12 January 2010

Unexpected Data

The phrase “seismic echo” recently popped into my head for no reason that I can explain. When I Googled it I discovered that it’s not a common phrase. However, I did come across an interesting paper from 1970, published by one A.I. Mukhamedzhanov.

The paper is titled The Nature of the Lunar Seismic Echo. Sadly, the full text is not available online, but the abstract alone was worth a read:

One of the interesting effects discovered during the recent Apollo 12 mission was the unexpectedly long duration of the Moon's seismic reaction to the impact of the ascent stage of the lunar module. Instead of the expected few minutes, the oscillations continued for about 55 minutes. No less unexpected and indeed somewhat disconcerting was the reaction among the specialists who were immediately faced with the difficulty of explaining this effect. The specialist world, under such circumstances, is reminiscent of a beehive reacting to an unexpected intrusion. However, the seeming chaos of conflicting opinions here involves three attitudes: (1) disbelief in the instrumentation; (2) the desire to construct, without delay, a new theory; (3) the attempt to account for the new phenomenon on the basis of some old theory. The duration of the seismic oscillations is calculated directly in terms of a theory worked out by the author back in 1965 (the multiple-cascade fall of material ejected by the impact of a meteorite on the lunar surface) and it is indicated just what was responsible for the Apollo 12 effect.

Regardless of how fascinating the science is (and it is!), what I particularly liked was Mr. Mukhamedzhanov’s observation of how specialists react to unexpected data. Basically, they:
1) Blame the sensor / disbelieve the data.
2) Accept the data and create a new theory.
3) Rely on an older, previously neglected theory to explain the data.

When faced with an unexpected outcome, is this a good or bad way to react? My first reaction was to chuckle at the “disconcerting reaction.” But the more I think about it, this reaction just might make sense.

And scientists aren't the only ones who face unexpected data. Project leaders get that all the time. How should we react? Blame the instrument? Create a new approach? Regroup?

Thoughts?

2 comments:

Dick Field said...

If these are "known unknowns", they should not be unexpected data. Much as an in-car GPS system sensing a different turn, we should proceed to "recalculate", based on all that is known in an enriched database.

Gabe said...

Isn't it all three in some sequential order? In experimentation, it seems the first choice is to examine the instrument. If it appears to be working correctly, then on to the next stage of trouble shooting. You may not get to finality about what happened, but its a good way to rule out possibilities.