance of foam strikes causing severe wing damage in successive
| shuttle missions, given the history, would be quite low.
|
| The overall probability is low, but both 107 *and the flight
| immediately before* suffered foam strikes, and both lost the left
| bipod ramp.
|
| That tells me that that using 'overall' statistics is shakey at best.
| (Which is what NASA did, and paid for it in blood.)

It shouldn't. Using overall statistics is fine, provided that you
don't ABUSE them. And assuming independence of events with no
reason to believe it is a very common abuse of statistics.

In this context, it raises the question of whether that was a simple
coincidence or due to a common cause. With the former, it would
be very unlikely to happen a third time; with the latter, it might
be almost certain to.


Regards,
Nick Maclaren.