Wednesday, January 5, 2011

Yes or no, take your pick: the implications of uncertainty

From Flickr via
Wikimedia Commons
It's holiday book-reading time, the best part of a holiday.  One isn't usually given a dirty book on these occasions, but a very interesting book under my tree this year was Dirt, by geologist David Montgomery.  This discusses a topic vital to survival: the soil.  The subtitle gives away the author's point:  "The Erosion of Civilization."

Dirt is about the repeated way that agriculture and other uses of the land by large human populations, such as cover much of the earth these days, erodes or degrades soil much faster than it can be replaced by natural processes.  Montgomery points out that soil buildup is slow, and erosion, too, may be so slow that we hardly notice it. In our current scientific, data-laden age, with huge populations and rapid growth, we seem more able to notice the changes than our ancestors may have been able to do.

Although we have only sparse data on relative rates of soil production vs soil erosion, the data do indicate that loss is greater than gain.  But in a pithy sentence the author notes that

This leaves the issue in a position not unlike global warming--while academics argue over the details, vested interests stake out positions to defend behind smokescreens of uncertainty.

And it's not just global warming.  We see the very same posturing in regard to biomedical genomics in which convenient self-serving positions are taken in regard to the effectiveness of genomic predictions of health and disease susceptibility.  Some argue that genetic causation is complex and that there are better ways to spend health resources (both in terms of research and application), while others promise near immortality from 'personalized genomic medicine'.

MT readers know where we stand on that particular issue.  But a more important and deeper point is the pervasiveness of the same kind of issues in our society.  We purport to be a fact-driven 'evidence based' society, that relies on science for important decisions.  But from foreign policy, economics, and many other fields, to genetics we face masses of data on phenomena that clearly involve many different contributing factors.  Often, we don't even know what all the factors are, much less being able to measure them accurately.

Normal science, the scientific method we pretend to follow so rigorously, is designed to answer very focused questions.  We do this by isolating one variable at a time under controlled conditions, so we can determine what that variable does.  But with complex causation this is difficult to do, if not impossible in practice, and single causes may not have much explanatory power even if we could estimate it accurately.  Often we deal with processes too slow to be accurately predicted by extrapolation from what we can confidently estimate from science.  We know that's true about evolution, but it applies also to soil abundance and climate change among others.

As 2011 starts out, dealing with such complexities is perhaps the most important general challenge to science today.  In many areas, it doesn't much matter how a question is approached or answered, and much of the perhaps properly ridiculed attention of academic research is rooted in trivia taken far too seriously (when professors should be teaching instead).  Who cares whether a hangnail led Shakespeare to write Othello, or how an ostrich leg-bone evolved?  Only a few professors struggling for dominance in their journals!

But a lot of major questions that science is being asked to answer do have major implications for our society, lifestyles, and behavior.  When that is the case, since those of us who are haves like to have more, more, more (or, at least, not to have to change from comfortable ways of living), it is only natural to see vested-interest posturing.  Each side thinks it's right, and resents the benighted opposition, and each side will tend to slant the facts or color their interpretations in their own self-interest.  Literature going back to the beginning of literature shows the natural resistance to change in human society, especially among the established generation at any given time.

There is no easy way out of this when facts are incomplete and phenomena too slow or complex, so that we need to extrapolate beyond direct observation.  Sometimes, the process is really slow in this sense.  In other times, things can change very rapidly, but we just don't understand them enough (or care to).  Major economic collapses or political dominoes leading to military disasters are clear examples.

Our lives, or those of our children, depend on acting on knowledge.  The history of dirt shows how entire civilizations can go under if they fail to act soon or definitively enough.  In the past, perhaps they had no knowledge or choice.  But we do have at least the archeological record, and some choice.  But we rarely seem to act before tragedy strikes.  We are each of us too rooted in the vested interests of our own lives to be as free to act as might be best, and uncertainty provides a handy excuse for inaction.

Can science adapt to the new landscape of complexity?

1 comment:

Arjun said...

I'm glad you brought up the scientific method's inadequacy with regard to dealing with issues of complex causation. It's an issue I've often wondered about in light of your final question.

The closest current remedy that comes to mind lies with response surface methodology (

Though the motivation for RSM lies in optimizing a process, it can in principle be used to at least characterize a complex explanatory-response profile.